SOCIOECOLOGICAL PREDICTORS OF MICROBIOME VARIATION IN WILD POPULATIONS OF AFRICAN MAMMALS By Connie A. Rojas A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Zoology – Doctor of Philosophy Ecology, Evolutionary Biology, and Behavior – Dual Major 2021 SOCIOECOLOGICAL PREDICTORS OF MICROBIOME VARIATION IN WILD POPULATIONS OF AFRICAN MAMMALS ABSTRACT By Connie A. Rojas Host-associated microbial communities (e.g. microbiomes) influence multiple aspects of their host’s phenotype. Over a decade of research shows that the microbiome can vary with both host factors and environmental factors. However, much of the existing literature has been limited to intestinal microbiomes and to laboratory and domesticated animals. Multi-body site and longitudinal analyses of the microbiomes of wild mammals are lacking. Here, I address these gaps in knowledge and use DNA sequencing to survey the microbiomes of a highly gregarious carnivore, the spotted hyena (Crocuta crocuta). Due to their complex societies, spotted hyenas offer an excellent model system for investigating how host physiology and ecology interact with the microbiome, and for elucidating the contributions of the microbiome to host function. In this dissertation, I leverage over three decades of data and samples collected by my adviser from wild hyenas residing in the Masai Mara National Reserve, Kenya (MMNR). Because this dissertation involved many collaborations with other scientists, I use the first person plural throughout this dissertation. In Chapter 1, I evaluate whether the microbiomes at six body-sites vary with host age, sex, and social rank in spotted hyenas, and find that the microbiome is distinct among body sites, and that this differentiation in microbiomes occurs early in life. For Chapter 2, I conduct a longitudinal analysis of the gut microbiome across 3 generations of spotted hyenas from 4 lineages, and elucidate the potential ways gut microbes may be contributing to their host’s digestion of animal carcasses. Findings show that the composition of the gut microbiome is highly variable across time, but its functional repertoire of genes is highly consistent. Furthermore, our analyses reveal that the abundances of bacterial taxa are associated with long-term ecological changes in livestock grazing, anthropogenic disturbance, and herbivore densities that occurred in the Masai Mara reserve. Chapter 3 inquires whether host social interactions and close associations between individuals shape gut microbiota similarity and diversity in a social group of spotted hyenas, which exhibit fission-fusion dynamics. Consistent with our hypothesis, close hyena affiliates share a greater number of bacterial types than hyena dyads that rarely encountered one another, but contrary to our hypothesis, more socially connected individuals do not harbor more diverse gut microbiotas than more isolated individuals. Chapter 4 compares the gut microbiomes of 11 species of sympatric African herbivores from the MMNR and Laikipia region in Kenya, and determines the relative influence of host diet and host phylogenetic relatedness in structuring the microbiome. My findings indicate that across distantly related hosts, herbivore gut microbiotas are strongly shaped by host phylogenetic relatedness and taxonomy, but among closely related hosts, host diet explains the most variation in the gut microbiota. Findings suggest that the gut microbiota is species-specific, but can be further modified by host ecology, including host diet and geography, especially among closely related host species. Overall, my dissertation provides novel insight regarding the factors shaping the gut microbiome in wild carnivores and herbivores, at individual, group-level, and ecosystem-wide scales. Copyright by CONNIE A. ROJAS 2021 ACKNOWLEDGEMENTS I am immensely proud to be writing this, knowing that I have finally accomplished one of the largest professional goals of my life. I was the first in my family to graduate from college, and now, will be the first in my family to obtain a PhD. I am proud and happy to represent my family, my Oaxacan heritage, and women of color in the academy. The past 6 years were filled with hard-work, terrific experiences, and wins, but were also accompanied by rejection, anxiety, and self-doubt. I doubted whether I was smart enough to accomplish the necessary PhD milestones, whether I had the knowledge to hold an engaging conversion with experts in my field, and whether I would actually finish my dissertation. Well, I did finish! I worked hard, persisted, and received invaluable guidance from mentors, and was supported by friends, family, and my therapist. I was also fortunate to have been fully funded during my PhD tenure, which allowed me to consistently make progress on my research and pursue interests outside of my research. I want to start by thanking my PhD adviser, Dr. Kay E. Holekamp. She has been an incredible adviser, and provided the guidance I needed to accomplish my goals. Whether that was writing a letter of recommendation, providing feedback on my manuscripts, sharing thoughtful advice, helping me obtain funds for research, or meeting with me, Kay was there to help with whatever I needed. I also want to thank Dr. Kevin R. Theis, an assistant professor at the Wayne State University School of Medicine and a member of my dissertation guidance committee whom I considered my “second” adviser”. Like Kay, Kevin was an invaluable resource during my PhD and contributed significantly to the development of my dissertation research and my professional development in general. I want to think him for hosting me at his laboratory during multiple summers, modeling strong microbiome research, offering a wealth of advice, and pushing my work to be the best it could be. I v thank the rest of my dissertation committee, Dr. Tom Getty, Dr. Elena Litchman, and Dr. Laura Smale for assisting me as I met my PhD milestones. Next, I thank Dr. Valeria Souza, from the Institute of Ecology at the Universidad Autonoma de Mexico (UNAM) for hosting me during a semester in her molecular evolution laboratory. In her lab, with the assistance of graduate students Mariette Viladomat and Yocelyn Gutierrez, I learned the fundamentals of metagenomic sequence analysis. With this training, I was able to analyze part of the data from the second chapter of my dissertation. I also thank my other collaborators, Dr. Andrew Winters, Dr. Julie W. Turner, and Dr. Santiago A. Ramirez for providing key microbiome, social network, and phylogenetic analyses, respectively that were central to my dissertation research. Their collaboration made my research possible. I am also grateful to my undergraduate adviser, Dr. Vanja Klepac-Ceraj for working tirelessly with me to get my undergraduate research published, shaping my research directions, and helping me navigate tough career decisions. Lastly, I am immensely grateful for all of the research assistants and graduate students of the Mara Hyena Project; their hard work and contributions ensured that I had access to over 3 decades of behavioral data and samples that formed the foundation of my research. Furthermore, this research would not be possible without the extensive financial and professional development resources I had access to while at MSU. I specially thank the National Science Foundation (NSF), Ford Foundation, NSF BEACON Center for the Study of Evolution in Action, and the MSU Alliances for Graduate Education and the Professoriate (AGEP) for their support. The BEACON Center fully funded 3 of my 4 dissertation projects, awarded me multiple travel grants so I could attend conferences, and had a great community of computational scientists that I was a part of. AGEP also deserves special mention because they connected me to a diverse and welcoming community of scholars at MSU, and provided significant research funding. They also enriched my professional vi development by selecting me to mentor a cohort of scholars for the Summer Research Opportunities Program (SROP), and by electing me to serve as Editor-in-Chief of their scholarly publication (Science Today). I am also grateful for funds from the MSU Graduate School, College of Natural Sciences, the Ecology, Evolution, and Behavior Program, Department of Biology, Office of Inclusion and Intercultural Initiatives, and the Council of Graduate Students, for financially supporting me in some capacity during my PhD. Similarly, I thank the Animal Behavior Society, Society for the Advancement of Chicanos/Hispanics and Native Americans in Science, and the American Society of Mammologists for their financial contributions. Professional and financial resources were key to my success and advancement, but so were the emotional support and encouragement I received from friends and family. Firstly, I thank my family, for always being proud of me, giving me space when things were difficult, and for trying their best to understand my world. They are a large motivating factor in everything I do, and I dedicate this dissertation to them. I want to tell them this directly: Familia, si estan leyendo esto, sepan que los aprecio muchisimo, y que ustedes me han motivado a siempre echarle ganas en todo lo que hago. Mis logros son sus logros y esta tesis doctoral se la dedico a ustedes! Furthermore, I thank my best friend Vanessa Diaz, my therapist, my Michigan Indigena / Chicanx Community Alliance familia, and the many friends I made while at MSU for positively impacting my life. I am happy to share this dissertation with everyone who assisted me in my journey! vii TABLE OF CONTENTS LIST OF TABLES ................................................................................................................. xii LIST OF FIGURES ............................................................................................................... xv INTRODUCTION .................................................................................................................... 1 REFERENCES ............................................................................................................ 5 CHAPTER ONE: BODY-SITE SPECIFIC MICROBIOTA REFLECT SEX AND AGE- CLASS AMONG WILD SPOTTED HYENAS ....................................................................... 10 ABSTRACT ............................................................................................................... 10 INTRODUCTION ....................................................................................................... 11 METHODS ................................................................................................................ 13 Behavioral and demographic data collection ................................................. 13 Sample collection ........................................................................................... 14 DNA extractions & 16S rRNA gene sequencing ............................................ 14 Sequence processing ..................................................................................... 15 Microbial community composition analyses ................................................... 16 Analyses of Alpha-diversity ............................................................................ 16 Analyses of Beta-diversity .............................................................................. 17 Data Availability .............................................................................................. 17 RESULTS .................................................................................................................. 18 Microbiota composition is body-site specific .................................................. 18 Microbiota α-diversity and β-diversity vary among body-sites ....................... 20 Variation in microbiota profiles is significantly associated with host sex and age-class ........................................................................................................ 23 Microbiota do not vary with host social rank, but are distinct among individuals ...................................................................................................... 26 DISCUSSION ............................................................................................................ 26 Principal findings of the study ........................................................................ 26 Ecological theory and niche structuring of the microbiota in spotted hyenas . 27 Why are the ear and oral microbiota highly diverse in spotted hyenas? ........ 27 Microbiota composition of spotted hyenas compared to those of other mammals ........................................................................................................ 28 Microbiota of juvenile spotted hyenas vary with host sex but not social rank 29 Differences between juvenile and adult microbiota profiles ........................... 30 Individual identity predicts microbiota profiles in hyenas and other mammals ....................................................................................................................... 30 APPENDICES ....................................................................................................................... 32 APPENDIX A: SUPPLEMENTAL TABLES ............................................................... 33 APPENDIX B: SUPPLEMENTAL FIGURES ............................................................. 41 REFERENCES ..................................................................................................................... 44 CHAPTER TWO: GUT MICROBIOME STRUCTURE AND FUNCTION VARIES WITH HOST ECOLOGY ACROSS MULTIPLE GENERATIONS OF WILD SPOTTED HYENAS 57 ABSTRACT ............................................................................................................... 57 INTRODUCTION ....................................................................................................... 58 viii METHODS ................................................................................................................ 61 Study site and study animals ......................................................................... 61 Sample and meta data collection ................................................................... 63 DNA extractions and genomic sequencing of amplicon and metagenomic reads .............................................................................................................. 64 Sequence processing of amplicon reads ....................................................... 65 Assessing temporal variability in gut microbiota composition ........................ 66 Identifying the host predictors of gut microbiota structure (b-diversity) .......... 67 Measuring gut microbiota stability within individuals ...................................... 67 Sequence processing of metagenomic reads ................................................ 68 Taxonomic and functional analyses of metagenomic reads .......................... 69 Data availability .............................................................................................. 69 Ethical approval .............................................................................................. 70 RESULTS .................................................................................................................. 70 Did the composition of the gut microbiota fluctuate across the two decades of sampling, and were changes associated with prey availability? .................... 70 Is host ecology a stronger predictor of gut microbiota similarity than host kinship or age? ............................................................................................... 75 Do gut microbiome functional profiles reveal adaptations to a carnivorous diet? ............................................................................................................... 77 Are gut microbiome functional profiles more stable than taxonomic profiles over short time-scales? .................................................................................. 80 DISCUSSION ............................................................................................................ 81 Principal findings of study .............................................................................. 81 Gut microbiota composition is highly plastic over the two decades of sampling ....................................................................................................................... 82 Local environmental conditions predict gut microbiota variation in wild spotted hyenas ............................................................................................................ 84 Gut microbiota composition reflects its host’s carnivorous diet ...................... 86 The gut microbiome of hyenas is implicated in host digestion and in host protection from pathogens ............................................................................. 88 Gut microbiome functional profiles are stable, yet flexible among individuals over time ........................................................................................................ 90 Limitations of the current study ...................................................................... 91 CONCLUSIONS ........................................................................................................ 92 APPENDICES ....................................................................................................................... 93 APPENDIX A: SUPPLEMENTAL TABLES ............................................................... 94 APPENDIX B: SUPPLEMENTAL FIGURES ........................................................... 100 REFERENCES ................................................................................................................... 104 CHAPTER THREE: DO HOST SOCIAL INTERACTIONS PREDICT GUT MICROBIOTA SIMILARITY AND DIVERSITY IN A FISSION-FUSION MAMMALIAN SOCIETY? .......... 115 ABSTRACT ............................................................................................................. 115 INTRODUCTION ..................................................................................................... 116 METHODS .............................................................................................................. 119 Study location and sampling ........................................................................ 119 Construction of social networks ................................................................... 120 Calculating hyena ‘sociability’ ...................................................................... 121 ix DNA extractions and 16S rRNA gene sequencing of fecal samples ............ 122 Characterizing gut microbiota profiles .......................................................... 123 Combined gut microbiota and social network analyses ............................... 124 Data availability ............................................................................................ 125 Ethical approval ............................................................................................ 126 RESULTS ................................................................................................................ 126 Gut microbiota composition across the eight sampled time-periods ............ 126 Effects of social interactions on gut microbiota similarity ............................. 128 Effects of individual sociability on gut microbiota alpha-diversity ................. 130 DISCUSSION .......................................................................................................... 132 Gut microbiota similarity ............................................................................... 132 Gut microbiota diversity ................................................................................ 134 CONCLUSIONS ...................................................................................................... 135 APPENDICES ..................................................................................................................... 137 APPENDIX A: SUPPLEMENTAL TABLES ............................................................. 138 APPENDIX B: SUPPLEMENTAL FIGURES ........................................................... 140 REFERENCES ................................................................................................................... 141 CHAPTER FOUR: HOST PHYLOGENY AND HOST ECOLOGY STRUCTURE THE MAMMALIAN GUT MICROBIOTA AT DIFFERENT TAXONOMIC SCALES ................... 148 ABSTRACT ............................................................................................................. 148 INTRODUCTION ..................................................................................................... 149 METHODS .............................................................................................................. 151 Study location and sampling ........................................................................ 151 DNA extraction and 16S rRNA gene sequencing ........................................ 152 Sequence processing and bioinformatics .................................................... 153 Microbiota composition analyses ................................................................. 153 Microbiota α-diversity statistical analyses .................................................... 154 Microbiota β-diversity analyses and testing for phylosymbiosis ................... 155 Comparisons of Masai Mara and Laikipia herbivores .................................. 157 Ethical Approval ........................................................................................... 158 Availability of data and materials .................................................................. 158 RESULTS ................................................................................................................ 159 Aim 1: Determine the strongest predictor of gut microbiota similarity among herbivore hosts at greater and lesser taxonomic scales .............................. 159 Aim 2: Evaluate the influences of host taxonomy and host dietary guild on gut microbiota composition and diversity ........................................................... 163 Aim 3: Examine the amount variance in the gut microbiota explained by geographic region among conspecific hosts ................................................ 169 DISCUSSION .......................................................................................................... 172 Principal findings of study ............................................................................ 172 Aim 1: Determine the strongest predictor of gut microbiota similarity among herbivore hosts at greater and lesser taxonomic scales .............................. 173 Aim 2: Evaluate the influences of host taxonomy and host dietary guild on gut microbiota composition and diversity ........................................................... 175 Aim 3: Examine the amount variance in the gut microbiota explained by geographic region among conspecific hosts ................................................ 177 CONCLUSIONS ...................................................................................................... 178 APPENDICES ..................................................................................................................... 180 x APPENDIX A: SUPPLEMENTAL TABLES ............................................................. 181 APPENDIX B: SUPPLEMENTAL FIGURES ........................................................... 189 REFERENCES ................................................................................................................... 192 xi LIST OF TABLES Table 1.1 Body-sites vary in their microbiota richness and evenness (a-diversity) ............. 20 Table 1.2 Juvenile female hyenas have distinct anal scent gland microbiota compared to juvenile male hyenas ............................................................................................................ 23 Table 1.3 Adult female hyenas have distinct microbiota compared to juvenile female hyenas .................................................................................................................................. 25 Table S1.1 Microbiota α-diversity among body-sites in adult and juvenile spotted hyenas (mean ± SD) ......................................................................................................................... 33 Table S1.2 List of taxa enriched in particular body-sites in adult spotted hyenas as determined by LEfSe ............................................................................................................ 34 Table S1.3 List of taxa enriched in particular body-sites in juvenile spotted hyenas as determined by LEfSe ............................................................................................................ 35 Table S1.4 Multiple-comparison testing of body-site microbiota alpha-diversity values in adults .................................................................................................................................... 37 Table S1.5 Multiple-comparison testing of body-site microbiota alpha-diversity values in juveniles ................................................................................................................................ 38 Table S1.6 Body-sites vary in their community dispersion in adults and juveniles .............. 39 Table 2.1 Longitudinal analysis of the hyena gut microbiota ............................................... 61 Table 2.2 Environmental conditions are the strongest predictors of gut microbiota similarity in adult female hyenas .......................................................................................................... 75 Table 2.3 Gut metagenome taxonomic and functional profiles are individual-specific and vary with host ecology .......................................................................................................... 81 Table S2.1 Distribution of samples from each hyena individual included in our longitudinal study ..................................................................................................................................... 94 Table S2.2 Statistics of the metagenome assemblies as determined by QUAST (v5.0.0) .. 95 Table S2.3 Temporal variation in the mean relative abundances of bacterial orders and ASVs ..................................................................................................................................... 95 Table S2.4 Number of bacterial ASVs retained over time in the gut microbiota of adult female hyenas ...................................................................................................................... 97 Table S2.5 Predictors of gut microbiota structure in adult female hyenas (N=301) ............. 97 xii Table S2.6 The abundances of KEGG Bacterial Pathways vary temporally and with prey availability ............................................................................................................................. 98 Table S2.7 The 45 most abundant KEGG functional genes in the hyena gut microbiome .. 98 Table 3.1 Breakdown of the samples included in this study .............................................. 119 Table 3.2 Do social bonds between hyena dyads predict their gut microbiota similarity? . 129 Table 3.3 Do more social hyenas harbor more diverse gut microbiotas? .......................... 131 Table S3.1 Samples included in this study and their relevant meta data ........................... 138 Table S3.2 Results of PERMANOVA model relating host factors to gut microbiota similarity ............................................................................................................................................ 138 Table S3.3 Mean±SE of the two microbiota similarity metrics calculated for the dyadic comparisons included in this study ..................................................................................... 138 Table S3.4 Mean±SE of social bond strength between hyena dyads included in this study ............................................................................................................................................ 139 Table S3.5 Mean±SE of the two gut microbiota α-diversity metrics calculated for hyena individuals included in this study ......................................................................................... 139 Table S3.6 Mean±SE of three network centrality metrics calculated for hyena individuals included in this study .......................................................................................................... 139 Table 4.1 List of host study species and their associated metadata .................................. 151 Table 4.2 The relative contributions of host phylogenetic relatedness and diet in predicting gut microbiota similarity ...................................................................................................... 160 Table 4.3 Host taxonomy and dietary guild shape the gut microbiotas of African herbivores ............................................................................................................................................ 163 Table 4.4 Microbiota richness, evenness, and phylogenetic diversity vary with host taxonomy and dietary guild ................................................................................................. 168 Table S4.1 Dietary C4 (%) data for the 11 herbivore species included in this study ......... 181 Table S4.2 Influences of host species and diet on the gut microbiota across 11 species of herbivores ........................................................................................................................... 181 Table S4.3 Multiple-comparison testing of gut microbiota alpha-diversity among host families ................................................................................................................................ 182 xiii Table S4.4 Multiple-comparison testing of gut microbiota alpha-diversity among host dietary guilds .................................................................................................................................. 183 Table S4.5 Multiple-comparison testing of gut microbiota alpha-diversity among host species (all study species) .................................................................................................. 184 Table S4.6 Multiple-comparison testing of gut microbiota alpha-diversity among host species in bovids ................................................................................................................ 185 Table S4.7 Influences of host species and diet on the gut microbiota remain despite differences in host geography ............................................................................................ 187 Table S4.8. The relative contributions of host phylogenetic relatedness and diet in predicting gut microbiota similarity in Masai Mara and Laikipia herbivores ........................ 187 Table S4.8 Percent of ASVs differentially abundant between Masai Mara and Laikipia herbivore populations ......................................................................................................... 188 Table S4.9 Sample distribution for the combined Masai Mara and Laikipia dataset .......... 188 xiv LIST OF FIGURES Figure 1.1 Microbiota composition at multiple body sites in adult and juvenile spotted hyenas .................................................................................................................................. 19 Figure 1.2 Body sites vary in their microbiota a-diversity .................................................... 21 Figure 1.3 Microbiota cluster by body site in spotted hyenas .............................................. 21 Figure 1.4 Microbiota structure and dispersion across body sites in juvenile and adult hyenas .................................................................................................................................. 22 Figure 1.5 Juvenile females and juvenile males have distinct scent gland microbiota ........ 24 Figure 1.6 Adult females and juvenile females have distinct prepuce and rectal microbiota ............................................................................................................................. 25 Figure S1.1 Rarefaction curves of microbial community richness across body-sites in spotted hyenas ..................................................................................................................... 41 Figure S1.2 Body-sites vary in their microbiota composition ............................................... 42 Figure S1.3 Top 21 most abundant bacterial genera inhabiting multiple body-sites in adult and juvenile spotted hyenas ................................................................................................. 43 Figure S1.4 Bacterial community dispersion in adult female hyenas compared to juvenile female hyenas ...................................................................................................................... 43 Figure 2.1 Temporal variability in the gut microbiota composition of wild spotted hyenas .. 72 Figure 2.2 The hyena gut microbiota is not stable within individuals over time ................... 74 Figure 2.3 Host socioecological determinants of gut microbiota similarity in wild spotted hyenas .................................................................................................................................. 76 Figure 2.4 Taxonomic and functional profiles of hyena gut metagenomes ......................... 79 Figure S2.1 Relative abundances of ASVs from DNA extraction kit control samples ....... 100 Figure S2.2 Rarefaction curves of gut microbiota ASV richness ....................................... 100 Figure S2.3 Predominant bacterial families of the hyena gut microbiome ......................... 101 Figure S2.4 The relative abundances of bacterial ASVs vary temporally .......................... 102 Figure S2.5 Gut microbiome similarity varies with year along PCoA axis 1 ...................... 103 xv Figure 3.1 Temporal variation in the gut microbiota composition and structure of wild spotted hyenas ................................................................................................................... 127 Figure 3.2 Association networks of the adult female hyenas included in this study .......... 128 Figure 3.3 Hyenas that associate more frequently share more bacterial types, but do not harbor similar abundances of these bacterial types ........................................................... 130 Figure 3.4 Gut microbiota richness as a function of hyena sociability ............................... 132 Figure S3.1 Predominant bacterial phyla in the hyena gut microbiota .............................. 140 Figure 4.1 African herbivore gut microbiotas exhibit patterns consistent with phylosymbiosis ............................................................................................................................................ 161 Figure 4.2 Gut microbiota composition of African herbivores ............................................ 166 Figure 4.3 Host taxonomy and dietary guild are associated with gut microbiota diversity in African herbivores ............................................................................................................... 168 Figure 4.4 The gut microbiotas of conspecific African herbivores broadly converge, but also exhibit differences in their ASV abundances ...................................................................... 171 Figure S4.1 Predominant bacterial phyla of African herbivore gut microbiotas ................. 189 Figure S4.2 Predominant bacterial genera of African herbivore gut microbiotas .............. 190 Figure S4.3 Relative abundance of 10 ASVs widely shared across host species ............. 190 Figure S4.4 Sample distribution by Month for Masai Mara herbivores .............................. 191 Figure S4.5 Rarefaction curves of gut microbiota ASV richness ....................................... 191 xvi KEY TO ABBREVIATIONS ASV Amplicon Sequence Variant DADA2 Divisive Amplicon Denoising Algorithm GLM Generalized linear model IACUC Institutional Animal Care and Use Committee LMM Linear mixed model LEfSe Linear discriminant analysis Effect Size MYA Millions of Years Ago NCBI National Center for Biotechnology Information OTU Operational taxonomic unit PCoA Principal Coordinates Analysis PD Faith’s Phylogenetic Diversity PERMANOVA Permutational Multivariate Analysis of Variance PERMDISP Permutation tests of multivariate dispersions RDP Ribosomal Database Project SCFA Short Chain Fatty Acids SRI Simple Ratio Association index UIC University of Illinois at Chicago xvii INTRODUCTION Organisms function not as single entities, but as holobionts; they are hosts that have dynamic interactions with their microbial communities, hereafter termed the microbiome (or microbiota) [1, 2]. A decade of research in the nascent field of host-microbe interactions demonstrates that hosts can alter their microbiome [3, 4], and microbes can influence their hosts [5–8], both of which have functional consequences for the holobiont as a whole. Host–microbe interactions shape host physiology, immunity, and behavior in a variety of animal systems, and may be fundamental to their host’s evolution [9–11]. Symbiotic gut microbes for example, may have enabled their hosts to access novel food resources and niches [10, 12]. However, outside of laboratory animals and primates, the causes and consequences of microbiome variation are not widely understood. In these less heavily studied systems, we have a limited understanding of the a) relative contributions of host physiological, behavioral, and environmental factors in shaping the microbiome, and b) the ways the microbiome potentially affects their host’s phenotype. In this dissertation, I address these gaps in knowledge, and combine over three decades of field data with genomic sequencing to profile the microbiome at multiple body sites in wild spotted hyenas residing in the Masai Mara National Reserve, Kenya (MMNR). Spotted hyenas are highly social carnivores that live in large, dynamic, and matriarchal societies [13] which offer an excellent model system for investigating how host physiology, behavior, and ecology interact with the microbiome, and for elucidating the contributions of the microbiome to host function. Microbiome variation can be examined at distinct scales: at the individual level, group level, ecosystem level, and at times, at a broader evolutionary level. Within individuals, the microbiome appears to be niche-specific, and can vary across major body- sites (e.g. oral cavity and skin) [14–18] or across locations within a body-site (e.g. jejunum, 1 ileum, and cecum) [19–21]. Additionally, the microbiome, particularly that of the gastrointestinal tract, can vary with aging, reproductive state, body condition, disease, and daily activity [22–28], suggesting that the microbiome is affected by changes in host physiology. However, it is unknown to the extent that these findings are observed in non- model organisms and wild mammals, and in body regions other than the gut. Thus, for Chapter 1 of my dissertation, I survey the microbiome at six distinct body niches in wild hyenas (ears, nares, oral cavity, prepuce, rectum and scent gland) and investigate whether the microbiome varies among body-sites, and within body-sites, and whether it is distinct between juveniles and adults, and males and females. These groups of hyenas possess distinct physiologies which could translate to differences in microbiome composition. Across and within individuals, microbiome composition tends to be highly-variable, and can vary with host identity, lifespan, and changes in the external environment, including changes in host diet, external temperature, and altitude [29–31]. However, this high variability in microbiome composition can be accompanied by high stability in microbiome function due to the widespread functional redundancy among microbes, as has been observed in humans [32, 33]. These two features of the microbiome are thought to be functionally adaptive for the holobiont; plasticity in gut microbiome composition may enable the holobiont to respond to changes in the environment (e.g. seasonal change in food source) while functional redundancy in microbial metabolisms may buffer microbiome function against disturbances (e.g. disease). Thus, due to the dynamism of the gut microbiome, longitudinal studies and repeated sampling of individuals are needed to provide a more complete picture of drivers of microbiome structure and function. Nonetheless, studies assessing temporal variability in gut microbiome composition concurrently with function across multiple decades in wild mammals are nonexistent. In Chapter 2 of this dissertation, I leverage over two decades of data and samples to quantify 2 variability in gut microbiome composition and function within individuals across time and across generations in a wild population of spotted hyenas. Additionally, I investigate whether host maternal relatedness, age, prey abundance, and local environmental conditions predict microbiome composition in adult females. Lastly, I determine whether the gut microbiome is enriched in genes implicated in the host digestion of their herbivore prey. Social animals like spotted hyenas, engage in a diversity of affiliative and aggressive interactions with conspecifics, which may provide pathways for microbial transmission between individuals. Because social interactions are known to affect pathogen infection rates and dynamics, they are likely also influencing the acquisition and transmission of beneficial and commensal microbes [34–36]. Indeed, in highly social primate species, the gut microbiome varies with host social group, and is most similar among group members that associate or groom frequently [37, 38]. Furthermore, in humans, individuals with greater levels of social interactions tend to have more diverse microbiomes [39]. However, the extent to which social interactions influence the gut microbiome in gregarious large carnivores is unknown. Chapter 3 of my dissertation directly addresses this question, ascertains whether host social interactions shape gut microbiome composition and diversity in adult female hyenas from a single clan. Furthermore, unlike many primate species, hyenas live in flexible fission-fusion societies [40, 41], where clan members form small groups (“fission”) that may later aggregate to form larger groups (“fusion”). It is unknown whether these fission-fusion dynamics will affect our expected patterns. Across larger host taxonomic scales (e.g. sponges to elephants), gut microbiome composition varies with host species and phylogenetic relatedness, such that closely related host species often have more similar gut microbiomes than more distantly related host species [42–46], a pattern termed phylosymbiosis [47–49]. Phylosymbiosis can arise when closely related host species provide potential colonizing microbes with similar 3 ecological niches. This can occur if closely related host species possess similar diets, physiologies and immune components, reside in similar habitats, or exhibit similar behaviors. Thus, to study phylosymbiosis, we need to disentangle the relative contributions of host phylogenetic relatedness and host ecology in shaping the microbiome. In Chapter 4 of my dissertation, I accomplish that goal and examine the potential influences of phylogenetic relatedness and host diet on the gut microbiome of 11 species of sympatric savanna herbivores residing in the MMNR. I also compare the microbiomes of these Mara herbivores to those from conspecifics in central Kenya to determine whether host geography and/or local habitat influence gut microbiota composition and patterns of phylosymbiosis. Collectively, my body of work will demonstrate the extent to which physiological, behavioral, and ecological forces potentially shape the gut microbiome of wild mammals and will further our understanding of the universal forces associated with microbiome variation across animal systems. This dissertation also examines the influences of the gut microbiome at distinct scales, from individuals to social groups to ecosystems. Novel findings from this research will show that the microbiome in spotted hyenas is body-site specific and this niche differentiation of the microbiome is present early in life. My research will constitute the longest survey of gut microbiome dynamics in a wild mammal (23 years and three generations). Lastly, my research will contribute to the literature and demonstrate the importance of host ecology in shaping the gut microbiome. 4 REFERENCES 5 REFERENCES 1. Bordenstein SR, Theis KR. Host biology in light of the microbiome: Ten principles of holobionts and hologenomes. PLoS Biol. 2015;13:1–23. 2. Theis KR, Dheilly NM, Klassen JL, Brucker RM, Baines JF, Bosch TCG, et al. Getting the Hologenome Concept Right: an Eco-Evolutionary Framework for Hosts and Their Microbiomes. mSystems. 2016;1:e00028-16. doi:10.1128/mSystems.00028-16. 3. Cho I, Yamanishi S, Cox L, Methé BA, Zavadil J, Li K, et al. Antibiotics in early life alter the murine colonic microbiome and adiposity. Nature. 2012;488:621–6. doi:10.1038/nature11400. 4. Daniel H, Gholami AM, Berry D, Desmarchelier C, Hahne H, Loh G, et al. High-fat diet alters gut microbiota physiology in mice. ISME J. 2014;8:295–308. doi:10.1038/ismej.2013.155. 5. Grisnik M, Bowers O, Moore AJ, Jones BF, Campbell JR, Walker DM. The cutaneous microbiota of bats has in vitro antifungal activity against the white nose pathogen. FEMS Microbiol Ecol. 2020;96:1–11. 6. Kohl KD, Oakeson KF, Orr TJ, Miller AW, Forbey JS, Phillips CD, et al. Metagenomic sequencing provides insights into microbial detoxification in the guts of small mammalian herbivores (Neotoma spp.). FEMS Microbiol Ecol. 2018;94:1–10. 7. Sommer F, Stå M, Ilkayeva O, Newgard CB, Frö O, Bä Ckhed Correspondence F, et al. The Gut Microbiota Modulates Energy Metabolism in the Hibernating Brown Bear Ursus arctos. Cell Rep. 2016;14:1655–61. doi:10.1016/j.celrep.2016.01.026. 8. Bäckhed F, Ding H, Wang T, Hooper L V., Gou YK, Nagy A, et al. The gut microbiota as an environmental factor that regulates fat storage. Proc Natl Acad Sci U S A. 2004;101:15718–23. 9. Kolodny O, Schulenburg H. Microbiome-mediated plasticity directs host evolution along several distinct time-scales. Philos Trans R Soc B2. 2020;375:20190589. 10. Moeller AH, Sanders JG. Roles of the gut microbiota in the adaptive evolution of mammalian species. Philos Trans R Soc B2. 2020;375:20190597. 11. Sharpton TJ. Role of the Gut Microbiome in Vertebrate Evolution. 2018;3:1–5. 12. Voolstra CR, Ziegler M. Adapting with microbial help: Microbiome flexibility facilitates rapid responses to environmental change. BioEssays. 2020;2000004:1–9. 13. Holekamp KE, Smith JE, Strelioff CC, Van Horn RC, Watts HE. Society, demography and genetic structure in the spotted hyena. Mol Ecol. 2012;21:613–32. 14. Cheng Y, Fox S, Pemberton D, Hogg C, Papenfuss AT, Belov K. The Tasmanian devil microbiome-implications for conservation and management. Microbiome. 2015;3:76. doi:10.1186/s40168-015-0143-0. 15. Bik EM, Costello EK, Switzer AD, Callahan BJ, Holmes SP, Wells RS, et al. Marine mammals harbor unique microbiotas shaped by and yet distinct from the sea. Nat Commun. 2016;7:10516. doi:10.1038/ncomms10516. 6 16. Alfano N, Courtiol A, Vielgrader H, Timms P, Roca AL, Greenwood AD. Variation in koala microbiomes within and between individuals: effect of body region and captivity status. Sci Rep. 2015;5:10189. doi:10.1038/srep10189. 17. Chen Z, Yeoh YK, Hui M, Wong PY, Chan MCW, Ip M, et al. Diversity of macaque microbiota compared to the human counterparts. Sci Rep. 2018;8:15573. doi:10.1038/s41598- 018-33950-6. 18. De Paepe AE, Sierpowska J, Garcia-Gorro C, Martinez-Horta S, Perez-Perez J, Kulisevsky J, et al. Stable and transient structural variation in lemur vaginal, labial, and axillary microbiomes: Patterns by species, body site, ovarian hormones, and forest access. J Chem Inf Model. 2019;53:1689–99. 19. He J, Yi L, Hai L, Ming L, Gao W, Ji R. Characterizing the bacterial microbiota in different gastrointestinal tract segments of the Bactrian camel. Sci Rep. 2018;8:1–9. 20. Kohl KD, Dearing MD, Bordenstein SR. Microbial communities exhibit host species distinguishability and phylosymbiosis along the length of the gastrointestinal tract. Mol Ecol. 2018;1:1–17. 21. Yasuda K, Oh K, Ren B, Tickle TL, Franzosa EA, Wachtman LM, et al. Biogeography of the intestinal mucosal and lumenal microbiome in the rhesus macaque. Cell Host Microbe. 2015;17:385–91. doi:10.1016/j.chom.2015.01.015. 22. Theis KR, Venkataraman A, Wagner AP, Holekamp KE, Schmidt TM. Age-related variation in the scent pouch bacterial communities of striped hyenas. In: Schulte BA, Goodwin TE, Ferkin MH, editors. Chemical Signals in Vertebrates 13. New York, NY: Springer; 2016. doi:10.1007/978-3-319-22026-0. 23. Tian J, Du J, Han J, Song X, Lu Z. Age-related differences in gut microbial community composition of captive spotted seals (Phoca largha). Mar Mammal Sci. 2020; May:1–10. 24. Lim MY, Song EJ, Kang KS, Nam Y. Age-related compositional and functional changes in micro-pig gut microbiome. GeroScience. 2019;41:935–44. 25. Suzuki TA, Martins FM, Phifer-Rixey M, Nachman MW. The gut microbiota and Bergmann’s rule in wild house mice. Mol Ecol. 2020;7071601132. 26. Chun JL, Ji SY, Lee SD, Lee YK, Kim B, Kim KH. Difference of gut microbiota composition based on the body condition scores in dogs. 2020;62:239–46. 27. DeCandia AL, Leverett KN, vonHoldt BM. Of Microbes and Mange: Consistent changes in the skin microbiome of three canid species infected with sarcoptic mange. bioRxiv. 2019;:709436. doi:10.1101/709436. 28. Vlckova K, Pafco B, Petrzelkova K, Modrý D, Todd A, Yeoman CJ, et al. Relationships Between Gastrointestinal Parasite Infections and the Fecal Microbiome in Free-Ranging Western Lowland Gorillas. Front Microbiol. 2018;9:1202. doi:10.3389/FMICB.2018.01202. 29. Zhu L, Wu Q, Deng C, Zhang M, Zhang C, Chen H, et al. Adaptive evolution to a high purine and fat diet of carnivorans revealed by gut microbiomes and host genomes. Environ Microbiol. 7 2018;20:1711–22. 30. Longo A V., Zamudio KR. Temperature variation, bacterial diversity and fungal infection dynamics in the amphibian skin. Mol Ecol. 2017;26:4787–97. 31. Zeng B, Zhang S, Xu H, Kong F, Yu X, Wang P, et al. Gut microbiota of Tibetans and Tibetan pigs varies between high and low altitude environments. Microbiol Res. 2020;235 November 2019:126447. doi:10.1016/j.micres.2020.126447. 32. Hollister EB, Riehle K, Luna RA, Weidler EM, Rubio-Gonzales M, Mistretta TA, et al. Structure and function of the healthy pre-adolescent pediatric gut microbiome. Microbiome. 2015;3:36. doi:10.1186/s40168-015-0101-x. 33. Huttenhower C, Gevers D, Knight R, Abubucker S, Badger JH, Chinwalla AT, et al. Structure, function and diversity of the healthy human microbiome. Nature. 2012;486:207–14. doi:10.1038/nature11234. 34. Sarkar A, Harty S, Johnson KVA, Moeller AH, Archie EA, Schell LD, et al. Microbial transmission in animal social networks and the social microbiome. Nat Ecol Evol. 2020. 35. Balasubramaniam KN, Beisner BA, Hubbard JA, Vandeleest JJ, Atwill ER, McCowan B. Affiliation and disease risk: social networks mediate gut microbial transmission among rhesus macaques. Anim Behav. 2019;151:131–43. doi:10.1016/j.anbehav.2019.03.009. 36. Lombardo MP. Access to mutualistic endosymbiotic microbes: An underappreciated benefit of group living. Behav Ecol Sociobiol. 2008;62:479–97. 37. Perofsky AC, Lewis RJ, Abondano LA, Di Fiore A, Meyers LA. Hierarchical social networks shape gut microbial composition in wild Verreaux’s sifaka. Proceedings Biol Sci. 2017;284. http://www.ncbi.nlm.nih.gov/pubmed/29212730. Accessed 29 Aug 2019. 38. Tung J, Barreiro LB, Burns MB, Grenier JC, Lynch J, Grieneisen LE, et al. Social networks predict gut microbiome composition in wild baboons. Elife. 2015;2015:1–18. 39. Johnson KVA. Gut microbiome composition and diversity are related to human personality traits. Hum Microbiome J. 2020;15 December 2019:100069. doi:10.1016/j.humic.2019.100069. 40. Kruuk H. The spotted hyena : a study of predation and social behavior. Chicago: University of Chicago Press; 1972. 41. Smith JE, Kolowski JM, Graham KE, Dawes SE, Holekamp KE. Social and ecological determinants of fission–fusion dynamics in the spotted hyaena. Anim Behav. 2008;76:619–36. doi:10.1016/J.ANBEHAV.2008.05.001. 42. Li J, Zhan S, Liu X, Lin Q, Jiang J, Li X. Divergence of Fecal Microbiota and Their Associations With Host Phylogeny in Cervinae. Front Microbiol. 2018;9:1–11. doi:10.3389/fmicb.2018.01823. 43. Amato KR, Sanders J, Song SJ, Nute M, Metcalf JL, Thompson LR, et al. Evolutionary trends in host physiology outweigh dietary niche in structuring primate gut microbiomes. ISME J. 2019;13:576–587. doi:10.1038/s41396-018-0175-0. 8 44. Nishida AH, Ochman H. Rates of gut microbiome divergence in mammals. Mol Ecol. 2018;27:1884–97. 45. Youngblut ND, Reischer GH, Walters W, Schuster N, Walzer C, Stalder G, et al. Host diet and evolutionary history explain different aspects of gut microbiome diversity among vertebrate clades. Nat Commun. 2019;10:1–15. doi:10.1038/s41467-019-10191-3. 46. Moeller AH, Suzuki TA, Lin D, Lacey EA, Wasser SK, Nachman MW. Dispersal limitation promotes the diversification of the mammalian gut microbiota. PNAS. 2017;114:13768–73. doi:10.1073/pnas.1700122114. 47. Kohl KD. Ecological and evolutionary mechanisms underlying patterns of phylosymbiosis in host-associated microbial communities. Philos Trans R Soc B Biol Sci. 2020;375. 48. Lim SJ, Bordenstein SR. An introduction to phylosymbiosis. Proc R Soc B Biol Sci. 2020;287. 49. Brooks AW, Kohl KD, Brucker RM, Van Opstal EJ, Bordenstein SR. Phylosymbiosis: Relationships and Functional Effects of Microbial Communities across Host Evolutionary History. PLoS Biol. 2016;14:1–19. doi:10.1371/journal.pbio.2000225. 9 BODY-SITE SPECIFIC MICROBIOTA REFLECT SEX AND AGE-CLASS AMONG WILD CHAPTER ONE: SPOTTED HYENAS PUBLISHED AS: Rojas, C.A., K.E. Holekamp, A.D. Winters, and K.R. Theis. 2020. Body-site specific microbiota reflect sex and age class among wild spotted hyenas. FEMS Microbiology Ecology: 96(2). https://doi.org/10.1093/femsec/fiaa007 ABSTRACT Host-associated microbial communities, henceforth ‘microbiota’, can affect the physiology and behavior of their hosts. In mammals, host ecological, social, and environmental variables are associated with variation in microbial communities. Within individuals in a given mammalian species, the microbiota also partitions by body-site. Here, we build on this work and sequence the bacterial 16S rRNA gene to profile the microbiota at six distinct body-sites (ear, nasal and oral cavities, prepuce, rectum, and anal scent gland) in a population of wild spotted hyenas (Crocuta crocuta), which are highly social, large African carnivores. We inquired whether microbiota at these body-sites vary with host sex or social rank among juvenile hyenas, and whether they differ between juvenile females and adult females. We found that the scent gland microbiota differed between juvenile males and juvenile females, whereas the prepuce and rectal microbiota differed between adult females and juvenile females. Social rank, however, was not a significant predictor of microbiota profiles. Additionally, the microbiota varied considerably among the six sampled body-sites and exhibited strong specificity among individual hyenas. Thus, our findings suggest that site-specific niche selection is a primary driver of microbiota structure in mammals, but endogenous host factors may also be influential. 10 INTRODUCTION Animal bodies are home to structurally and functionally organized microbial communities, termed ‘microbiota’, which can strongly affect their host’s physiology, behavior and fitness [1–3]. For example, in the gastrointestinal tracts of folivorous and myrmecophagous mammals [4, 5] resident microbes can convert tough compounds (i.e. cellulose and chitin) into readily available nutrients and energy for their host. In multiple species of carnivores [6–9], microbiota inhabiting scent gland secretions co-vary with the gland’s odorous biochemical profiles and include odor-producing bacteria, indicating that microbes likely contribute to their host’s chemical signals. Furthermore, in insectivorous bats, bacteria from the skin exhibit anti-fungal properties against the pathogen that causes white-nose syndrome, implicating these microbes in the pathogen defenses of their hosts [10–12]. Thus, resident microbes and their genomes, collectively referred to as the ‘microbiome,’ are functionally important in shaping the phenotypes of their hosts. Hence, identifying the environmental and host factors that affect variation in the microbiota and microbiome, and asking how this may affect host phenotype, are key lines of inquiry in host- microbial ecology [13–15]. Numerous environmental, social and physiological host factors are associated with variation in the microbiota within and among mammalian host species. Among the key drivers of variation in the mammalian microbiota, particularly the gut microbiota, are host diet and phylogeny [16–20]. The microbiota might also be sensitive to the host’s social and ecological environment, as some variation in the microbiota can be attributed to their host’s social interactions, season, habitat, and geography [21–25] Lastly, microbiota are also known to vary with physiological and host-specific factors such as genetic variation, sex, age, and individual identity [7, 26–29]. Within individual hosts however, microbiota are often strongly structured by anatomical body region. In a range of mammalian hosts including 11 humans, primates, marine mammals, marsupials, bats, carnivores, and domestic animals, the microbiota vary among body-sites [4, 30–37]. Host-associated microbial communities may also vary within specific regions of a body-site, as has been demonstrated for the mammalian gastrointestinal tract (e.g. stomach, small intestine, cecum, large intestine, feces) [38–41], the skin (e.g. axillae, nose, ears, digits, limbs) [29, 42, 43], and the oral cavity (e.g. gums, molars, plaque) [37]. Here, we build upon this prior work by using 16S rRNA gene sequencing to characterize the diversity and structure of host-associated microbiota at six distinct body- sites in a gregarious large carnivore, the spotted hyena (Crocuta crocuta). Spotted hyenas inhabit much of sub-Saharan Africa [44] and live in social groups, called ‘clans.’ Clans may contain over 90 individuals, and usually consist of multiple overlapping generations of natal females and their offspring, along with a few immigrant males. Their societies are structured by linear dominance hierarchies, in which an individual’s position within the hierarchy determines its priority of access to resources [45, 46]. Hyena societies also characterized by female dominance, male-biased dispersal, and a high degree of fission-fusion dynamics, such that individuals move among subgroups several times per day [45–48]. Hyenas are matrilineal, and each new offspring inherits the rank immediately below that of its mother but above those of its older siblings [49, 50]. Spotted hyenas bear litters of one or two cubs, which are reared at communal dens for the first 9 to 12 mo of life; they are weaned at 12-18 mo, and reach reproductive maturity at 24 mo, although most females do not bear offspring of their own until they are at least 36 mo of age [51, 52]. To communicate, spotted hyenas utilize signals via multiple sensory modalities, including a rich repertoire of vocalizations [53] and odorous secretions from their scent glands [6, 54–58]. Specifically, in this study, we inquire whether body-site specificity of the microbiota is observed in adult and juvenile hyenas. We also evaluate whether these bacterial 12 communities vary with host sex or social rank among juvenile hyenas. Lastly, we investigate whether the microbiota differs between juvenile females and adult females at each body- site. Prior research on spotted hyenas has shown that the anal scent gland microbiota varies with host sex, social group, and reproductive state [6], and that gut microbiota diversity varies with host age [59]. However, we know little about the microbiota occupying other hyena body regions. Thus, our study will help establish a baseline understanding of the microbiota occupying six anatomical body-sites of a large carnivore species in its natural habitat. METHODS Behavioral and demographic data collection We identified individual hyenas by their unique spot patterns, determined their sex based on phallic morphology [60], and calculated birthdates to ± 7 days based on the appearance of cubs when first observed [51]. We defined juveniles as hyenas less than 24 months old, and older animals were considered to be adults. Hyenas were assigned a dominance rank based on their position in a matrix ordered by submissive behaviors displayed during dyadic agonistic encounters [61]. In our analyses, social rank was normalized (to values between 0 and 1), such that the highest-ranking hyena had a value of 1, and the lowest ranking-hyena had a value of 0. Juveniles were assigned the same ranks as their mothers. Our statistical power was insufficient to test whether the microbiota of adult females varied with clan membership or social rank, so these samples were pooled and only used when determining whether the microbiota varied among body-sites or age- classes. Sample metadata can be accessed online at https://academic.oup.com/femsec/article/96/2/fiaa007/5700710?searchresult=1 under name “Supplemental file 1”. 13 Sample collection Bacterial swabs were collected from 12 adult and 24 juvenile spotted hyenas inhabiting the Masai Mara National Reserve, Kenya between May 2012 and July 2014. The adult hyenas were all females, and represented the Talek (N=5), Fig Tree (N=3), and Serena South (N=4) clans within the Reserve. The juveniles included 13 females and 11 males from the Talek clan. The hyenas were anesthetized with Telazol (6.5 mg/kg), and swabs were obtained from the anal scent gland, rectum, prepuce, oral cavity (gum line above the upper 3rd premolar), nares, and ear. Swabs were stored in cryogenic vials in liquid nitrogen until transport to Michigan State University, where they were stored at -80°C until DNA extraction. DNA extractions & 16S rRNA gene sequencing DNA was extracted from bacterial swabs using PowerSoil DNA Isolation Kits (MOBIO Laboratories, Inc; Carlsbad, CA), following the manufacturer’s recommended protocol, with two minor modifications. For each sample, we removed 200 µL Bead Solution and replaced it with 200 µL phenol:chloroform:isoamyl alcohol (25:24:1, v/v; Thermo Scientific, Waltham, MA). We also incubated the swab in Bead Solution within the Bead Tube for 10min, and vortexed the Bead Tube for 1 min before removing the swab aseptically and resuming the DNA extraction protocol. These modifications were implemented to increase the DNA yield of our low-biomass samples. The order of DNA extractions was randomized to minimize sampling bias (i.e. extracting samples from only one body-site or from the same individual hyena). We also completed 6 blank DNA extraction kit controls (i.e. DNA extractions of sterile swabs) to control for potential background DNA contamination. The V4 region of the bacterial 16S rRNA gene was targeted for paired-end sequencing (~253 bp per sequence) on the Illumina MiSeq platform at the Michigan State University Genomics Core (East Lansing, MI, USA). Sample 14 preparation, nucleotide sequencing, and preliminary quality filtering were completed as described previously [62, 63]. Sequence processing All sequence processing was conducted using mothur software (v.1.36.1), [64], following the MiSeq standard operating procedure (https://www.mothur.org/wiki/MiSeq_SOP). Briefly, forward and reverse reads were joined into contigs, generating a total of 11,331,400 paired-end sequences. After initial quality filtering, the remaining sequences (8,745,743) were aligned to the Silva reference database (v.4) [65]. Chimeric sequences were detected and removed using the UCHIME algorithm [66], and the remaining sequences were taxonomically classified using the Ribosomal Database Project reference files (RDP; v.9) [67] . Sequences deemed to have come from Chloroplasts, Mitochondria, Archaea, or Eukarya were filtered from our dataset, leaving 8,363,519 total sequences, which were clustered de novo into operational taxonomic units (OTUs) at 97% nucleotide similarity [68]. A total of 16 OTUs had an average relative abundance of >1% across our blank DNA extraction kit controls; most were previously documented contaminants of DNA extraction kits and/or reagents [69, 70]. We removed these OTUs from the dataset, with the exception of Providencia (OTU0006). Providencia was kept because it had an average relative abundance >1% among both biological and technical control samples, and members of this genus are common residents of the mammalian gastrointestinal tract [71– 73]. We subsampled individual samples to 13,340 sequence reads/sample prior to analysis to avoid biases due to sequencing effort. This subsampling cutoff was the third lowest number of sequences found in our dataset and was selected because it satisfied saturation for the majority of our samples and minimized data loss (i.e. a higher cutoff would have resulted in additional samples excluded from statistical analyses). The table of OTUs and 15 their associated taxonomic classifications can be downloaded online at https://academic.oup.com/femsec/article/96/2/fiaa007/5700710?searchresult=1 under names “Supplemental file 3” and “Supplemental file 4.” Microbial community composition analyses To visualize microbiota composition, we constructed heatmaps and stacked barplots in R v.3.4.3 [74]. For the raw relative abundances of each bacterial taxon at each body-site, see “Supplemental file 5” and “Supplemental file 6” from https://academic.oup.com/femsec/article/96/2/fiaa007/5700710?searchresult=1. We used Linear Discriminant Analysis (LDA) Effect Size (LEfSe) to identify the taxa that were enriched in particular samples (i.e. in one body-site relative to others) using default parameters [75]. LEfSe output for body-site vs. body-site, age and sex comparisons can be accessed using the link from above. Analyses of Alpha-diversity Microbiota a-diversity was estimated in mothur prior to deleting singletons (sequences observed only once in the dataset) and doubletons (sequences observed only twice). Specifically, community richness was characterized using the Chao1 nonparametric richness estimator, and community evenness was characterized using Shannon diversity and Simpson’s diversity indices. Chao 1 values were log-transformed prior to analyses, due to their skewed, high values (max 5000). The majority of rarefaction curves of OTU richness for a given number of sequences reached saturation, and Good’s coverage values of all body-sites averaged greater than 95%, indicating that sequencing depth was sufficient for analysis of these communities (Figure S1.1, Table S1.1). The effect of predictor variables (i.e. age-class, sex, social rank) on each measure of a-diversity was statistically evaluated using linear mixed effects models in R with the lme4 package, specifying hyena identity as a 16 random effect and body-site as one of the two fixed variables (e.g. y ~ bodysite + sex + 1|hyena id) [76]. The significance of the effect of each predictor variable on microbiota a- diversity was assessed via Wald Chi Square Tests on the linear mixed effects model using the R car package [77]. If a particular main effect was deemed statistically significant (p<0.05), we followed up with multiple comparison testing using the multcomp R package and report Benjamini-Hochberg adjusted p-values [78]. Analyses of Beta-diversity Prior to analysis of β-diversity, singletons and doubletons were removed from the dataset. For all the β-diversity analyses, we used the vegan package in R [74, 79]. β- diversity among samples was assessed using Jaccard (presence/absence data) and Bray- Curtis (relative abundance data) distance measures. To visualize microbiota similarity, we generated principal coordinate analysis (PCoA) plots from the distance matrices and coupled these with permutational multivariate analysis of variance (PERMANOVA) tests. To assess the effect of continuous predictor variables on microbiota similarity (i.e. social rank), we used Mantel tests with Spearman correlations. Lastly, to test for differences in microbiota dispersion (i.e. between-sample variance), we ran permutation tests of multivariate dispersions (PERMDISP2) [80]. To visualize the degree of dispersion in the microbiota, the output from PERMDISP2 was also plotted in R [74]. Data Availability 16S rRNA raw sequence reads were deposited in NCBI’s Sequence Read Archive, under BioProject PRJNA540066 and accession numbers SRR9006239-SRR9006445. Sample metadata and supplementary material for this article (see list and descriptions below) can be found in the Appendix for this chapter. 17 RESULTS Microbiota composition is body-site specific The microbiota of adult and juvenile spotted hyenas were niche-specific, and body- sites varied greatly in the relative abundances of their bacterial phyla, families and genera (Figure 1.1, Figure S1.2, Figure S1.3). In adult females, the ear microbiota were not composed of a few dominant taxa, but rather of many taxa found at low abundances. The nasal communities, however, contained a single predominant bacterial family, Moraxellaceae (51%). The oral cavity was mostly inhabited by Pasteurellaceae (23%), Leptotrichiaceae (12%), and Porphyromonadaceae (12%), and the rectum by Clostridiales_XI (Anaerococcus, 19%), Corynebacteriaceae (Corynebacterium; 9%), unclassified Clostridia (6%), and Bacteroidaceae (5%). The prepuce and anal scent gland microbiota were similar in composition; both were dominated by Clostridiales_XI (mostly Anaerococcus, 19% in prepuce; 29% in scent gland) and Corynebacteriaceae (Corynebacterium, 31% in prepuce; 11% in scent gland). Many of the abundant taxa at each body-site were identified by Linear Discriminant Analysis (LDA) Effect Size (LEfSe) as being differentially abundant among body-sites, particularly in the ears, nose, and mouth (Table S1.2). 18 Figure 1.1 Microbiota composition at multiple body sites in adult and juvenile spotted hyenas. Stacked bar plots showing the relative frequency of 16S rRNA gene sequences assigned to each bacterial family across samples in the ears, nose, mouth, prepuce, rectum and anal scent gland of adults (top) and juveniles (bottom). Each individual bar represents a sample and 1.00 equals 100%. Not all individual bars reach 1.00 because the rest of the taxa were not among the top 21 most abundant. Note how the keys from the two panels are almost identical, except for the names of the first two and last two taxa. As was observed in adults, microbiota composition also varied among body-sites in juvenile hyenas (Figure 1.1, Figure S1.2, Figure S1.3). Here again, the juvenile ear microbiota were not dominated by a single bacterial type, but the nasal microbiota mostly contained Moraxellaceae (68%), and the scent gland harbored high abundances of Anaerococcus, Corynebacterium, and Clostridia (20%, 19%, and 15%, respectively). The oral cavity was primarily colonized by Pasteurellaceae (18%), Porphyromonadaceae (11%), Leptotrichiaceae (11%), and Moraxellaceae (11%). The juvenile rectum was not dominated by any particular bacterial family, but harbored equal numbers of Bacteroidaceae, Lachnospiraceae, Fusobacteriaceae, and Streptococcaceae (all ~8%). The juvenile prepuce had high Corynebacteriaceae (Corynebacterium; 16%) and Enterobacteriaceae (Providencia, 10%) relative abundances. As in adults, many of the aforementioned taxa 19 were among those that LEfSe identified as being differentially abundant among body-sites (Table S1.3). Microbiota a-diversity and β-diversity vary among body-sites Body-sites also varied in their microbiota richness and evenness in both adult and juvenile hyenas (Table 1.1, Figure 1.2). Generally, the microbiota of the ear, mouth, and rectum were the most diverse, whereas the preputial, nasal, and anal scent gland microbiota were the least diverse (Figure 1.2); for post-hoc comparisons, see Table S1.4 and Table S1.5. Furthermore, β-diversity analyses confirmed that microbiota structure also varied among body-sites, both when taking into account the relative abundances of taxa (Bray-Curtis PERMANOVA, R2=0.42, p=0.001 for adults, R2=0.37, p=0.001 for juveniles) or only their presence and absence (Jaccard PERMANOVA, R2=0.15, p=0.001 for both adults and juveniles). PCoA ordinations using Bray-Curtis distances showed that the nasal and oral microbiota were different from one another and from those at other body-sites (Figure 1.3); therefore, we also plotted the remaining body-sites separately from the nose and mouth to better visualize their variation (Figure 1.4). This ordination shows that in adults, the microbiota from the ears, prepuce, anal scent gland and rectum formed unique, mostly non- overlapping clusters, but that in juveniles there was significant overlap between the ear and prepuce microbiota (Figure 1.4). Table 1.1 Body-sites vary in their microbiota richness and evenness (a-diversity). Predictor Body-site (12 adults; 6 sites; a-diversity metric Chao 1 Richness Shannon Diversity Index Simpson’s index (1-D) Chao 1 Richness Shannon Diversity Index Simpson’s index (1-D) DF c2 5 5 5 5 5 5 p-value 0.023* 12.99 1.69 x 10-13*** 68.95 5.4 x 10-13*** 66.49 2.32 x 10-7*** 39.04 258.9 < 2.2 x 10-16 *** 127.83 < 2.2 x 10-16 *** 71 samples) Body-site (24 juveniles; 6 sites; 143 samples) Shown are the Chi-Sq. values and p-values for linear mixed effects models specifying body site as a predictor variable, hyena identity as a random effect, and an alpha-diversity metric as a dependent variable. The data are shown separately for adults and juveniles. *p<0.05 **p<0.01 ***p<0.001. 20 Figure 1.2 Body sites vary in their microbiota α-diversity. Box plots of microbiota alpha-diversity (Chao 1 richness, Shannon diversity, Simpson’s index) at each body site in adults (A–C) and juveniles (D–F). Boxed Xs are outlier values. Figure 1.3 Microbiota cluster by body site in spotted hyenas. PCoA plots from Bray–Curtis dissimilarity matrices in (A) adults and (B) juveniles. Each point represents a sample and is color-coded by body site. Closeness of points indicates high community similarity. The % of variance accounted for by each principal coordinate axis is shown in the axis labels. 21 Figure 1.4 Microbiota structure and dispersion across body sites in juvenile and adult hyenas. PCoA plots from Bray–Curtis dissimilarity matrices in adults (A, B) and juveniles (C, D). Because the nasal and oral microbiota were drastically different from those of other body-sites (see Fig. 3), they are also plotted separately (B, D) in order to better visualize the variation in other body sites (A, C). Each point represents a sample and is color coded by body site. Closeness of points indicates high community similarity. The % of variance accounted for by each PCo axis is shown in the axis labels. Lastly, body-sites also varied in their degree of microbial community dispersion in both adults and juveniles (PERMDISP Bray-Curtis, F=6.54, p<0.0001 for adults, F=27.81, p<0.0001 for juveniles; PERMDISP Jaccard, F=32.20, p<0.0001 for adults, F=38.26, p<0.0001 for juveniles). However, the differences were modest. In adults, the ear microbiota was more homogenous among individuals than were microbiota at other body-sites. In juveniles, the ear and prepuce microbiota showed less individual variation than the nasal and oral microbiota (Table S1.6). 22 Variation in microbiota profiles is significantly associated with host sex and age- class Our results reveal that host sex and age-class were associated with variation in the microbiota of hyenas at multiple body-sites. Among juveniles, microbiota richness differed between females and males across body-sites (LMM Chao1 c2=4.79, p=0.02) but microbiota evenness did not (LMM Shannon c2=0.16, p=0.68; LMM Simpson c2=0.01, p=0.91). Specifically, juvenile males tended to have richer microbial communities than juvenile females. Additionally, host sex explained 10% of the structural variation in the anal scent gland microbiota (Table 1.2, Figure 1.5). Linear Discriminant Analysis (LDA) Effect Size (LEfSe) indicated that juvenile males harbored greater abundances of unclassified Clostridia, Prevotella, and unclassified Firmicutes in their anal scent glands, whereas juvenile females had more Corynebacterium and unclassified Clostridiales. Table 1.2 Juvenile female hyenas have distinct anal scent gland microbiota compared to juvenile male hyenas. Factor Body-site R2 p-value adjusted p-value Bray-Curtis Sex ears nasal oral prepuce rectum scent gland Jaccard Sex ears nasal oral prepuce rectum scent gland 0.03 0.05 0.16 0.06 0.038 0.17 0.069 0.03 0.040 0.51 0.102 0.007 0.05 0.032 0.06 0.141 0.042 0.48 0.059 0.067 0.043 0.040 0.011 0.08 0.060 0.204 0.204 0.060 0.510 0.042 0.08 0.16 0.48 0.10 0.08 0.06 Shown are the PERMANOVA tests assessing whether microbiota structure vary among the sexes (juvenile females vs. juvenile males) in spotted hyenas. PERMANOVA tests based on Bray-Curtis (proportions of taxa) distance matrices are shown on top and those based on Jaccard (presence/absence) distance matrices are shown at the bottom. p-values were adjusted for multiple comparisons using the Benjamini-Hochberg method and bolded if they were <0.05. 23 Figure 1.5 Juvenile females and juvenile males have distinct scent gland microbiota. PCoA plot from Bray–Curtis dissimilarity matrices showing the anal scent gland microbiota in juvenile females (orange) and juvenile males (green). Closeness of points indicates high community similarity. The % of variance accounted for by each PCo axis is shown in the axis labels. The microbiota also differed between juvenile females and adult females (Bray- Curtis PERMANOVA R2=0.01, p=0.001, Jaccard PERMANOVA R2=0.008, p=0.005). However, the percent variance in microbiota explained by age varied among body sites (Table 3). In the prepuce and rectum, host age-class accounted for 15% and 11% of the variation in microbiota structure, respectively (Table 3, Figure 6). LEfSe analyses indicated that the prepuce microbiota of adult female hyenas, compared to those of juvenile females, were enriched in Corynebacterium, Finegolidia, and Clostridiales. In the rectum, adult females harbored greater abundances of Anaerococcus and Corynebacterium, whereas juveniles contained greater abundances of Erysipelotrichaceae, Lachnospiraceae, and Helicobacteraceae. Furthermore, the preputial microbiota of adult females but not those of other body sites, tended to be more variable among individuals than did those of juvenile females (Bray-Curtis PERMDISP adults vs. juveniles: ears F=0.66, p=0.42; nasal F=2.13, p=0.15; oral F=0.87, p=0.35; prepuce F=0.21, p=0.02; rectum F=0.006, p=0.93, anal scent gland F=1.67, p=0.20) (Figure S1.4). Lastly, no differences in alpha-diversity were evident 24 between the microbiota of adult females and juvenile females across body-sites (LMM Chao1 c2=0.15, p=0.69, Shannon c2=0, p=0.99; LMM Simpson c2=1.16, p=0.28). Table 1.3 Adult female hyenas have distinct microbiota compared to juvenile female hyenas. Factor Body-site R2 p-value adjusted p-value Bray-Curtis Age class Jaccard Age class ears nasal oral prepuce rectum scent gland 0.04 0.043 0.44 0.057 0.15 0.033 0.72 0.151 0.001 0.113 0.004 0.37 ears nasal oral prepuce rectum scent gland 0.045 0.111 0.042 0.26 0.044 0.21 0.068 0.001 0.06 0.004 0.048 0.021 0.53 0.30 0.72 0.006 0.012 0.53 0.166 0.26 0.26 0.006 0.012 0.042 Shown are the PERMANOVA tests assessing whether microbiota structure vary among age classes (adult females vs. juvenile females) in spotted hyenas. PERMANOVA tests based on Bray-Curtis (proportions of taxa) distance matrices are shown on top and those based on Jaccard (presence/absence) distance matrices are shown at the bottom. p-values were adjusted for multiple comparisons using the Benjamini-Hochberg method and bolded if they were <0.05. Figure 1.6 Adult females and juvenile females have distinct prepuce and rectal microbiota. PCoA plot from Bray–Curtis dissimilarity matrices showing the prepuce (left) and rectal (right) microbiota in adult females (turquoise) and juvenile females (purple). Closeness of points indicates high community similarity. The % of variance accounted for by each PCo axis is shown in the axis labels. Host age-class was significantly associated with PC1 in both the prepuce (LM β = 0.43 ± 0.05, P < 0.0001) and rectum (LM β = 0.30 ± 0.10, P < 0.005). 25 Microbiota do not vary with host social rank, but are distinct among individuals Neither microbiota α-diversity (LMM Chao1 c2=0.059, p=0.80; Shannon diversity c2=0.064, p=0.80; Simpson’s index c2=0.16, p=0.68) nor β-diversity varied with host social rank in juvenile hyenas (both sexes included; Bray-Curtis Mantel test rho=0.003, p=0.45; Jaccard Mantel test rho=0.028, p=0.19). However, in both adult and juvenile hyenas, individual identity accounted for >11% of variation in microbiota structure across body-sites (adult females: PERMANOVA R2=0.11, p=0.001 for Bray-Curtis, R2=0.15, p=0.001 for Jaccard; juvenile females and males: R2=0.12, p=0.003 for Bray-Curtis ; R2=0.13, p=0.01 for Jaccard). DISCUSSION Principal findings of the study The purpose of this study was to characterize the diversity and structure of microbiota at six distinct body-sites in juvenile and adult spotted hyenas. We determined whether microbiota varied with host traits such as sex and social rank in juveniles of both sexes, and also whether microbiota differed between adult females and juvenile females. We found that the microbiota of spotted hyenas were body-site-specific, with respect to composition, structure, and diversity in both adult females and juveniles. Despite the body- site-specific structuring of the microbiota, these bacterial communities still exhibited strong specificity among individual hyenas, with host identity accounting for >11% of the total variation in microbiota structure. Additionally, the microbiota differed between adult females and juvenile females, particularly in the prepuce and rectum, indicating that age-related variation in diet, physiology, and/or social interactions might underlie the differences in their microbial communities. Furthermore, the anal scent gland microbiota of juvenile females were distinct from those of juvenile males, suggesting the potential role of hormones or sex- 26 specific life experiences in shaping these communities, even at this early life stage. Lastly, the microbiota of juvenile hyenas did not vary with host social rank. Future studies that include a larger number of adult hyenas of both sexes from a single clan will be required to determine whether social rank is associated with variation in the microbiota of mature hyenas. Ecological theory and niche structuring of the microbiota in spotted hyenas Our results support prior findings that anatomical body-site predominantly structures microbiota α-diversity and β-diversity in mammals [4, 31–34, 37, 81]. According to ecological theory, body-sites often act as environmental filters and impede colonization and persistence of bacterial groups that do not possess suitable functional traits for surviving and competing in their respective environments [82–85]. Body-sites are known to vary in their chemical and nutrient gradients, as well as in host immune activity [37, 86–88], and this likely contributed to the body-site specificity of the microbiota observed in this study. Why are the ear and oral microbiota highly diverse in spotted hyenas? In terms of microbiota α-diversity, in both adults and juveniles, the ears and mouth were the most taxa-rich, whereas the preputial and nasal communities were the least rich. In spotted hyenas, the oral and rectal microbial communities might be highly diverse due to hyenas’ varied diet; hyenas eat various tissue types (e.g. skin, meat, bone, viscera) and prey species [45, 89, 90], exposing them to many prey-associated microbiota and potentially giving rise to a diverse community of oral microbes that are able to utilize these varied substrates. A functional explanation for the high alpha-diversity of the hyena ear microbiota is lacking; however, one potential explanation is that hyena’s ear is often immersed deep in carcasses; this may facilitate colonization by a wide diversity of bacteria. Indeed, the most abundant microbes from the hyena ear have been found in the skin and abdominal cavity of decomposing animals and the skin and hindgut of vultures [91–93]. The 27 nasal microbiota of hyenas might harbor low-diversity due to the aerobic and mucous-rich environment of the nares, which may be inhospitable for many microbes [87, 94]. Lastly, the hyena prepuce in both males and females is part of a long and narrow organ that has a small aperture; it is composed of thick epidermal tissue and likely contains sebaceous glands as in other carnivores [45, 95, 96]. This unique physiology and morphology might be contributing to the site’s low bacterial diversity. Microbiota composition of spotted hyenas compared to those of other mammals With the exception of the ear, body-sites harbored microbes that also reside in the oral cavity, skin, scent gland, genitalia, and gut of other animals. A few dominant taxa of the hyena ear microbiota (Staphylococcaceae; Sphingobacteriaceae) are generalist bacterial taxa that are present across skin sites in both terrestrial and aquatic mammals [97–100]. However, unlike the skin of humans and apes [98, 100, 101], the hyena’s ear did not harbor appreciable numbers of Corynebacterium and Propionibacterium, and its microbiota profile only modestly resembled that of domestic dogs [102, 103]. Hyena noses and oral cavities exhibited fairly typical mammalian nasal and oral microbiota profiles [4, 33, 81, 87, 104– 107]. The only difference was that hyena oral microbial communities did not contain significant numbers of Prevotellaceae and Staphylococcaceae; as these taxa are positively associated with diets high in carbohydrates, fruits, and vegetables [108–111], which hyenas eat very rarely, if at all. In the prepuce, bacterial communities were dominated by taxa (i.e. Corynebacterium) that inhabit the urogenital and reproductive tract, skin, and scent gland of other mammals [8, 58, 98, 112–115]. Additionally, the hyena rectum was inhabited by bacteria that typically reside in the guts of meat- and insect-eating mammals [5, 116–120], as well as in dogs fed high-protein diets [111, 121]. Thus, although rectal microbiomes do not represent a proxy for gut microbiomes, it is not surprising to find that some taxa are 28 shared between the two sites. Members of these bacterial families (i.e. Lachnospiraceae, Streptococcaceae, Clostridiales_XI) can ferment protein, and in the process, synthesize short-chain fatty acids (SCFAs), branched-chain amino acids, ammonia, phenols, and indoles, which can then be used by the host and other bacteria [122–125]. Lastly, the anal scent gland microbiota profiles of hyenas strongly resembled the microbiota from other scent-producing areas in mammals (ie. axillae, musk gland, scent glands) [6, 8, 26, 58, 115, 126, 127]. Dominant bacterial taxa in these regions (i.e. Anaerococcus, Corynebacterium, and Porphyromonas) can also produce SCFAs [128–131]; however, in scent-producing glands, SCFAs, as well as medium-chain fatty acids, are hypothesized to function as volatile odorants that are employed by their mammalian hosts during chemical signaling [1, 132–134]. Microbiota of juvenile spotted hyenas vary with host sex but not social rank Within the body-sites of juvenile spotted hyenas, host sex was a significant predictor of microbiota composition and structure. Microbiota a-diversity differed between males and females across all of the surveyed body-sites, and host sex also predicted microbial community structure in the anal scent gland. Similarly, sex differences have also been observed in the gut, skin, and scent gland microbiota of primates, rodents, marsupials, carnivores, bats, and marine mammals [6, 7, 22, 98, 99, 118, 135–138]. In adult mammals, sex differences in the microbiota are often attributed to sex differences in physiology, morphology, hormones, and behavior [7, 35, 135, 139, 140]. In hyena societies, juvenile females associate with more individuals, and spend less time alone, than do juvenile males [141]. Male cubs are also known to scent mark more than female cubs, and male subadults scent overmark more than do subadult females [142]. These early behavioral differences 29 between the sexes might be modulating microbial exposure and consequently microbiota structure and composition in the anal scent gland of juvenile hyenas. In juvenile hyenas, host social rank did not consistently predict microbiota profiles at any body-site. Similarly, rank also failed to predict the gut microbiota in a Tanzanian population of spotted hyenas [59]. In contrast, the glandular microbiota of adult meerkats and sifakas have been found to vary with host social status [7, 8, 127]. In our study, the 24 juvenile hyenas (11-21 mos. of age) were still in the process of developing their ranks; young hyenas typically do not assume their proper positions in the clan’s hierarchy until they are at least 18 months of age [49, 50]. Future studies should investigate rank effects using a large sample of adult male and female hyenas from a single clan. Differences between juvenile and adult microbiota profiles Our results show that the microbiota of juvenile females and adult female hyenas differed in the prepuce and rectum, suggesting that life stage accounts for significant variation in the microbiota. This is also characteristic of primate, rodent, carnivore, and marine mammal microbiomes [7, 26, 99, 143–149]. Most notably, in the rectum, the microbiota of juvenile female hyenas were enriched in Erysipelotrichaceae and Helicobacter. High abundances of Erysipelotrichaceae have been associated with high fat diets [150–152]. Hyena milk has one of the highest fat contents of milks produced by land mammals [153], and some of the juveniles sampled here were still nursing, suggesting that perhaps this relatively high-fat diet might be related to the higher concentration of Erysipelotrichaceae in their rectums. Individual identity predicts microbiota profiles in hyenas and other mammals Despite the large amount of variation in microbiota profiles accounted for by body- site, sex, and age-class, individual hyenas still consistently harbored unique microbial communities. We found that individual identity was significantly associated with variation in 30 the microbiota across all sampled body-sites in both adults and juveniles, and accounted for >11% of the variation. In many mammals, host identity is one of the primary predictors of the skin or gut microbiota [29, 99, 103, 154–158]. Individual differences in immune function, early-life experiences, social interactions, and stress responses have been documented extensively for a range of mammalian taxa, and all of these variables may act individually or in concert to structure mammalian microbiomes [159–162]. 31 APPENDICES 32 APPENDIX A: SUPPLEMENTAL TABLES Table S1.1 Microbiota α-diversity among body-sites in adult and juvenile spotted hyenas (mean ± SD). Group Sample size (N) Good’s coverage Chao 1 Richness Shannon Diversity Simpson’s Index (1- D) 0.97 ± 0.02 0.78 ± 0.14 0.95 ± 0.02 0.88 ± 0.05 0.95 ± 0.01 0.85 ± 0.12 0.97 ± 0.01 0.68 ± 0.12 0.95 ± 0.02 0.89 ± 0.12 0.94 ± 0.04 0.81 ± 0.15 0.96 ± 0.016 0.95 ± 0.017 0.95 ± 0.007 0.96 ± 0.011 0.95 ± 0.010 0.95 ± 0.000 A. Microbiota α-diversity at multiple body-sites in adults 2287.68 ± 1011.19 5.29 ± 0.50 ears 3.31 ± 1.39 2205.56 ± 755.02 nasal 4.20 ± 0.39 2636.90 ± 555.26 oral 1669.71 ± 624.82 3.30 ± 0.41 prepuce 4.25 ± 0.38 2425.80 ± 655.85 rectum scent-gland 2427.06 ± 468.07 3.42 ± 0.71 B. Microbiota α-diversity at multiple body-sites in juveniles 5.58 ± 0.62 2635.02 ± 781.18 ears 2.31 ± 0.96 1845.60 ± 365.03 nasal 4.42 ± 0.48 2600.87 ± 605.82 oral 4.41 ± 1.12 prepuce 2192.69 ± 694.31 2595.38 ± 521.62 rectum 4.36 ± 0.47 2165.29 ± 817.339 3.05 ± 0.86 scent-gland 0.95 ± 0.018 0.96 ± 0.007 0.95 ± 0.008 0.96 ± 0.012 0.95 ± 0.009 0.96 ± 0.011 11 12 12 12 12 12 24 24 24 23 24 24 Microbiota alpha-diversity was estimated using the Chao1 nonparametric richness estimator, Shannon diversity, and Simpson’s index. We rarefied samples to 13,340 sequence reads/sample prior to analysis to control for sequencing effort. In doing so, 2 samples (1 juvenile prepuce swab, 1 adult ear swab) did not meet this cutoff and were excluded from analysis. 33 Table S1.2 List of taxa enriched in particular body-sites in adult spotted hyenas as determined by LEfSe. We used Linear discriminant analysis (LDA) Effect Size (LEfSe) to identity differentially abundant taxa at each body-site. These taxa characterize the microbiota at a particular body-site and consistently explain the differences between the microbiota at two or more body-sites. p-values were adjusted for multiple comparisons using the Benjamini-Hochberg method. 34 Table S1.3 List of taxa enriched in particular body-sites in juvenile spotted hyenas as determined by LEfSe. Group Mean OTU 3.92746 ears Otu000036.Firmicutes.Clostridia.Clostridiales.Peptostreptococcaceae.Clostridium_XI 2.47974 ears Otu000636.Firmicutes.Bacilli.Bacillales.Planococcaceae.Planococcaceae_unclassified 3.68639 ears Otu000127.Firmicutes.Bacilli.Lactobacillales.Enterococcaceae.Vagococcus 2.22506 ears Otu000613.Actinobacteria.Actinobacteria.Actinomycetales.Micrococcaceae.Micrococcaceae_unclassified Otu000567.Bacteroidetes.Sphingobacteria.Sphingobacteriales.Chitinophagaceae.Flavisolibacter 2.85683 ears Otu000864.Actinobacteria.Actinobacteria.Actinomycetales.Micromonosporaceae.Micromonosporaceae_unclassified 2.64521 ears 3.37773 ears Otu000231.Firmicutes.Bacilli.Bacillales.Bacillaceae_1.Bacillaceae_1_unclassified 2.60011 ears Otu000708.Proteobacteria.Alphaproteobacteria.Rhizobiales.Methylobacteriaceae.Methylobacterium Otu000726.Proteobacteria.Alphaproteobacteria.Rhizobiales.Methylobacteriaceae.Methylobacterium 2.56228 ears 3.39548 ears Otu000207.Proteobacteria.Alphaproteobacteria.Rhizobiales.Methylobacteriaceae.Methylobacterium 2.70693 ears Otu000894.Proteobacteria.Alphaproteobacteria.Rhizobiales.Rhizobiaceae.Rhizobium 3.2928 ears Otu000246.Firmicutes.Clostridia.Clostridiales.Clostridiaceae_1.Clostridium_sensu_stricto 2.91599 ears Otu000430.Actinobacteria.Actinobacteria.Actinomycetales.Microbacteriaceae.Microbacteriaceae_unclassified 3.00549 ears Otu000568.Proteobacteria.Alphaproteobacteria.Sphingomonadales.Sphingomonadaceae.Sphingomonas Otu000001.Proteobacteria.Gammaproteobacteria.Pseudomonadales.Moraxellaceae.Moraxellaceae_unclassified 5.69285 nasal 5.17674 nasal Otu000004.Proteobacteria.Gammaproteobacteria.Pseudomonadales.Moraxellaceae.Moraxellaceae_unclassified 4.49627 nasal Otu000029.unclassified.unclassified_unclassified.unclassified_unclassified.unclassified_unclassified.unclassified_unclassified 3.99009 nasal Otu000071.Actinobacteria.Actinobacteria.Actinomycetales.Microbacteriaceae.Microbacteriaceae_unclassified 4.91874 oral Otu000008.Proteobacteria.Gammaproteobacteria.Pasteurellales.Pasteurellaceae.Pasteurellaceae_unclassified 4.87842 oral Otu000009.Proteobacteria.Gammaproteobacteria.Pseudomonadales.Moraxellaceae.Moraxella Otu000010.Proteobacteria.Gammaproteobacteria.Pasteurellales.Pasteurellaceae.Pasteurella 4.68835 oral 4.66105 oral Otu000013.Fusobacteria.Fusobacteria.Fusobacteriales.Leptotrichiaceae.Streptobacillus 4.57885 oral Otu000018.Fusobacteria.Fusobacteria.Fusobacteriales.Leptotrichiaceae.Streptobacillus 4.4497 oral Otu000034.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidales_unclassified.Bacteroidales_unclassified 4.44257 oral Otu000024.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonadaceae_unclassified 4.36934 oral Otu000031.Fusobacteria.Fusobacteria.Fusobacteriales.Fusobacteriaceae.Fusobacterium Otu000014.Proteobacteria.Gammaproteobacteria.Pasteurellales.Pasteurellaceae.Pasteurellaceae_unclassified 4.30171 oral 4.21999 oral Otu000053.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonas 4.30185 oral Otu000048.Fusobacteria.Fusobacteria.Fusobacteriales.Leptotrichiaceae.Leptotrichiaceae_unclassified 4.21214 oral Otu000049.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidales_unclassified.Bacteroidales_unclassified 4.20179 oral Otu000043.Proteobacteria.Gammaproteobacteria.Pseudomonadales.Moraxellaceae.Moraxella 4.1362 oral Otu000077.Proteobacteria.Betaproteobacteria.Neisseriales.Neisseriaceae.Neisseriaceae_unclassified Otu000089.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidaceae.Bacteroides 3.99422 oral 4.00321 oral Otu000062.Proteobacteria.Betaproteobacteria.Neisseriales.Neisseriaceae.Neisseria 3.9759 oral Otu000044.Proteobacteria.Gammaproteobacteria.Pasteurellales.Pasteurellaceae.Pasteurellaceae_unclassified 3.91886 oral Otu000084.Firmicutes.Clostridia.Clostridiales.Peptostreptococcaceae.Peptostreptococcaceae_incertae_sedis 3.93846 oral Otu000072.Firmicutes.Clostridia.Clostridiales.Clostridiales_Incertae_Sedis_XI.Parvimonas 3.84203 oral Otu000128.Proteobacteria.Gammaproteobacteria.Pseudomonadales.Moraxellaceae.Moraxellaceae_unclassified Otu000119.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonas 3.84288 oral 3.83249 oral Otu000093.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonadaceae_unclassified 3.69983 oral Otu000153.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonas 3.72081 oral Otu000097.Firmicutes.Bacilli.Bacillales.Bacillales_Incertae_Sedis_XI.Gemella 3.66775 oral Otu000145.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidales_unclassified 3.65976 oral Otu000129.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Catonella Otu000154.Proteobacteria.Epsilonproteobacteria.Campylobacterales.Campylobacteraceae.Campylobacter 3.63928 oral 3.56764 oral Otu000101.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonadaceae_unclassified 3.58027 oral Otu000191.Firmicutes.Negativicutes.Selenomonadales.Veillonellaceae.Veillonella 3.00147 oral Otu000369.Actinobacteria.Actinobacteria.Actinomycetales.Actinomycetaceae.Actinomyces 2.85349 oral Otu000441.Firmicutes.Erysipelotrichia.Erysipelotrichales.Erysipelotrichaceae.Holdemania 2.77438 oral Otu000641.Spirochaetes.Spirochaetes.Spirochaetales.Spirochaetaceae.Treponema Otu000166.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonadaceae_unclassified 3.52701 oral 2.66649 oral Otu000553.Actinobacteria.Actinobacteria.Actinomycetales.Actinomycetaceae.Actinomyces 3.48251 oral Otu000142.Firmicutes.Bacilli.Lactobacillales.Lactobacillales_unclassified.Lactobacillales_unclassified 2.99941 oral Otu000471.Firmicutes.Clostridia.Clostridiales.Clostridiales_unclassified.Clostridiales_unclassified 3.46244 oral Otu000218.Firmicutes.Clostridia.Clostridiales.Peptostreptococcaceae.Peptostreptococcaceae_unclassified 3.41059 oral Otu000168.Proteobacteria.Gammaproteobacteria.Pseudomonadales.Moraxellaceae.Moraxellaceae_unclassified Otu000459.Actinobacteria.Actinobacteria.Actinomycetales.Corynebacteriaceae.Corynebacterium 2.86417 oral 3.23286 oral Otu000244.Firmicutes.Clostridia.Clostridiales.Clostridiales_Incertae_Sedis_XI.Clostridiales_Incertae_Sedis_XI_unclassified 2.64251 oral Otu000666.Firmicutes.Clostridia.Clostridiales.Clostridiales_unclassified 3.18243 oral Otu000136.Actinobacteria.Actinobacteria.Actinomycetales.Actinomycetaceae.Actinomyces 3.27749 oral Otu000268.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonas 2.94293 oral Otu000463.Firmicutes.Clostridia.Clostridiales.Clostridiales_unclassified.Clostridiales_unclassified Otu000343.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonas 3.03655 oral 2.72411 oral Otu000564.Proteobacteria.Betaproteobacteria.Burkholderiales.Burkholderiales_unclassified 5.19178 prepuce Otu000005.Actinobacteria.Actinobacteria.Actinomycetales.Corynebacteriaceae.Corynebacterium 4.51374 prepuce Otu000022.Firmicutes.Clostridia.Clostridiales.Clostridiales_Incertae_Sedis_XI.Anaerococcus 3.40132 prepuce Otu000142.Firmicutes.Bacilli.Lactobacillales.Lactobacillales_unclassified.Lactobacillales_unclassified 3.14013 prepuce Otu000334.Proteobacteria.Gammaproteobacteria.Pseudomonadales.Pseudomonadaceae.Pseudomonas Otu000130.Proteobacteria.Betaproteobacteria.Neisseriales.Neisseriaceae.Neisseria 3.4963 prepuce 2.62901 prepuce Otu000091.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Lachnospiraceae_unclassified 4.59809 rectum Otu000023.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidaceae.Bacteroides 4.5641 rectum Otu000028.Firmicutes.Bacilli.Lactobacillales.Streptococcaceae.Streptococcus 4.5207 rectum Otu000030.Actinobacteria.Actinobacteria.Actinomycetales.Corynebacteriaceae.Corynebacterium 4.42852 rectum Otu000040.Bacteroidetes.Bacteroidia.Bacteroidales.Prevotellaceae.Prevotellaceae_unclassified Otu000021.Fusobacteria.Fusobacteria.Fusobacteriales.Fusobacteriaceae.Fusobacterium 4.44414 rectum 4.44426 rectum Otu000011.Firmicutes.Bacilli.Bacillales.Staphylococcaceae.Staphylococcus 3.06447 rectum Otu000302.Actinobacteria.Actinobacteria.Coriobacteriales.Coriobacteriaceae.Collinsella 4.33233 rectum Otu000046.Fusobacteria.Fusobacteria.Fusobacteriales.Fusobacteriaceae.Fusobacterium 35 4.27817 rectum Otu000020.Actinobacteria.Actinobacteria.Actinomycetales.Actinomycetaceae.Actinomycetaceae_unclassified 3.11886 rectum Otu000323.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Blautia 4.23963 rectum Otu000057.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidales_unclassified 4.22894 rectum Otu000060.Bacteroidetes.Bacteroidetes_unclassified Otu000064.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidales_unclassified 4.14204 rectum p.adj LDA 3.66025 0.001906 3.47741 0.009867 3.42212 2.36E-05 3.39999 0.002104 3.28804 0.000449 3.22307 0.000109 3.22057 7.02E-08 3.21215 0.003688 3.18464 0.00557 3.17565 1.99E-06 3.13906 0.010156 3.0949 1.14E-07 3.00143 4.77E-07 2.94389 0.001169 5.38798 9.77E-11 4.89327 2.32E-10 4.2175 6.86E-11 3.70748 1.08E-10 4.59872 3.66E-11 4.57078 2.18E-10 4.37142 5.68E-10 4.36321 2.81E-10 4.25996 9.87E-12 4.17697 2.83E-11 4.16787 1.97E-10 4.06263 3.95E-11 3.98928 9.87E-12 3.96479 9.77E-11 3.96045 4.27E-08 3.92606 5.55E-09 3.91493 8.98E-07 3.85064 6.09E-10 3.73796 3.95E-11 3.69649 8.06E-10 3.672 3.01E-09 3.65259 1.22E-11 3.63568 9.75E-11 3.60319 1.79E-08 3.56005 2.31E-05 3.55466 9.87E-12 3.47905 6.21E-10 3.473 1.50E-09 3.42424 3.31E-10 3.40254 1.22E-11 3.37572 9.87E-12 3.3723 1.40E-10 3.34289 3.13E-10 3.33336 3.90E-08 3.30416 1.97E-10 3.28655 5.61E-05 3.27504 1.93E-08 3.26282 1.11E-05 3.25075 1.85E-10 3.24792 3.76E-10 3.23786 1.94E-10 3.18484 9.11E-10 3.16468 1.16E-08 3.08038 6.53E-10 3.07664 1.25E-05 3.04968 8.21E-10 3.04009 6.21E-10 2.92214 1.27E-06 2.89058 2.14E-07 2.81156 7.67E-06 4.92105 6.48E-09 4.25089 1.31E-05 3.33126 0.000623 3.30856 0.007987 3.2606 0.000108 2.74245 0.001181 4.31754 4.26E-10 4.25088 8.65E-08 4.23039 3.74E-05 4.17098 7.50E-07 4.13665 1.02E-10 4.05985 5.10E-07 4.04068 0.000311 4.00306 1.39E-09 3.97884 6.02E-06 3.95632 0.001211 3.9175 4.78E-08 3.90244 4.42E-05 3.87348 1.23E-06 Table S1.3 (cont’d) Otu000641.Spirochaetes.Spirochaetes.Spirochaetales.Spirochaetaceae.Treponema 3.28655 5.61E-05 2.77438 oral Otu000166.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonadaceae_unclassified 3.27504 1.93E-08 3.52701 oral Otu000553.Actinobacteria.Actinobacteria.Actinomycetales.Actinomycetaceae.Actinomyces 3.26282 1.11E-05 2.66649 oral Otu000142.Firmicutes.Bacilli.Lactobacillales.Lactobacillales_unclassified.Lactobacillales_unclassified 3.25075 1.85E-10 3.48251 oral Otu000471.Firmicutes.Clostridia.Clostridiales.Clostridiales_unclassified.Clostridiales_unclassified 3.24792 3.76E-10 2.99941 oral Otu000218.Firmicutes.Clostridia.Clostridiales.Peptostreptococcaceae.Peptostreptococcaceae_unclassified 3.23786 1.94E-10 3.46244 oral Otu000168.Proteobacteria.Gammaproteobacteria.Pseudomonadales.Moraxellaceae.Moraxellaceae_unclassified 3.18484 9.11E-10 3.41059 oral Otu000459.Actinobacteria.Actinobacteria.Actinomycetales.Corynebacteriaceae.Corynebacterium 3.16468 1.16E-08 2.86417 oral Otu000244.Firmicutes.Clostridia.Clostridiales.Clostridiales_Incertae_Sedis_XI.Clostridiales_Incertae_Sedis_XI_unclassified 3.08038 6.53E-10 3.23286 oral Otu000666.Firmicutes.Clostridia.Clostridiales.Clostridiales_unclassified 3.07664 1.25E-05 2.64251 oral Otu000136.Actinobacteria.Actinobacteria.Actinomycetales.Actinomycetaceae.Actinomyces 3.04968 8.21E-10 3.18243 oral Otu000268.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonas 3.04009 6.21E-10 3.27749 oral Otu000463.Firmicutes.Clostridia.Clostridiales.Clostridiales_unclassified.Clostridiales_unclassified 2.92214 1.27E-06 2.94293 oral Otu000343.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonas 2.89058 2.14E-07 3.03655 oral Otu000564.Proteobacteria.Betaproteobacteria.Burkholderiales.Burkholderiales_unclassified 2.81156 7.67E-06 2.72411 oral Otu000005.Actinobacteria.Actinobacteria.Actinomycetales.Corynebacteriaceae.Corynebacterium 4.92105 6.48E-09 5.19178 prepuce Otu000022.Firmicutes.Clostridia.Clostridiales.Clostridiales_Incertae_Sedis_XI.Anaerococcus 4.25089 1.31E-05 4.51374 prepuce Otu000142.Firmicutes.Bacilli.Lactobacillales.Lactobacillales_unclassified.Lactobacillales_unclassified 3.33126 0.000623 3.40132 prepuce Otu000334.Proteobacteria.Gammaproteobacteria.Pseudomonadales.Pseudomonadaceae.Pseudomonas 3.30856 0.007987 3.14013 prepuce Otu000130.Proteobacteria.Betaproteobacteria.Neisseriales.Neisseriaceae.Neisseria 3.2606 0.000108 3.4963 prepuce Otu000091.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Lachnospiraceae_unclassified 2.74245 0.001181 2.62901 prepuce Otu000023.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidaceae.Bacteroides 4.31754 4.26E-10 4.59809 rectum Otu000028.Firmicutes.Bacilli.Lactobacillales.Streptococcaceae.Streptococcus 4.25088 8.65E-08 4.5641 rectum Otu000030.Actinobacteria.Actinobacteria.Actinomycetales.Corynebacteriaceae.Corynebacterium 4.23039 3.74E-05 4.5207 rectum Otu000040.Bacteroidetes.Bacteroidia.Bacteroidales.Prevotellaceae.Prevotellaceae_unclassified 4.17098 7.50E-07 4.42852 rectum Otu000021.Fusobacteria.Fusobacteria.Fusobacteriales.Fusobacteriaceae.Fusobacterium 4.13665 1.02E-10 4.44414 rectum Otu000011.Firmicutes.Bacilli.Bacillales.Staphylococcaceae.Staphylococcus 4.05985 5.10E-07 4.44426 rectum Otu000302.Actinobacteria.Actinobacteria.Coriobacteriales.Coriobacteriaceae.Collinsella 4.04068 0.000311 3.06447 rectum Otu000046.Fusobacteria.Fusobacteria.Fusobacteriales.Fusobacteriaceae.Fusobacterium 4.00306 1.39E-09 4.33233 rectum Otu000020.Actinobacteria.Actinobacteria.Actinomycetales.Actinomycetaceae.Actinomycetaceae_unclassified 3.97884 6.02E-06 4.27817 rectum Otu000323.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Blautia 3.95632 0.001211 3.11886 rectum Otu000057.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidales_unclassified 3.9175 4.78E-08 4.23963 rectum Otu000060.Bacteroidetes.Bacteroidetes_unclassified 3.90244 4.42E-05 4.22894 rectum Otu000064.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidales_unclassified 3.87348 1.23E-06 4.14204 rectum Otu000520.Firmicutes.Erysipelotrichia.Erysipelotrichales.Erysipelotrichaceae.Coprobacillus 3.87205 3.49E-07 2.90422 rectum Otu000074.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Lachnospiraceae_unclassified 3.8585 3.95E-11 4.12686 rectum Otu000068.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidales_unclassified 3.84846 1.09E-05 4.12034 rectum Otu000039.Firmicutes.Clostridia.Clostridiales.Clostridiaceae_1.Clostridium_sensu_stricto 3.82977 6.21E-10 4.08599 rectum Otu000078.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidaceae.Bacteroides 3.74505 2.54E-07 4.07313 rectum Otu000134.Fusobacteria.Fusobacteria.Fusobacteriales.Fusobacteriaceae.Fusobacteriaceae_unclassified 3.70083 3.01E-09 3.75451 rectum Otu000092.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Lachnospiracea_incertae_sedis 3.69828 6.48E-09 4.01207 rectum Otu000086.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Blautia 3.69023 8.06E-10 3.98647 rectum Otu000091.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Lachnospiraceae_unclassified 3.68101 1.07E-08 3.97937 rectum Otu000095.Firmicutes.Erysipelotrichia.Erysipelotrichales.Erysipelotrichaceae.Erysipelotrichaceae_incertae_sedis 3.66816 8.54E-08 3.97607 rectum Otu000103.Bacteroidetes.Bacteroidia.Bacteroidales.Bacteroidaceae.Bacteroides 3.60405 0.000796 3.85341 rectum Otu000113.Proteobacteria.Betaproteobacteria.Burkholderiales.Sutterellaceae.Sutterella 3.53919 1.54E-08 3.85065 rectum Otu000208.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Lachnospiraceae_unclassified 3.43802 2.07E-09 3.51722 rectum Otu000233.Firmicutes.Erysipelotrichia.Erysipelotrichales.Erysipelotrichaceae.Erysipelotrichaceae_unclassified 3.42049 2.11E-07 3.42506 rectum Otu000255.Firmicutes.Erysipelotrichia.Erysipelotrichales.Erysipelotrichaceae.Erysipelotrichaceae_unclassified 3.32734 4.31E-09 3.40712 rectum Otu000104.Firmicutes.Clostridia.Clostridiales.Clostridiales_unclassified 3.28753 8.64E-06 3.54334 rectum Otu000100.Firmicutes.Clostridia.Clostridiales.Peptostreptococcaceae.Peptostreptococcus 3.26407 1.88E-05 3.46185 rectum Otu000222.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Lachnospiraceae_unclassified 3.35963 rectum 3.22408 0.000312 Otu000003.Actinobacteria.Actinobacteria.Actinomycetales.Corynebacteriaceae.Corynebacterium 5.25754 scentgland 4.95885 5.19E-11 Otu000002.Firmicutes.Clostridia.Clostridiales.Clostridiales_Incertae_Sedis_XI.Anaerococcus 5.2115 scentgland 4.88357 1.34E-07 Otu000007.Firmicutes.Clostridia.Clostridia_unclassified.Clostridia_unclassified.Clostridia_unclassified 5.03071 scentgland 4.73859 1.21E-10 Otu000015.Bacteroidetes.Bacteroidia.Bacteroidales.Porphyromonadaceae.Porphyromonas 4.68288 scentgland 4.41495 7.82E-08 Otu000025.Firmicutes.Clostridia.Clostridia_unclassified.Clostridia_unclassified.Clostridia_unclassified 4.52923 scentgland 4.23317 4.97E-06 Otu000032.Firmicutes.Negativicutes.Selenomonadales.Veillonellaceae.Veillonellaceae_unclassified 4.46 scentgland 4.1508 1.90E-09 Otu000033.Firmicutes.Clostridia.Clostridiales.Clostridiales_unclassified.Clostridiales_unclassified 4.35209 scentgland 4.04488 1.45E-05 Otu000050.Actinobacteria.Actinobacteria.Actinomycetales.Propionibacteriaceae.Propionibacterium 4.26644 scentgland 3.95708 3.95E-11 Otu000058.Firmicutes.Clostridia.Clostridiales.Clostridiales_unclassified 4.07725 scentgland 3.79965 1.94E-05 Otu000045.Firmicutes.Firmicutes_unclassified.Firmicutes_unclassified.Firmicutes_unclassified.Firmicutes_unclassified3.87688 scentgland 3.59435 4.37E-05 3.46994 scentgland 3.29902 0.006999 Otu000118.Firmicutes.Firmicutes_unclassified Otu000222.Firmicutes.Clostridia.Clostridiales.Lachnospiraceae.Lachnospiraceae_unclassified 2.59068 scentgland 3.00808 0.000133 We used Linear discriminant analysis (LDA) Effect Size (LEfSe) to identity differentially abundant taxa at each body-site. These taxa characterize the microbiota at a particular body-site and consistently explain the differences between the microbiota at two or more body-sites. p-values were adjusted for multiple comparisons using the Benjamini-Hochberg method. 36 Table S1.4 Multiple-comparison testing of body-site microbiota alpha-diversity values in adults. nasal oral prepuce rectum Scent gland A)Chao 1 Richness ears nasal oral prepuce rectum B)Shannon Diversity Index ears nasal oral prepuce rectum 111.74 288 0.74 -325.30 288 0.51 -437 288 0.32 592.45 288 0.14 480.71 288 0.28 917.75 288 0.02* -202.92 288 0.72 -314.67 288 0.51 122.38 288 0.74 -795.37 288 0.04* -168.65 288 0.73 -280.39 288 0.55 156.65 288 0.73 -761.10 288 0.04* 34.27 288 0.09 nasal oral prepuce rectum 2.19 0.30 <0.0001*** 1.08 0.30 0.001** -1.10 0.30 <0.0001*** 1.94 0.30 <0.0001*** -0.24 0.30 0.49 0.86 0.30 0.008** 0.94 0.30 0.004** -1.24 0.30 <0.0001*** -0.14 0.30 0.64 -1.00 0.30 0.002** Scent gland 1.74 0.30 <0.0001** -0.44 0.30 0.18 0.66 0.30 0.04* -0.19 0.30 0.55 0.80 0.30 0.01* C)Simpson’s index (1-D) ears nasal nasal 0.20 0.03 <0.0001** * oral 0.02 0.03 0.55 -0.18 0.03 <0.0001** * prepuce 0.09 0.03 0.006** -0.10 0.03 0.004** rectum Scent gland 0.11 0.03 0.001** 0.01 0.03 0.61 -0.18 0.03 <0.0001 *** -0.08 0.03 0.01* 37 Table S1.4 (cont’d) oral prepuce rectum 0.07 0.03 0.04* -0.006 0.03 0.83 -0.07 0.03 0.02* 0.09 0.03 0.01* 0.01 0.03 0.61 0.09 0.03 0.006** Microbiota alpha-diversity (richness and evenness) varied among body-sites in adult hyenas when evaluated using a linear mixed effects model. Shown are the estimate, std. error, and Benjamini- Hochberg adjusted p-values of multiple comparison testing to see exactly which body-sites varied from one another. A) Chao Richness, B) Shannon diversity index, and C) Simpson’s index. Comparisons are read from left to right. Table S1.5 Multiple-comparison testing of body-site microbiota alpha-diversity values in juveniles. A)Chao 1 Richness ears nasal oral prepuce rectum B)Shannon Diversity Index ears nasal oral nasal oral prepuce rectum Scent gland 869.89 174.06 <0.0001*** 117.66 174.06 0.62 -752.24 174.06 <0.0001*** 512.23 174.06 0.009** -357.67 174.06 0.05 394.57 174.06 0.03* 67.73 174.06 0.80 -802.17 174.06 <0.0001*** -49.93 174.06 0.82 -444.50 174.06 0.02* nasal oral prepuce rectum 546.72 174.06 0.006** -323.17 174.06 0.08 429.07 174.06 0.02* 34.50 174.06 0.84 -479.00 174.06 0.01* Scent gland 3.25 0.23 <0.0001*** 1.18 0.23 <0.0001*** -2.06 0.23 <0.0001*** 1.17 0.23 <0.0001*** -2.08 0.23 <0.0001*** -0.01 0.23 0.96 1.22 0.23 <0.0001*** -2.03 0.23 <0.0001*** 0.03 0.23 0.94 2.62 0.23 <0.0001** -0.62 0.23 0.007** 1.44 0.23 <0.0001*** 38 Table S1.5 (cont’d) prepuce rectum 0.04 0.23 0.94 1.45 0.23 <0.0001*** 1.40 0.23 <0.0001*** C)Simpson’s index (1-D) ears nasal oral prepuce rectum nasal oral prepuce rectum 0.28 0.02 <0.0001*** 0.01 0.02 0.58 -0.26 0.02 <0.0001*** 0.08 0.02 0.007** -0.20 0.02 <0.0001*** 0.06 0.02 0.03* 0.02 0.02 0.41 -0.25 0.02 <0.0001*** 0.009 0.02 0.75 -0.05 0.02 0.07 Scent gland 0.16 0.02 <0.0001*** -0.11 0.02 <0.0001*** 0.15 0.02 <0.0001*** 0.08 0.02 0.005** 0.14 0.02 <0.0001*** Microbiota alpha-diversity (richness and evenness) varied among body-sites in adult hyenas when evaluated using a linear mixed effects model. Shown are the estimate, std. error, and Benjamini- Hochberg adjusted p-values of multiple comparison testing to see exactly which body-sites varied from one another. A) Chao Richness, B) Shannon diversity index, and C) Simpson’s index. Comparisons are read from left to right. Table S1.6 Body-sites vary in their community dispersion in adults and juveniles. ADULTS Bray- Curtis nasal oral prepuce rectum scent gland ears nasal oral prepuce rectum -0.17 0.003 -0.24 <0.0001 -0.16 0.008 -0.11 0.13 -0.19 0.001 0.06 0.73 0.01 0.99 0.06 0.71 -0.01 0.99 0.07 0.53 0.12 0.06 0.04 0.89 0.04 0.88 -0.02 0.98 -0.07 0.51 39 Table S1.6 (cont’d) oral Jaccard ears nasal -0.01 0.87 -0.15 <0.0001 -0.03 0.16 -0.05 0.003 -0.04 0.02 scent gland prepuce rectum Bray- Curtis ears oral prepuce rectum scent gland ears -0.35 <0.0001 -0.22 <0.0001 -0.01 0.99 -0.09 0.07 -0.12 0.008 nasal oral -0.13 <0.0001 -0.01 0.76 -0.03 0.06 -0.02 0.29 -0.13 <0.0001 -0.01 0.76 -0.03 0.06 JUVENILES oral nasal prepuce rectum -0.01 0.67 -0.01 0.97 0.009 0.97 prepuce rectum 0.13 0.005 0.34 <0.0001 0.25 <0.0001 0.22 <0.0001 0.21 <0.0001 0.12 0.008 0.09 0.08 -0.08 0.16 -0.11 0.02 -0.02 0.97 Jaccard nasal oral prepuce rectum scent gland ears -0.02 0.36 -0.12 <0.0001 0.012 0.90 -0.06 <0.0001 -0.02 0.41 nasal -0.10 <0.0001 0.03 0.038 -0.04 0.007 0.0008 0.99 oral 0.13 <0.0001 0.06 <0.0001 0.10 <0.0001 prepuce rectum -0.07 <0.0001 -0.03 0.04 0.04 0.006 Shown are the PERMDISP mean differences and adjusted p-values for body-site vs body-site comparisons of microbiota dispersion using Bray-Curtis (left) and Jaccard (right) distances in adults (top) and juveniles (bottom). PERMDISP measures the degree of community homogeneity. Comparisons are read from left to right. 40 APPENDIX B: SUPPLEMENTAL FIGURES Figure S1.1 Rarefaction curves of microbial community richness across body-sites in spotted hyenas. Plotted are the number of OTUs (OTU Richness) that are recovered with an increasing number of sequences, after subsampling to 13,340 sequences/sample. The maximum number of OTUs found in a sample was 2846. Each curve represents a unique sample and is color-coded by body-site origin. The diagonal line is a y=x linear function line, which highlights what the curve would like if with every sequence came a new OTU. 41 Figure S1.2 Body-sites vary in their microbiota composition. Stacked bar plots showing the relative frequency of 16S rRNA gene sequences assigned to each bacterial phylum across samples in A) adult and B) juvenile spotted hyenas. Samples are grouped by body-site, each color represents a bacterial phylum, and 1.00 equals 100%. 42 Figure S1.3 Top 21 most abundant bacterial genera inhabiting multiple body-sites in adult and juvenile spotted hyenas. Stacked bar plots showing the relative frequency of 16S rRNA gene sequences assigned to each bacterial genera across samples in the ears, nose, mouth, prepuce, rectum, and anal scent gland of adults (top) and juveniles (bottom). Each individual bar represents a sample and 1.00 equals 100%. Not all individual bars reach 1.00 because the rest of the taxa were not among the top 21 most abundant. Note how the keys from the two panels are not identical, and thus, the two panels are not comparable. Figure S1.4. Bacterial community dispersion in adult female hyenas compared to juvenile female hyenas. Plots showing the degree of microbiota dispersion (i.e. heterogeneity) in the ears, nose, mouth, prepuce, rectum, and anal scent gland of adult females (black points) and juvenile females (red points). Larger shapes indicate high between-sample variation. Overlapping shapes indicate similarity in microbial communities between the two groups. 43 REFERENCES 44 REFERENCES 1. Archie EA, Theis KR. Animal behaviour meets microbial ecology. Anim Behav. 2011;82:425– 36. 2. Archie EA, Tung J. Social behavior and the microbiome. Curr Opin Behav Sci. 2015;6:28–34. 3. Vuong HE, Yano JM, Fung TC, Hsiao EY. The Microbiome and Host Behavior. Annu Rev Neurosci. 2017;40:21–49. doi:10.1146/annurev-neuro-072116. 4. Alfano N, Courtiol A, Vielgrader H, Timms P, Roca AL, Greenwood AD. Variation in koala microbiomes within and between individuals: effect of body region and captivity status. Sci Rep. 2015;5:10189. doi:10.1038/srep10189. 5. Delsuc F, Metcalf JL, Wegener Parfrey L, Song SJ, González A, Knight R. Convergence of gut microbiomes in myrmecophagous mammals. Mol Ecol. 2014;23:1301–17. 6. Theis KR, Venkataraman A, Dycus JA, Koonter KD, Schmitt-Matzen EN, Wagner AP, et al. Symbiotic bacteria appear to mediate hyena social odors. PNAS. 2013;110:19832–19837. doi:10.1073/pnas.1306477110. 7. Leclaire S, Nielsen JF, Drea CM. Bacterial communities in meerkat anal scent secretions vary with host sex, age, and group membership. Behav Ecol. 2014;25:996–1004. 8. Leclaire S, Jacob S, Greene LK, Dubay GR, Drea CM. Social odours covary with bacterial community in the anal secretions of wild meerkats. Sci Rep. 2017;7:3240. doi:10.1038/s41598- 017-03356-x. 9. Buesching CD, Tinnesand HV, Sin Y, Rosell F, Burke T, Macdonald DW. Coding of Group Odor in the Subcaudal Gland Secretion of the European Badger Meles meles: Chemical Composition and Pouch Microbiota. In: Schulte BA, Goodwin TE, Ferkin MH, editors. Chemical Signals in Vertebrates 13. New York, NY; 2016. p. 45–62. 10. Hamm PS, Caimi NA, Northup DE, Valdez EW, Buecher DC, Dunlap CA, et al. Western Bats as a Reservoir of Novel Streptomyces Species with Antifungal Activity. Appl Environ Microbiol. 2017;83:e03057. doi:10.1128/AEM.03057-16. 11. Hoyt JR, Cheng TL, Langwig KE, Hee MM, Frick WF, Kilpatrick AM. Bacteria Isolated from Bats Inhibit the Growth of Pseudogymnoascus destructans, the Causative Agent of White-Nose Syndrome. PLoS One. 2015;10:1–12. doi:10.1371/journal.pone.0121329. 12. Lemieux-Labonté V, Simard A, Willis CKR, Lapointe F-J. Enrichment of beneficial bacteria in the skin microbiota of bats persisting with white-nose syndrome. Microbiome. 2017;5:1–15. doi:10.1186/s40168-017-0334-y. 13. Antwis RE, Griffiths SM, Harrison XA, Aranega-Bou P, Arce A, Bettridge AS, et al. Fifty important research questions in microbial ecology. FEMS Microbiol Ecol. 2017;93:1–10. 14. McFall-Ngai M, Hadfield MG, Bosch TCG, Carey H V., Domazet-Lošo T, Douglas AE, et al. Animals in a bacterial world, a new imperative for the life sciences. PNAS. 2013;110:3229–36. 45 15. Bordenstein SR, Theis KR. Host biology in light of the microbiome: Ten principles of holobionts and hologenomes. PLoS Biol. 2015;13:1–23. 16. Youngblut ND, Reischer GH, Walters W, Schuster N, Walzer C, Stalder G, et al. Host diet and evolutionary history explain different aspects of gut microbiome diversity among vertebrate clades. Nat Commun. 2019;10:1–15. doi:10.1038/s41467-019-10191-3. 17. Groussin M, Mazel F, Sanders JG, Smillie CS, Lavergne S, Thuiller W, et al. Unraveling the processes shaping mammalian gut microbiomes over evolutionary time. Nat Commun. 2017;8:14319. doi:10.1038/ncomms14319. 18. Muegge BD, Kuczynski J, Kinghts D, Clemente JC, Fontana L, Henrissat B, et al. Diet drives Convergence in Gut Microbiome Functions Across Mammalian Phylogeny and Within Humans. Science (80- ). 2017;332:970–4. 19. Ley RE, Hamady M, Lozupone C, Turnbaugh PJ, Ramey RR, Bircher JS, et al. Evolution of Mammals and Their Gut Microbes. Science (80- ). 2008;320:1647. 20. Li J, Zhan S, Liu X, Lin Q, Jiang J, Li X. Divergence of Fecal Microbiota and Their Associations With Host Phylogeny in Cervinae. Front Microbiol. 2018;9:1–11. doi:10.3389/fmicb.2018.01823. 21. Greene LK, Clayton JB, Rothman RS, Semel BP, Semel MA, Gillespie TR, et al. Local habitat , not phylogenetic relatedness , predicts gut microbiota better within folivorous than frugivorous lemur lineages. Biol Lett. 2019;15:5–11. 22. Watson SE, Hauffe HC, Bull M, McKinney MA, Atwood TA, Perkins SE. Global change- driven use of onshore habitat impacts polar bear faecal microbiota. ISME J. 2019. doi:10.1038/s41396-019-0480-2. 23. Sommer F, Stå M, Ilkayeva O, Newgard CB, Frö O, BäCkhed Correspondence F. The Gut Microbiota Modulates Energy Metabolism in the Hibernating Brown Bear Ursus arctos. Cell Rep. 2016;14:1655–61. 24. Perofsky AC, Lewis RJ, Abondano LA, Di Fiore A, Meyers LA. Hierarchical social networks shape gut microbial composition in wild Verreaux’s sifaka. Proceedings Biol Sci. 2017;284. 25. Tung J, Barreiro LB, Burns MB, Grenier JC, Lynch J, Grieneisen LE, et al. Social networks predict gut microbiome composition in wild baboons. Elife. 2015;2015:1–18. 26. Theis KR, Venkataraman A, Wagner AP, Holekamp KE, Schmidt TM. Age-related variation in the scent pouch bacterial communities of striped hyenas. In: Schulte BA, Goodwin TE, Ferkin MH, editors. Chemical Signals in Vertebrates 13. New York, NY: Springer; 2016. doi:10.1007/978-3-319-22026-0. 27. Blekhman R, Goodrich JK, Huang K, Sun Q, Bukowski R, Bell JT, et al. Host genetic variation impacts microbiome composition across human body sites. Genome Biol. 2015;16:191. doi:10.1186/s13059-015-0759-1. 28. Grosser S, Sauer J, Paijmans AJ, Caspers BA, Forcada J, Wolf JBW, et al. Fur seal microbiota are shaped by the social and physical environment, show mother–offspring similarities and are associated with host genetic quality. Mol Ecol. 2019;28:2406–22. 46 29. Cuscó A, Belanger JM, Gershony L, Islas-Trejo A, Levy K, Medrano JF, et al. Individual signatures and environmental factors shape skin microbiota in healthy dogs. Microbiome. 2017;5:139. doi:10.1186/s40168-017-0355-6. 30. Rothschild D, Weissbrod O, Barkan E, Kurilshikov A, Korem T, Zeevi D, et al. Environment dominates over host genetics in shaping human gut microbiota. Nature. 2018;555:210. https://doi.org/10.1038/nature25973. 31. Carvalho VM, Vanstreels RET, Paula CD, Kolesnikovas CKM, Ramos MCC, Coutinho SD, et al. Nasal, oral and rectal microbiota of Black lion tamarins (Leontopithecus chrysopygus). Brazilian J Microbiol. 2014;45:1531–9. 32. Bik EM, Costello EK, Switzer AD, Callahan BJ, Holmes SP, Wells RS, et al. Marine mammals harbor unique microbiotas shaped by and yet distinct from the sea. Nat Commun. 2016;7:10516. doi:10.1038/ncomms10516. 33. Cheng Y, Fox S, Pemberton D, Hogg C, Papenfuss AT, Belov K. The Tasmanian devil microbiome-implications for conservation and management. Microbiome. 2015;3:76. doi:10.1186/s40168-015-0143-0. 34. Dietrich M, Kearney T, Seamark ECJ, Markotter W. The excreted microbiota of bats: Evidence of niche specialisation based on multiple body habitats. FEMS Microbiol Lett. 2017;364:1–7. 35. Dietrich M, Kearney T, Seamark ECJ, Paweska JT, Markotter W. Synchronized shift of oral, faecal and urinary microbiotas in bats and natural infection dynamics during seasonal reproduction. R Soc Open Sci. 2018;5:180041. 36. Strube ML, Hansen JE, Rasmussen S, Pedersen K. A detailed investigation of the porcine skin and nose microbiome using universal and Staphylococcus specific primers. Sci Rep. 2018;8:1–9. 37. Huttenhower C, Gevers D, Knight R, Abubucker S, Badger JH, Chinwalla AT, et al. Structure, function and diversity of the healthy human microbiome. Nature. 2012;486:207–14. doi:10.1038/nature11234. 38. Li H, Li T, Berasategui A, Rui J, Zhang X, Li C, et al. Gut region influences the diversity and interactions of bacterial communities in pikas (Ochotona curzoniae and Ochotona daurica). FEMS Microbiol Ecol. 2017;93:1–13. 39. He J, Yi L, Hai L, Ming L, Gao W, Ji R. Characterizing the bacterial microbiota in different gastrointestinal tract segments of the Bactrian camel. Sci Rep. 2018;8:1–9. 40. Dougal K, Harris PA, Edwards A, Pachebat JA, Blackmore TM, Worgan HJ, et al. A comparison of the microbiome and the metabolome of different regions of the equine hindgut. FEMS Microbiol Ecol. 2012;82:642–52. 41. Greene LK, Mckenney EA. The inside tract: The appendicular, cecal, and colonic microbiome of captive aye-ayes. Am J Phys Anthropol. 2018;166:1–8. 42. Kamus LJ, Theoret C, Costa MC. Use of next generation sequencing to investigate the 47 microbiota of experimentally induced wounds and the effect of bandaging in horses. PLoS One. 2018;13:1–18. 43. Hoffmann AR, Patterson AP, Diesel A, Lawhon SD, Ly HJ, Stephenson CE, et al. The skin microbiome in healthy and allergic dogs. PLoS One. 2014;9. 44. Mills G, Hofer H. Hyaenas: Status survey and conservation action plan. Gland, Switzerland: IUCN; 1998. 45. Kruuk H. The spotted hyena : a study of predation and social behavior. Chicago: University of Chicago Press; 1972. 46. Frank LG. Social organization of the spotted hyaena Crocuta crocuta. II. Dominance and reproduction. Anim Behav. 1986;34:1510–27. 47. Smith JE, Kolowski JM, Graham KE, Dawes SE, Holekamp KE. Social and ecological determinants of fission–fusion dynamics in the spotted hyaena. Anim Behav. 2008;76:619–36. doi:10.1016/J.ANBEHAV.2008.05.001. 48. Höner OP, Wachter B, East ML, Streich WJ, Wilhelm K, Burke T, et al. Female mate-choice drives the evolution of male-biased dispersal in a social mammal. Nature. 2007;448:798–801. doi:10.1038/nature06040. 49. Smale L, Frank LG, Holekamp KE. Ontogeny of dominance in free-living spotted hyaenas: juvenile rank relations with adult females and immigrant males. Anim Behav. 1993;46:467–77. doi:10.1006/anbe.1993.1215. 50. Holekamp KE, Smale L. Ontogeny of dominance in free-living spotted hyaenas: juvenile rank relations with other immature individuals. Anim Behav. 1993;46:451–66. 51. Holekamp KE, Smale L, Szykman M. Rank and reproduction in the female spotted hyaena. J Reprod Fertil. 1996;108:229–37. 52. Hofer H, East M. Population dynamics, population size, and the commuting system of Serengeti spotted hyaenas. In: Sinclair ARE, Arcese P, editors. Serengeti II : dynamics, management, and conservation of an ecosystem. Chicago: University of Chicago Press; 1995. p. 665. 53. Gersick AS, Cheney DL, Schneider JM, Seyfarth RM, Holekamp KE. Long-distance communication facilitates cooperation among wild spotted hyaenas, Crocuta crocuta. Anim Behav. 2015;103:107–16. doi:10.1016/j.anbehav.2015.02.003. 54. Burgener N, East ML, Hofer H, Dehnhard M. Do Spotted Hyena Scent Marks Code for Clan Membership? In: Hurst JL, Beynon RJ, Roberts SC, Wyatt TD, editors. Chemical Signals in Vertebrates 11. New York, NY: Springer; 2008. p. 169–77. doi:10.1007/978-0-387-73945-8_16. 55. Drea CM, Vignieri SN, Kim HS, Weldele ML, Glickman SE. Responses to olfactory stimuli in spotted hyenas (Crocuta crocuts): II. Discrimination of conspecific scent. J Comp Psychol. 2002;116:342–9. doi:10.1037/0735-7036.116.4.342. 56. East ML, Burgener N, Hofer H. Sex Differences in Olfactory Behaviours Reflect the 48 Importance of Scent Marking for Social Integration in Adult Females and Competition Between Reproductively Active Males in the Spotted Hyena. In: East M, Dehnhar M, editors. Chemical Signals in Vertebrates 12. New York, NY: Springer; 2013. p. 149–60. http://link.springer.com/10.1007/978-1-4614-5927-9_11. Accessed 29 Aug 2019. 57. Woodmansee KB, Zabel CJ, Glickman SE, Frank LG, Keppel G. Scent marking (pasting) in a colony of immature spotted hyenas (Crocuta crocuta): a developmental study. J Comp Psychol. 1991;105:10–4. http://www.ncbi.nlm.nih.gov/pubmed/2032450. Accessed 29 Aug 2019. 58. Theis KR, Schmidt TM, Holekamp KE. Evidence for a bacterial mechanism for group- specific social odors among hyenas. Sci Rep. 2012;2:615. doi:10.1038/srep00615. 59. Heitlinger E, Ferreira SCM, Thierer D, Hofer H, East ML. The Intestinal Eukaryotic and Bacterial Biome of Spotted Hyenas: The Impact of Social Status and Age on Diversity and Composition. Front Cell Infect Microbiol. 2017;7. 60. Frank LG, Glickman SE, Powch I. Sexual dimorphism in the spotted hyaena (Crocuta crocuta). J Zool. 1990;221:308–13. doi:10.1111/j.1469-7998.1990.tb04001.x. 61. Strauss ED, Holekamp KE. Inferring longitudinal hierarchies: Framework and methods for studying the dynamics of dominance. J Anim Ecol. 2019;88:521–36. doi:10.1111/1365- 2656.12951. 62. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. Development of a Dual- Index Sequencing Strategy and Curation Pipeline for Analyzing Amplicon Sequence Data on the MiSeq Illumina Sequencing Platform. Appl Environ Microbiol. 2013;79:5112–20. doi:10.1128/AEM.01043-13. 63. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, et al. Ultra-high- throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 2012;6:1621–4. doi:10.1038/ismej.2012.8. 64. Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol. 2009;75:7537–41. doi:10.1128/AEM.01541-09. 65. Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 2013;41:D590-6. doi:10.1093/nar/gks1219. 66. Edgar RC, Haas BJ, Clemente JC, Quince C, Knight R. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics. 2011;27:2194–200. doi:10.1093/bioinformatics/btr381. 67. Cole JR, Wang Q, Fish JA, Chai B, McGarrell DM, Sun Y, et al. Ribosomal Database Project: data and tools for high throughput rRNA analysis. Nucleic Acids Res. 2014;42:D633– 42. doi:10.1093/nar/gkt1244. 68. Westcott SL, Schloss PD. OptiClust, an Improved Method for Assigning Amplicon-Based 49 Sequence Data to Operational Taxonomic Units. mSphere. 2017;2:e00073-17. doi:10.1128/mSphereDirect.00073-17. 69. Glassing A, Dowd SE, Galandiuk S, Davis B, Chiodini RJ. Inherent bacterial DNA contamination of extraction and sequencing reagents may affect interpretation of microbiota in low bacterial biomass samples. Gut Pathog. 2016;8:24. doi:10.1186/S13099-016-0103-7. 70. Salter SJ, Cox MJ, Turek EM, Calus ST, Cookson WO, Moffatt MF, et al. Reagent and laboratory contamination can critically impact sequence-based microbiome analyses. BMC Biol. 2014;12:87. doi:10.1186/s12915-014-0087-z. 71. Li J, Jia H, Cai X, Zhong H, Feng Q, Sunagawa S, et al. An integrated catalog of reference genes in the human gut microbiome. Nat Biotechnol. 2014;32:834–41. doi:10.1038/nbt.2942. 72. Yadav M, Verma MK, Chauhan NS. A review of metabolic potential of human gut microbiome in human nutrition. Arch Microbiol. 2018;200:203–17. doi:10.1007/s00203-017- 1459-x. 73. Manos J, Belas R. The Genera Proteus, Providencia, and Morganella. In: Dworkin M, Falkow S, Rosenberg E, Schleifer K, Stackebrand E, editors. The Prokaryotes: Proteobacteria: Gamma Subclass. New York, NY: Springer; 2006. p. 245–69. doi:10.1007/0-387-30746-x. 74. R Core Team. R: a language and environment for statistical computing. In: R Foundation for Statistical Computing. Vienna, Austria; 2019. https://www.gbif.org/tool/81287/r-a-language-and- environment-for-statistical-computing. Accessed 28 Aug 2019. 75. Segata N, Izard J, Waldron L, Gevers D, Miropolsky L, Garrett WS, et al. Metagenomic biomarker discovery and explanation. Genome Biol. 2011;12:R60. doi:10.1186/gb-2011-12-6- r60. 76. Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw. 2015;67:1–48. doi:10.18637/jss.v067.i01. 77. Fox J, Weisberg S, Fox J. An R companion to applied regression. Thousand Oaks, CA: Sage; 2011. 78. Hothorn T, Bretz F, Westfall P. Simultaneous Inference in General Parametric Models. Biometrical J. 2008;50:346–63. doi:10.1002/bimj.200810425. 79. Oksanen J, Blanchet F, Friendly M, Kindt R, Legendre P, McGlinn D, et al. vegan: Community Ecology Package. R Packag version 24-6. 2018. https://cran.r- project.org/web/packages/vegan/index.html. 80. Anderson MJ. Distance-Based Tests for Homogeneity of Multivariate Dispersions. Biometrics. 2006;62:245–53. doi:10.1111/j.1541-0420.2005.00440.x. 81. Chen Z, Yeoh YK, Hui M, Wong PY, Chan MCW, Ip M, et al. Diversity of macaque microbiota compared to the human counterparts. Sci Rep. 2018;8:15573. doi:10.1038/s41598- 018-33950-6. 82. Costello E, Stagaman K, Dethlefsen L, Bohannan B, Relman D. The Application of 50 Ecological Theory Toward an Understanding of the Human Microbiome. Science (80- ). 2012;336:1255–62. http://science.sciencemag.org/content/sci/336/6086/1255.full.pdf. Accessed 28 Jul 2017. 83. Nemergut DR, Schmidt SK, Fukami T, O’Neill SP, Bilinski TM, Stanish LF, et al. Patterns and processes of microbial community assembly. Microbiol Mol Biol Rev. 2013;77:342–56. doi:10.1128/MMBR.00051-12. 84. Widder S, Allen RJ, Pfeiffer T, Curtis TP, Wiuf C, Sloan WT, et al. Challenges in microbial ecology: building predictive understanding of community function and dynamics. ISME J. 2016;10:2557–68. doi:10.1038/ismej.2016.45. 85. García-Bayona L, Comstock LE. Bacterial antagonism in host-associated microbial communities. Science (80- ). 2018;361:eaat2456. 86. Pereira FC, Berry D. Microbial nutrient niches in the gut. Environ Microbiol. 2017;19:1366– 78. doi:10.1111/1462-2920.13659. 87. Proctor DM, Relman DA. The Landscape Ecology and Microbiota of the Human Nose , Mouth , and Throat. Cell Host Microbe. 2017;21:421–32. doi:10.1016/j.chom.2017.03.011. 88. Park W. Gut microbiomes and their metabolites shape human and animal health. J Microbiol. 2018;56:151–3. doi:10.1007/s12275-018-0577-8. 89. Cooper SM, Holekamp KE, Smale L. A seasonal feast: long-term analysis of feeding behaviour in the spotted hyaena ( Crocuta crocuta ). Afr J Ecol. 1999;37:149–60. doi:10.1046/j.1365-2028.1999.00161.x. 90. Hofer H, East M. The communiting system of Serengeti spotted hyaenas: how a predator copes with migratory prey. I. Social organization. Animal Behaviour. 1993;:547–57. 91. Cobaugh KL, Schaeffer SM, DeBruyn JM. Functional and structural succession of soil microbial communities below decomposing human cadavers. PLoS One. 2015;10:1–20. 92. Metcalf JL, Xu ZZ, Weiss S, Lax S, Treuren W Van, Hyde ER, et al. Microbial community assembly and metabolic function during mammalian corpse decomposition. Science (80- ). 2016;351:158–62. 93. Roggenbuck M, Bærholm Schnell I, Blom N, Bælum J, Bertelsen MF, Pontén TS, et al. The microbiome of New World vultures. Nat Commun. 2014;5:1–8. 94. Biswas K, Hoggard M, Jain R, Taylor MW, Douglas RG. The nasal microbiota in health and disease: Variation within and between subjects. Front Microbiol. 2015;6:1–9. 95. Clapperton BK, Fordham RA, Sparksman RI. Preputial glands of the ferret Mustela furo (Carnivora: Mustelidae). J Zool London. 1987;212:356–61. 96. Neaves W, Griffin J, Wilson J. Sexual dimorphism of the phallus in spotted hyaena (Crocuta crocuta). J Reprod Fertil. 1980;59:509–13. 97. Ross AA, Rodrigues Hoffmann A, Neufeld JD. The skin microbiome of vertebrates. 51 Microbiome. 2019;7:79. doi:10.1186/s40168-019-0694-6. 98. Ross AA, Muller K, Weese JS, Neufeld J. Comprehensive skin microbiome analysis reveals the uniqueness of human-associated microbial communities among the class Mammalia. PNAS. 2018;115:e5786–95. doi:10.1101/201434. 99. Chiarello M, Villéger S, Bouvier C, Auguet JC, Bouvier T. Captive bottlenose dolphins and killer whales harbor a species-specific skin microbiota that varies among individuals. Sci Rep. 2017;7:1–12. 100. Council SE, Savage AM, Urban JM, Ehlers ME, Pate Skene JH, Platt ML, et al. Diversity and evolution of the primate skin microbiome. Proc R Soc B Biol Sci. 2016;283. 101. Weese JS. The canine and feline skin microbiome in health and disease. Vet Dermatol. 2013;24:137–46. 102. Ngo J, Taminiau B, Fall PA, Daube G, Fontaine J. Ear canal microbiota - a comparison between healthy dogs and atopic dogs without clinical signs of otitis externa. Vet Dermatol. 2018;29:425. doi:10.1111/vde.12674. 103. Cuscó A, Sánchez A, Altet L, Ferrer L, Francino O. Individual Signatures Define Canine Skin Microbiota Composition and Variability. Front Vet Sci. 2017;4:1–12. doi:10.3389/fvets.2017.00006. 104. Tress B, Dorn ES, Suchodolski JS, Nisar T, Ravindran P, Weber K, et al. Bacterial microbiome of the nose of healthy dogs and dogs with nasal disease. PLoS One. 2017;12:e0176736. doi:10.1371/journal.pone.0176736. 105. Dorn ES, Tress B, Suchodolski JS, Nisar T, Ravindran P, Weber K, et al. Bacterial microbiome in the nose of healthy cats and in cats with nasal disease. PLoS One. 2017;12:1– 23. 106. Adler CJ, Malik R, Browne G V., Norris JM. Diet may influence the oral microbiome composition in cats. Microbiome. 2016;4:1–9. doi:10.1186/s40168-016-0169-y. 107. Mugisha L, Köndgen S, Kaddu-Mulindwa D, Gaffikin L, Leendertz FH. Nasopharyngeal colonization by potentially pathogenic bacteria found in healthy semi-captive wild-born chimpanzees in Uganda. Am J Primatol. 2014;76:103–10. 108. Lory S. The Family Staphylococcaceae. In: Rosenberg E, DeLong E, Lory S, Stackebrandt E, Thompson E, editors. The Prokaryotes: Firmicutes and Tenericutes. Berlin, Heidelberg: Springer; 2014. p. 363–6. doi:10.1007/978-3-642-30120-9_350. 109. Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR, et al. Enterotypes of the human gut microbiome. Nature. 2011;473:174–80. doi:10.1038/nature09944. 110. Sun B, Wang X, Bernstein S, Huffman MA, Xia DP, Gu Z, et al. Marked variation between winter and spring gut microbiota in free-ranging Tibetan Macaques (Macaca thibetana). Sci Rep. 2016;6:1–8. doi:10.1038/srep26035. 111. Li Q, Lauber CL, Czarnecki-Maulden G, Pan Y, Hannah SS. Effects of the Dietary Protein 52 and Carbohydrate Ratio on Gut Microbiomes in Dogs of Different Body Conditions. MBio. 2017;8:1–14. 112. Courchay M. Molecular characterization of the penile microbiome of Dorper rams ( Ovis aries ). Stellenbosch University; 2017. 113. Ström Holst B, Bergström A, Lagerstedt AS, Karlstam E, Englund L, Båverud V. Characterization of the bacterial population of the genital tract of adult cats. Am J Vet Res. 2003;64:963–8. 114. Costello EK, Lauber CL, Hamady M, Fierer N, Gordon JI, Knight R. Bacterial Community Variation in Human Body Habitats Across Space and Time. Science (80- ). 2009;326:1694–7. doi:10.1126/science.1177486. 115. Li Y, Zhang T, Qi L, Yang S, Xu S, Cha M, et al. Microbiota Changes in the Musk Gland of Male Forest Musk Deer During Musk Maturation. Front Microbiol. 2018;9:1–9. doi:10.3389/fmicb.2018.03048. 116. Xue Z, Zhang W, Wang L, Hou R, Zhang M, Fei L, et al. The Bamboo-Eating Giant Panda Harbors a Carnivore-Like Gut Microbiota, with Excessive Seasonal Variations. MBio. 2015;6:1– 12. 117. Han S, Guan Y, Dou H, Yang H, Yao M, Ge J, et al. Comparison of the fecal microbiota of two free-ranging Chinese subspecies of the leopard (Panthera pardus) using high-throughput sequencing. PeerJ. 2019;7:e6684. doi:10.7717/peerj.6684. 118. Menke S, Meier M, Mfune JKE, Melzheimer J, Wachter B, Sommer S. Effects of host traits and land-use changes on the gut microbiota of the Namibian black-backed jackal (Canis mesomelas). FEMS Microbiol Ecol. 2017;93:1–16. 119. Wasimuddin, Menke S, Melzheimer J, Thalwitzer S, Heinrich S, Wachter B, et al. Gut microbiomes of free-ranging and captive Namibian cheetahs: Diversity, putative functions and occurrence of potential pathogens. Mol Ecol. 2017;26:5515–27. 120. He F, Liu D, Zhang L, Zhai J, Ma Y, Xu Y, et al. Metagenomic analysis of captive Amur tiger faecal microbiome. BMC Vet Res. 2018;14:1–8. 121. Kim J, An J-U, Kim W, Lee S, Cho S. Differences in the gut microbiota of dogs (Canis lupus familiaris) fed a natural diet or a commercial feed revealed by the Illumina MiSeq platform. Gut Pathog. 2017;9:68. doi:10.1186/s13099-017-0218-5. 122. Yao CK, Muir JG, Gibson PR. Review article: Insights into colonic protein fermentation, its modulation and potential health implications. Aliment Pharmacol Ther. 2016;43:181–96. 123. Korpela K. Diet, Microbiota, and Metabolic Health: Trade-Off Between Saccharolytic and Proteolytic Fermentation. Annu Rev Food Sci Technol. 2018;9:65–84. 124. Diether N, Willing B. Microbial Fermentation of Dietary Protein: An Important Factor in Diet–Microbe–Host Interaction. Microorganisms. 2019;7:19. 125. Dai ZL, Wu G, Zhu WY. Amino acid metabolism in intestinal bacteria: Links between gut 53 ecology and host health. Front Biosci. 2011;16:1768–86. 126. Troccaz M, Gaïa N, Beccucci S, Schrenzel J, Cayeux I, Starkenmann C, et al. Mapping axillary microbiota responsible for body odours using a culture-independent approach. Microbiome. 2015;3:1–15. 127. Greene LK, Bornbusch SL, McKenney EA, Harris RL, Gorvetzian SR, Yoder AD, et al. The importance of scale in comparative microbiome research: New insights from the gut and glands of captive and wild lemurs. Am J Primatol. 2019;:e22974. doi:10.1002/ajp.22974. 128. Sakamoto M. The Family Porphyromonadaceae. In: Rosenberg E, DeLong EF, Lory S, Stackebrandt E, Thompson F, editors. The Prokaryotes: Other Major Lineages of Bacteria and The Archaea. Berlin, Heidelberg: Springer; 2014. p. 811–24. doi:10.1007/978-3-642-38954- 2_132. 129. James AG, Hyliands D, Johnston H. Generation of volatile fatty acids by axillary bacteria. Int J Cosmet Sci. 2004;26:149–56. 130. Ezaki T, Kawamura Y, Li N, Li ZY, Zhao L, Shu SE. Proposal of the genera Anaerococcus gen. nov., Peptoniphilus gen. nov. and Gallicola gen. nov for members of the genus Peptostreptococcus. Int J Syst Evol Microbiol. 2001;51:1521–8. 131. Albone ES, Shirley SG. Mammalian semiochemistry : the investigation of chemical signals between mammals. Chichester: Wiley; 1984. 132. Gorman M, Nedwell DB, Smith RM. An analysis of the contents of the anal scent pockets of Herpestes auropunctatus (Carnivora: Viverridae). J Zool. 2009;172:389–99. doi:10.1111/j.1469-7998.1974.tb04115.x. 133. Albone ES, Eglinton G, Walker JM, Ware GC. The anal sac secretion of the red fox (Vulpes vulpes); its chemistry and microbiology. A comparison with the anal sac secretion of the lion (Panthera leo). Life Sci. 1974;14:387–400. doi:10.1016/0024-3205(74)90069-1. 134. Carthey AJR, Gillings MR, Blumstein DT. The Extended Genotype: Microbially Mediated Olfactory Communication. Trends Ecol Evol. 2018;33:885–94. doi:10.1016/j.tree.2018.08.010. 135. Dominianni C, Sinha R, Goedert JJ, Pei Z, Yang L, Hayes RB, et al. Sex, body mass index, and dietary fiber intake influence the human gut microbiome. PLoS One. 2015;10:1–14. 136. Liu X, Fan P, Yi L, Zhao N, Garber PA. Fecal bacterial diversity of wild Sichuan snub- nosed monkeys ( Rhinopithecus roxellana ). Am J Primatol. 2018;80:1–9. 137. Ren T, Boutin S, Humphries MM, Dantzer B, Gorrell JC, Coltman DW, et al. Seasonal, spatial, and maternal effects on gut microbiome in wild red squirrels. Microbiome. 2017;5:163. doi:10.1186/s40168-017-0382-3. 138. Voigt CC, Caspers B, Speck S. Bats, Bacteria, and Bat Smell: Sex-Specific Diversity of Microbes in a Sexually Selected Scent Organ. J Mammal. 2005;86:745–9. 139. Aivelo T, Norberg A. Parasite–microbiota interactions potentially affect intestinal communities in wild mammals. J Anim Ecol. 2018;87:438–47. 54 140. Grieneisen LE, Livermore J, Alberts S, Tung J, Archie EA. Group Living and Male Dispersal Predict the Core Gut Microbiome in Wild Baboons. Integr Comp Biol. 2017;57:770– 85. 141. Turner JW, Bills PS, Holekamp KE. Ontogenetic change in determinants of social network position in the spotted hyena. Behav Ecol Sociobiol. 2018;72:10. http://link.springer.com/10.1007/s00265-017-2426-x. Accessed 22 Jan 2018. 142. Theis KR, Heckla AL, Verge JR, Holekamp KE. The Ontogeny of Pasting Behavior in Free- living Spotted Hyenas, Crocuta crocuta. In: Hurst J, Beynon R, Roberts S, Wyatt T, editors. Chemical Signals in Vertebrates 11. New York, NY: Springer; 2008. p. 179–87. doi:10.1007/978-0-387-73945-8_17. 143. Langille MGI, Meehan CJ, Koenig JE, Dhanani AS, Rose RA, Howlett SE, et al. Microbial shifts in the aging mouse gut. Microbiome. 2014;2:1–12. http://www.microbiomejournal.com/content/2/1/50. 144. Smith SC, Chalker A, Dewar ML, Arnould JPY. Age-related differences revealed in australian fur seal arctocephalus pusillus doriferus gut microbiota. FEMS Microbiol Ecol. 2013;86:246–55. 145. Uchihashi M, Bergin IL, Bassis CM, Hashway SA, Chai D, Bell JD. Influence of age, reproductive cycling status, and menstruation on the vaginal microbiome in baboons (Papio anubis). Am J Primatol. 2015;77:563–78. 146. Sin YW, Buesching CD, Burke T, MacDonald DW. Molecular characterization of the microbial communities in the subcaudal gland secretion of the European badger (Meles meles). FEMS Microbiol Ecol. 2012;81:648–59. 147. Jia T, Zhao S, Knott K, Li X, Liu Y, Li Y, et al. The gastrointestinal tract microbiota of northern white- cheeked gibbons ( Nomascus leucogenys ) varies with age and captive condition. Sci Rep. 2018;8:1–14. doi:10.1038/s41598-018-21117-2. 148. Reveles KR, Patel S, Forney L, Ross CN. Age-related changes in the marmoset gut microbiome. Am J Primatol. 2019;81. 149. Mizukami K, Uchiyama J, Igarashi H, Murakami H, Shima A, Ishiahra G, et al. Age-related analysis of the gut microbiome in a purebred dog colony. FEMS Microbiol Lett. 2019;366. 150. Kaakoush NO. Insights into the Role of Erysipelotrichaceae in the Human Host. Front Cell Infect Microbiol. 2015;5:1–4. doi:10.3389/fcimb.2015.00084. 151. Turnbaugh PJ, Bäckhed F, Fulton L, Gordon JI. Diet-Induced Obesity Is Linked to Marked but Reversible Alterations in the Mouse Distal Gut Microbiome. Cell Host Microbe. 2008;3:213– 23. doi:10.1016/j.chom.2008.02.015. 152. Bermingham EN, Maclean P, Thomas DG, Cave NJ, Young W. Key bacterial families (Clostridiaceae, Erysipelotrichaceae and Bacteroidaceae) are related to the digestion of protein and energy in dogs. PeerJ. 2017;5:e3019. doi:10.7717/peerj.3019. 153. Hofer H, East ML. The components of parental care and their fitness consequences: a life 55 history perspective. Verh Deut Zool Ges. 1996;89:149–64. 154. Kolodny O, Weinberg M, Reshef L, Harten L, Hefetz A, Gophna U, et al. Coordinated change at the colony level in fruit bat fur microbiomes through time. Nat Ecol Evol. 2019;3. doi:10.1038/s41559-018-0731-z. 155. Antwis RE, Lea JMD, Unwin B, Shultz S. Gut microbiome composition is associated with spatial structuring and social interactions in semi-feral Welsh Mountain ponies. Microbiome. 2018;6:1–11. 156. Trosvik P, Muinck EJ De, Rueness EK, Fashing PJ, Beierschmitt EC, Callingham KR, et al. Multilevel social structure and diet shape the gut microbiota of the gelada monkey , the only grazing primate. Microbiome. 2018;6:1–18. 157. Raulo A, Ruokolainen L, Lane A, Amato K, Knight R, Leigh S, et al. Social behaviour and gut microbiota in red-bellied lemurs ( Eulemur rubriventer ): In search of the role of immunity in the evolution of sociality. J Anim Ecol. 2017;87:1–12. doi:10.1111/1365-2656.12781. 158. Wos-Oxley ML, Chaves-Moreno D, Jáuregui R, Oxley APA, Kaspar U, Plumeier I, et al. Exploring the bacterial assemblages along the human nasal passage. Environ Microbiol. 2016;18:2259–71. 159. Suzuki TA. Links between Natural Variation in the Microbiome and Host Fitness in Wild Mammals. Integr Comp Biol. 2017;57:1–14. doi:10.1093/icb/icx104. 160. Ren T. Ecological forces shaping gut microbiota in wild animal populations. University of Virginia; 2016. https://libraetd.lib.virginia.edu/public_view/z603qx417. 161. Grieneisen LE, Archie EA. Social, environmental and genetic predictors of microbiome composition in wild baboons. University of Notre Dame; 2017. 162. Björk JR, Dasari M, Grieneisen L, Archie EA. Primate microbiomes over time: Longitudinal answers to standing questions in microbiome research. Am J Primatol. 2019;:e22970. doi:10.1002/ajp.22970. 56 ABSTRACT GUT MICROBIOME STRUCTURE AND FUNCTION VARIES WITH HOST ECOLOGY ACROSS MULTIPLE GENERATIONS OF WILD SPOTTED HYENAS CHAPTER TWO: Rojas, C.A., K.E. Holekamp, V. Souza, M. Viladomat-Jasso, and K.R. Theis. The gut microbiome performs an array of important functions for their hosts. Although much is known regarding the host-related and ecological factors potentially shaping the gut microbiome in a diverse array of animals, longitudinal research on the stability and functional repertoire of the gut microbiome is lacking, particularly across generations. Here, we address these gaps in knowledge, and use both amplicon and metagenomic sequencing to profile gut microbiome taxonomic composition and metabolic function across three generations of wild spotted hyenas from four matrilines residing in the Masai Mara National Reserve. Specifically, we a) assay temporal variability in the gut microbiota over time within and across individuals, b) determine whether local environmental conditions and prey abundance outweigh host kinship or age in predicting gut microbiota structure, and c) characterize the functional repertoire of the gut microbiome and elucidate microbial functions potentially implicated in host digestion. Our results showed that the taxonomic composition of the gut microbiota was highly variable across the two decades of sampling, and few microbes were retained within individuals over time. Furthermore, shifts in the abundances of bacterial groups were associated with short-term and long-term ecological changes occurring in the reserve. Unlike gut microbiota taxonomic profiles, metabolic functional profiles exhibited considerable redundancy and stability among individuals over time. Nonetheless, functional profiles also varied with host identity and prey abundance, indicating that gut microbiomes are also flexible. Lastly, the most abundant metabolic pathways of the hyena gut microbiome were related to the digestion of protein, and the synthesis of secondary antibiotics, which has implications for host health. 57 INTRODUCTION Across mammals, the taxonomic composition of gut bacterial communities (hereafter termed the “microbiota”) varies widely among and across individuals over time, due to the shifts in host diet, reproductive state, age, disease, habitat and social interactions that occur across the host’s lifespan [1–10]. This plasticity in gut microbiota composition is thought to be adaptive for the host, as it allows hosts to accommodate changes (e.g., seasonal shifts in diet; parasite infection) in their internal and external environments [11, 12]. Thus, to accurately capture the dynamism of the gut microbiota and the host predictors that affect its assembly and structure, longitudinal studies and repeated sampling of individuals are required. So far, studies have conducted repeated sampling of the gut microbiota in individuals on the order of weeks, months, or years. These studies for example, examined the progression of the gut microbiota across gestation in cattle [13], following the onset of ulcerative colitis in humans [14], post-helminth infection in red voles [15], and from birth to weaning in goat kids [16]. The longest survey of the gut microbiota to date was conducted across 13 years in a population of wild baboons [17]. However, to our knowledge, no study thus far has surveyed the gut microbiota across individuals over multiple decades and generations in a wild mammal. Thus, our understanding of the long-term dynamics of the gut microbiome and its potential effects on host functioning are limited. Unlike the gut microbiota, the metabolic functional capacity of gut bacterial communities (hereafter termed the “microbiome”), may be more stable within individuals over time, at least in humans [18, 19]. This stability may arise from functional redundancy in microbial metabolisms; each metabolic function can be performed by multiple taxonomically distinct microbes that can coexist or occupy specific niches within the human gut [20]. From a host perspective, functional redundancy among microbes enables metabolic functions to be maintained within the gut over time despite fluctuations in the taxonomic composition of 58 the gut microbiota. This is particularly important as the metabolic activities of the gut microbiome directly affect host physiology and fitness [21–23]. In Malayan pangolins (Manis javanica), and potentially other myrmecophagous (i.e. ant and termite-eating) mammals, the functional gene repertoire of the gut microbiome includes over 114 gene modules related to the synthesis of chitin-degrading enzymes and digestion of insect exoskeletons [24]. Stephen’s woodrat (Neotoma stephensi) are dietary specialists of juniper and their gut microbiome harbors genes associated with the detoxification of juniper secondary metabolites (oxalate, monoterpenes and phenolics) [25]. Experiments show that antibiotics impair the woodrat’s ability to feed on juniper, and transfer of the gut microbiotas of experienced to naïve animals maintains the microbiome’s capacity for detoxification. Lastly, in the gut of humans, who are more generalist feeders, fermentative microbes produce short-chain fatty-acids (SCFAs) that are used by the host as substrates for glucose, cholesterol, and lipid metabolism [26, 27]. Thus, it is evident that gut microbes and their functional repertoires are essential for host health, metabolism, and homeostasis. Nonetheless, the extent to which the gut microbiome changes over time is largely unknown, particularly in wild mammals. We also do not know whether the research findings that have been primarily observed in humans and laboratory animals also manifest in wild systems. Here, we fill these gaps in knowledge and employ over two decades of data collected by the Mara Hyena Project to conduct a longitudinal analysis of the gut microbiota and gut microbiome in three generations of wild spotted hyenas inhabiting the Masai Mara National Reserve, Kenya (MMNR) (Table 2.1). Spotted hyenas are large, gregarious carnivores and apex predators and make an excellent system for assaying longitudinal variation in the gut microbiota and microbiome, and for determining the relative influences of host endogenous and ecological variables on the gut microbiome. Hyenas live in matrilineal societies that are structured by linear dominance hierarchies [28–30]. An individual’s social 59 rank, which is not dependent on body size or fighting ability, determines access to resources and fitness [29, 31, 32]. Food competition is intense in this species [29, 33], and high-ranking hyenas can usurp food from all other clan members at all times. Hyenas consume medium-to-large sized antelope that they hunt themselves [34], although they may also scavenge opportunistically, and likely possess adaptations to avoid pathogen infections [35]. In the MMNR, rain is bimodal [36, 37], and during July-October of every year, it is home to millions of wildebeest and zebra that migrate from Tanzania [34, 38], which results in a prey surplus. Apart from annual ecological changes, the MMNR also underwent significant changes in its landscape and ecology over the two decades of sampling. A consistent decline in the herbivore and lion densities over time, was accompanied by an increase in anthropogenic disturbance (# of tourist lodges) and livestock grazing within the reserve over time [36, 39, 40]. With this rich dataset, we conduct the longest longitudinal survey (23 yrs) of the gut microbiota and microbiome in a wild mammal, and determine whether this variation is associated with changes in host ecology. We also identify the bacterial functional pathways that may contributing to host digestion in this hyper-carnivore, which has not been previously documented. Specifically, in this study we use 16S rRNA gene sequencing and shotgun metagenomic sequencing to survey variation in gut microbiota composition and gut microbiome function in 12 adult females over short and long temporal scales (Table 2.1, Table S2.1). Hyenas belonged to one of four matrilines that varied in their social rank, and collectively constituted three generations of sampled individuals (i.e., mother, daughter, and granddaughter were sampled from four matrilines over thirty years). First, in accordance with prior findings from human studies [18, 19], we predicted a high degree of plasticity in the taxonomic composition of the gut microbiota within and among individuals over time, but stability in gut microbiome function. Furthermore, because the gut microbiota and 60 microbiome in other mammals are strongly shaped by the host’s external environment, and because it is unknown the extent to which vertically-transmitted microbes during birth persist to adulthood [41], we predicted that hyena gut microbiota composition would be more strongly associated with host prey abundance and local environmental conditions than with maternal relatedness (e.g., kinship) or age. Lastly, we anticipated that the gut microbiome of wild spotted hyenas would reveal adaptions to the host’s carnivorous diet and be enriched in microbial pathways involved in digesting the prey consumed by this species. Collectively, our findings provide a novel perspective on the long-term dynamics of gut microbiome composition and function in a wild African carnivore, and the potential forces that shape these symbiotic gut communities. Table 2.1 Longitudinal analysis of the hyena gut microbiota. Matriline Hyena Samples years samples were collected Age range (yrs) during Shown are the number of fecal samples collected for the 12 adult female hyenas included in our study. The matriline and lifespan of each hyena is also shown, as is the time period represented by their fecal samples. M-mother, D-daughter, and G-granddaughter. See Table S1 for a more detailed, visual representation of this table. METHODS Study site and study animals The Masai Mara National Reserve, Kenya (MMNR; 1,530 km2) in southwestern Kenya (1°40’S, 35°50’E) is a rolling savanna habitat that constitutes the northernmost 61 #1 High- rank #2 Med- high rank #3 Med- low rank #4 Low- rank ID M1 D1 G1 M2 D2 G2 M3 D3 G3 M4 D4 G4 (N) 13 35 33 33 24 17 14 49 16 18 27 24 1993-1999 1999-2007, 2011 2003-2010, 2012-2015 1997-2007, 2009-2012 2006-2007, 2009-2014, 2016 2011-2016 1993-1995 1995-2009, 2011-2012, 2015 2011, 2013-2016 1993-1995, 1997-2000 1994-1997, 2000-2004, 2006 2006, 2008-2014 study 10-16 2-14 2-14 3-17 2-13 2-7 8-10 2-22 3-8 6-12 2-14 3-11 portion of the Mara-Serengeti ecosystem and boasts high densities of resident herbivores and carnivores [42–46]. The Reserve has two dry seasons (late December-March, late June-mid Nov) and two rainy seasons (late November-early December, April-early June) [36, 37]. Spotted hyenas at this location have been monitored continuously since 1988 by the Mara Hyena Project. Spotted hyenas live in large social groups, called clans, which resemble troops of cercopithecine primates [47, 48]. Clans may contain over 120 individuals, including multiple overlapping generations of adult female kin and their young, along with breeding males that immigrated from other clans [29, 40, 49]. As mentioned before, hyena societies are structured by strict linear dominance hierarchies [28–30], and ranks are not dependent on hyena size or fighting ability but are instead acquired via learning, where each new offspring comes to occupy the rank immediately below that of its mother but above that of its older siblings [49–51]. Thus, maternally related hyenas hold adjacent social ranks which may be stable for many years. While genetic relatedness is higher within matrilines than between matrilines, on average relatedness within clans is very low given that, at least in the MMNR, 97% of cubs are sired by immigrant males born in myriad neighboring clans [52]. Hyena societies are also characterized by female-dominance, male-biased dispersal [31, 32, 53], and strong fission-fusion dynamics where individuals mainly travel, rest and forage alone (“fission”) or in small subgroups that may come together (“fusion”) as larger groups, particularly during defense of resources [28, 54, 55]. Female hyenas bear litters of one or two cubs, and rear them at communal dens for the first 9 to 12 mo of life [49, 56]; dens are the places where groups often “fuse” and larger subgroups form. Hyena cubs nurse up to 24 months and reach reproductive maturity shortly thereafter, although reproduction is often delayed and most females do not bear young until they are at least 36 months of age [49, 56]. Individual hyenas can live up to 26 years in the wild [47]. In the 62 MMNR, hyenas prey mainly on topi (Damaliscus lunatus) and Thomson’s gazelles (Eudorcas thomsonii), but during July-October of every year, they switch to feed mainly on large migratory herds of wildebeest (Connochaetes taurinus) and zebra (Equus grevyi) [34, 38] coming from the Serengeti. This mass migration event, termed “the great migration,” occurs annually and lasts for approximately 4 months. Sample and meta data collection Fecal samples (N=303) were collected opportunistically from 12 adult female spotted hyenas from a single clan, inhabiting the MMNR between 1993 and 2016 (Table 2.1, Table S2.1). The hyenas were all adults at the time of sampling, and represented four matrilines that spanned 3 generations, with each sampled lineage containing a mother, daughter, and granddaughter. Individuals ranged in age from 2.4-22 yrs, and averaged ~25 fecal samples each (Table 2.1). Upon collection, fecal samples were stored in cryogenic vials in liquid nitrogen until transport on dry ice to Michigan State University, where they were stored at - 80°F until DNA extractions. In the field, hyenas are identified as individuals by their unique spot patterns, sexed based on the dimorphic morphology of their erect phalluses [57], and their birthdates calculated to ± 7 days based on their appearance as cubs when first observed [58]. Hyenas were assigned a dominance rank based on their position in a matrix ordered by submissive behaviors displayed during dyadic agonistic encounters [30]. As paternity was unknown for some of our sampled females, relatedness was estimated between individual hyenas as follows: 0 for unrelated pairs, 0.25 for grandmother-granddaughter pairs, and 0.50 for mother-daughter pairs. To assay prey abundance, three 4-km line-transects in the clan’s territory were sampled biweekly on which all mammalian herbivores were counted within 100 m of each transect centerline, as detailed in Holekamp et al. 1999 [59], and the number of herbivores was summed across the three transects. For our analyses, we calculated prey 63 abundance as the mean number of herbivores during the 30 days preceding sample collection. Additionally, we classified samples as being collected “during” the annual wildebeest migration or “outside” of the migration. To do this, we determined the start and end dates of the wildebeest migration for each year (as the first and last date where there were >100 total wildebeest across any 2 transects) and determined whether the sampling date fell within this migration interval (True or False?). Relevant metadata for each hyena fecal sample are available on the GitHub repository for this project (see Data availability). DNA extractions and genomic sequencing of amplicon and metagenomic reads Genomic DNA was extracted from fecal samples using the QIAGEN DNeasy PowerSoil Kits (QIAGEN, Valencia, CA), following the manufacturers’ recommended protocol. The order of extractions was randomized. Each sample yielded amplifiable (i.e., evident via gel electrophoresis), 16S rDNA using the bacterial-specific primers (8F: 5’ – AGAGTTTGATCCTGGCTCAG – 3’; 1492R: 5’ – ACGGCTACCTTGTTACGACTT – 3’) , while DNA extraction kit controls did not; nevertheless, the kit controls were sequenced. DNA concentrations were quantified using QUBIT (Invitrogen). All samples (N=303) were sent for 16S rRNA gene sequencing (V4 region) using the Illumina MiSeq v2 platform (250 bp length, paired end reads, pooled) at the Michigan State University Genomics Core. Sequencing, library preparation, and preliminary quality filtering were completed according to Caporaso et al. 2012 [60] and Kozich et al. 2013 [61]. A subset of fecal samples (N=32) from two mother-daughter pairs (8 samples/hyena) were also selected for shotgun metagenomics sequencing. The hyena individuals belonged to matrilines 1 and 3, and their samples were collected within 2-year periods: 2000-2001 for samples from mothers and 2013-2015 for samples from daughters. The samples were sequenced on the Illumina HiSeq 4000 platform at the Michigan State University Genomics Core (150+150 bp paired end reads). Libraries were prepared using the Rubicon ThruPLEX 64 DNA-Seq Library Preparation Kit following manufacturer’s recommendations. Base calling was done by Illumina Real Time Analysis (RTA; v2.7.7 ) and output of RTA was demultiplexed and converted to FastQ format with Illumina Bcl2fastq (v2.19.1). Sequence processing of amplicon reads Raw Illumina amplicon sequence reads were processed, filtered for quality, and classified into amplicon sequence variants (ASVs) using the Divisive Amplicon Denoising Algorithm (DADA2) pipeline in R (v3.6.2) [62, 63]. Briefly, reads were filtered for quality, allowing for 2 and 3 errors per forward and reverse read, respectively. To remove the low‐ quality portion of the sequences, forward reads were trimmed to 250 bp while reverse reads were trimmed to 220 bp. After calculating sequencing errors, reads were clustered into ASVs, and forward and reverse reads were merged into contiguous sequences. Chimeric sequences were then removed, leaving an average of 13,411 ± 5431 sequences per sample. The resulting ASVs were assigned a taxonomy using the SILVA rRNA gene reference database (v.132) [64] and those classified as Eukarya, Chloroplasts, or Mitochondria, were removed from the dataset, as were those from unknown Kingdoms. The table of ASV abundances and a table of ASV taxonomic designations are available on GitHub (see availability of data below). On average, samples retained over 64.05% (± 6.06%) of their total sequences after processing in DADA2. Two samples did not amplify well (<100 sequences after processing) and were filtered from the dataset. We exported the final ASV table, taxonomy table, and sample metadata into R (v3.6.2) [62] and these files are stored on the public GitHub repository for this project (see Data availability). Prior to analyses, we used the R decontam package [65] to identify and remove contaminant ASVs based on their prevalence in control samples compared to biological samples. We had 14 extraction kit control samples, which yielded an average of 1,860 65 sequences [range 31 – 5,646] and their 16S rRNA gene profiles differed from those of biological samples (PERMANOVA Jaccard F=3.06, p=0.0001; Bray-Curtis F=8.34, p=0.0001). A total of 4 bacterial ASVs (ASV276 Micrococcaceae, ASV1412 Planoco- ccaceae, ASV1797 Delftia, ASV1979 Stenotrophomonas) were present in at least 50% of control samples at relative abundances >1% and had decontamination scores below our specified threshold (0.5) (Fig S2.1). These ASVs were filtered from the dataset, leaving a total of 1,974 ASVs. To assess sequencing coverage, samples were temporarily subsampled to 2900 reads per sample using mothur [66]. Rarefaction curves of ASV richness reached saturation, indicating that sequencing depth was sufficient for analyzing these communities (Fig S2.2). Assessing temporal variability in gut microbiota composition Unless otherwise stated, all statistical analyses and figures were made in R. Gut microbiota composition was visualized through stacked bar plots using the ggplot2 package [67]. The plots showed the relative abundances of dominant bacterial phyla, orders, and families across time (1993-2016). Because of the high temporal variability observed in gut microbiota compositions, we also conducted linear mixed models to determine whether the abundances of particular bacterial orders or ASVs varied with the annual wildebeest migration or sample year while accounting for variation attributable to host individual identity. The model was constructed using the lmer function from the lme4 package [68], and was specified as follows, Bacterial abundance (y)~ wildebeest migration (T/F) + sample year + (1 | hyena identity). Only bacterial orders with average relative abundances >0.5% across samples and bacterial ASVs with average relative abundances >1% across samples were considered for analyses. The relative abundances of bacterial taxa and ASVs that yielded significant p-values from likelihood ratio tests on the linear models were plotted across years in ggplot2. 66 Identifying the host predictors of gut microbiota structure (b-diversity) Microbiota b-diversity was estimated using Bray-Curtis distances calculated from bacterial taxa abundance data, and Jaccard distances calculated from bacterial taxa presence / absence data using the vegan package [69]. To determine whether gut microbiota similarity was correlated with host kinship, age, prey availability, and local environmental conditions present at the time of sampling (e.g. sampling date), we conducted partial Mantel tests using vegan. For this, pairwise gut microbiota similarity values were compared against the age difference between hyena dyads, their maternal relatedness, the difference in mean monthly prey abundance between sample pairs, and the length of time elapsed between sample pairs, while accounting for variation attributable to host identity. Partial Mantel tests employed Spearman correlations and 999 permutations. To supplement these analyses, we also constructed permutational multivariate analysis of variance (PERMANOVA) tests that evaluated whether gut microbiota structure varied with host individual identity, matriline, sample month, sample year (in categories, e.g. 1993- 1995), and the annual wildebeest migration (T/F). PCoA ordinations based on Bray-Curtis distances were constructed in ggplot2 and color coded by sample year. Measuring gut microbiota stability within individuals To measure the degree of stability in hyena gut microbiotas over time, we calculated Jensen-Shannon distances (JSD) for each pair of samples from each hyena individual, also using the vegan package. These distances reflect the overall turnover between two microbial communities; a value of 1 signified that the two communities were highly heterogenous and a value of 0 indicated that the two communities were identical. JSD values for each hyena individual were visualized as boxplots in ggplot2. 67 Sequence processing of metagenomic reads A subset of samples (N=32) from two mother-daughter pairs were submitted for metagenome sequencing on the Illumina HiSeq platform. On average, these samples contained a total of ~20 million paired-end reads (range: 14-23 million) with high-quality phred scores (28-30). Trimmomatic (v0.38) [70] was used to remove sequence adapters and low-quality bases from raw reads using the program’s default parameters, leaving samples with an average of 16,374,385 sequences (± 3,049,668). To ensure that we only retained microbial DNA for analyses, reads mapping to the hyena reference genome were removed using the alignment program HISAT2 (v2.1.0) [71]. For taxonomic profiling of metagenomes, forward and reverse reads were merged with the Paired-End reAd mergeR (PEAR) [72] and assigned a taxonomy with MetaPhlAn2 [73], which utilizes unique clade- specific marker genes from bacterial and archaeal reference genomes. To functionally annotate metagenomes, reads were assembled into contigs with Megahit (v1.2.9), specifying default parameters [74], and the quality of the assemblies were evaluated using the Quality Assessment Tool for Genome Assemblies (QUAST) (v.5.0.0) [75] (Table S2). We used MetaGeneMark (v.3.96) [76] to predict gene open-reading frames (ORFs) within our contigs, and CD-HIT (v.4.6.8) [77] to cluster predicted ORFs at 95% sequence identity and 90% coverage to form the final gene catalogue. These ORFs (e.g. amino acid sequences) were then functionally annotated against the Kyoto Encyclopedia of Genes and Genomes (KEGG) database on the GhostKoala online platform [78, 79]. Ghost Koala provided the abundance of ORFs mapping to specific KEGG orthologs (e.g. “genes”), and to broader microbial pathways, which are composed of KEGG orthologs. KEGG pathways are hierarchical; Level 1 pathways are the broadest (e.g., Metabolism) and Level 3 are the most specific (e.g., xylene degradation). On average, ~ 38% of the predicted ORFs from each sample could be assigned KEGG orthology (Table S2.2). 68 Taxonomic and functional analyses of metagenomic reads Two types of data were analyzed from hyena metagenomes in R: i) taxonomic profiles generated by MetaphIan2 and ii) functional KEGG gene profiles generated by GhostKoala. The taxonomic profiles had counts of bacterial taxa in each sample and the functional profiles had counts of bacterial genes in each sample. Metagenome taxonomic and functional profiles were visualized via stacked bar plots using ggplot2, which showed the a) relative abundance of metagenomic reads assigned to bacterial kingdoms and orders, or b) the relative abundance of ORFs assigned to KEGG pathways. Furthermore, linear mixed models (lme4 package) evaluated whether the abundances of 10 microbial metabolic pathways of interest (Table S2.6) varied with sample year, or mean monthly prey abundance while controlling for variation attributable to host individual identity. Because the majority of samples from this dataset were collected outside of the wildebeest migration, we used monthly prey abundance as our predictor for prey availability. If a particular pathway yielded significant likelihood ratio test statistics, their abundances were illustrated in the form of boxplots. Moreover, we conducted PERMANOVA statistics on both taxonomic and functional profiles to determine whether gut microbiome variation was associated with host individual identity, matriline, mean monthly prey abundance and year. Lastly, we generated PCoA ordinations from these taxonomic and functional profiles in ggplot2, which were based on Bray-Curtis distances [80]. Data availability Raw sequence files will be deposited in NCBI’s Sequence Read Archive (SRA) prior to submission to an academic journal. All sample metadata and R scripts for the statistical analyses and figures included in this manuscript are available on the GitHub repository for this project (https://github.com/rojascon/HyenaGutMicrobiome_AcrossGenerations). 69 Ethical approval Our research procedures were approved by the MSU IACUC on January 8, 2020 (approval no. PROTO201900126), and comply with the ethical standards set by Michigan State University, the American Society of Mammalogists [81], and Kenya. RESULTS Did the composition of the gut microbiota fluctuate across the two decades of sampling, and were changes associated with prey availability? We conducted a longitudinal survey of the gut microbiota in a wild population of adult female hyenas across 23 years. Given that the gut microbiota of many mammalian species are highly plastic, and that the Masai Mara reserve underwent significant changes over the sampling period, we expected a high degree of temporal variability in hyena gut microbiotas. Gut microbiota 16S rRNA gene profiles revealed that they were dominated by a single bacterial phylum, Firmicutes (>70% average relative abundance across samples), but also harbored notable proportions of Actinobacteria (11%) and Bacteroidetes (8%) (Fig 2.1A). Fusobacteria and Proteobacteria were found at much lower abundances (<3%), while all other phyla were part of the rare biosphere (<1% relative abundance for each). However, the relative abundances of these bacterial phyla varied over the two decades of sampling (Fig 2.1A). Specifically, Bacteroidetes and Fusobacteria abundances increased over time and Actinobacteria abundances decreased over time among hyena individuals. Similar patterns were evident at the bacterial order and family levels. Generally, when Bacillales abundances decreased, Clostridiales abundances increased, and starting in ~2008, the abundances of Bacteroidales (Prevotellaceae) and Fusobacteriales (Fusobacteriaceae) significantly increased while those of Bacillales (Planococcaceae) decreased (Fig 2.1B, Fig S2.3). 70 Next, we tested whether the observed temporal variation in bacterial abundances was associated with prey availability, specifically with the arrival of large migratory herds of wildebeest and zebra from the Serengeti. We constructed linear mixed models that evaluated whether the abundances of 11 bacterial orders varied with time (sample year), and during periods of the wildebeest migration. The mean abundances of 8/11 bacterial orders were highly variable across time, and these included Bacillales, Bacteroidales, Coriobacteriales, Erysipelotrichales, Fusobacteriales, Lactobacillales, Methanobacteriales, and Micrococcales (Fig 2.1C, Table S2.3). Specifically, the mean abundances of Bacteroidales, Erysipelotrichales, and Fusobacteriales increased over time, while those of Bacillales, Coriobacteriales, Lactobacillales, Methanobacteriales, and Micrococcales decreased over time (Fig 2.1C). Many of these taxa underwent dramatic shifts in abundance during 2007-2009. For example, Bacillales were moderately abundant (~10- 20% relative abundance) during 1993-2006, but in 2007, their abundances decreased precipitously to <3% and remained unchanged in the following years. Lastly, linear models indicated that the abundances of Bacteroidales and Clostridiales increased during periods of the wildebeest migration while the opposite was true of Bifidobacteriales and Lactobacillales abundances (Table S2.3, Fig 2.1D). 71 Figure 2.1 Temporal variability in the gut microbiota composition of wild spotted hyenas. B) 1993 2000 2005 2010 2015 1993 2000 2005 2010 2015 A) C) D) E) During the Migration FALSE TRUE During the Migration 72 Figure 2.1 (cont’d) A)Stacked bar plots showing the relative frequency of 16S rRNA gene sequences assigned to each bacterial phylum across samples. Samples are ordered by sampling date, and each color represents a bacterial phylum. B) Stacked bar plots showing the relative frequency of 16S rRNA gene sequences assigned to each bacterial order across samples. C) Annual mean relative abundances (± standard error) of dominant bacterial orders. The abundances of these taxa varied with sample year as determined by linear mixed models. Shaded line represents a best fit elm model line. D) Average relative abundances (± standard error) of bacterial orders during the annual wildebeest migration vs. periods outside of the wildebeest migration. The abundances of these 4 taxa were statistically significant between the two categories. E) Bacterial ASVs that were differentially abundant between periods of the annual wildebeest migration vs. periods outside of the migration, as determined by linear mixed models. Plot shows parameter estimates. As with bacterial orders, we also ascertained whether temporal and dietary-related variation was observed at the level of bacterial ASVs. Linear models correlated the abundance of the 19 most abundant ASVs across samples with sample year and the annual wildebeest migration (during vs. outside of the migration). We found that the abundances of 13/19 ASVs varied across the sampled years, 4 of these increased in abundance over time and 9 decreased in abundance over time (Table S2.3, Fig S2.4). Additionally, of the 13 ASVs six were overrepresented in hyena gut microbiotas when migratory wildebeest were present (Fig 2.1E), and among them were ASVs classified as Lachnospiraceae, Clostridium, or Bacteroides. Conversely, two ASVs (Enterococcus, Planococcaceae) were more abundant during the months when migratory wildebeest were absent. Within individual hyenas, gut microbiota composition was also highly variable across time. Overall, the gut microbiotas of each hyena were not stable (mean JSD: 0.75), and some hyenas tended to have slightly more variable gut microbiotas than others (LRT on GLM χ2=426.12, p<0.0001) (Fig 2.2A). Ordination plots based on Bray-Curtis dissimilarities demonstrate that adult female hyenas exhibit considerable dispersion in their gut microbiotas, again, suggesting that these are not stable over time (Fig 2.2B). Given this instability, it was not surprising that individual hyenas only retained ~4.4% of their bacterial ASVs over time (range: 35-74 total ASVs) (Table S2.4). Most of these ASVs constituted 73 small fractions of the gut community and were found at low relative abundances (<4%) across samples from each hyena individual (Fig 2.2C). A) C) B) Figure 2.2 The hyena gut microbiota is not stable within individuals over time. A) Jensen-Shannon Distances for each hyena individual. Samples are ordered by matriline (from 1-highest to 4-lowest ranking), and generation: mother (M), daughter (D), and granddaughter (G). B) PCoA ordinations based on Bray-Curtis distances for each hyena individual. Each point represents a sample and samples are color-coded by their year category. C) Number of ASVs present in 75% of samples from each hyena individual. Each bar represents an individual hyena and colors represent ASV relative abundances. Abundance categories are: <1%, 2-3.9%, 4-6.9%, 7-9.9%, and >10%. Collectively, these findings show that gut microbiota composition is highly variable across time in wild spotted hyenas. Our analyses indicated that the relative abundances of a number of bacterial ASVs fluctuated with time, but these changes were also reflected at broader bacterial taxonomic levels. During the two decades of sampling, the abundances of 74 bacterial orders varied temporally, and at times, entire bacterial groups (e.g., Bacillales) were replaced by other bacterial groups (e.g., Bacteroidales). Some of this variation was related to prey availability, and potentially, to the large-scale ecological changes happening in the reserve throughout this time (e.g. increase in anthropogenic activity and livestock grazing). Lastly, few bacterial ASVs were retained within individuals over time, indicating that hyenas gut microbiotas are plastic and are colonized by taxonomically distinct bacterial types that may or may not be performing similar functions. Is host ecology a stronger predictor of gut microbiota similarity than host kinship or age? We predicted that gut microbiota structure would be more strongly correlated with local environmental conditions present at the time of sampling (e.g., sampling date), and monthly prey abundance than with host kinship or age. To test this, we conducted partial mantel tests that evaluated whether each of these four host variables predicted gut microbiota similarity, while accounting for variation associated with individual identity. We found no relationship between gut microbiota similarity and host age, maternal relatedness or mean monthly prey abundance (Table 2.2). A significant correlation was observed between the length of time elapsed between sample pairs (e.g., sampling interval) and gut microbiota similarity (r=0.23; Fig 2.3B), such that samples that were collected around the same time were more similar than samples that were collected years apart. Table 2.2 Environmental conditions are the strongest predictors of gut microbiota similarity in adult female hyenas. Pairwise predictor Age difference (# years) Maternal relatedness (0-1) # Weeks elapsed between samples Difference in mean monthly prey abundance Jaccard r= 0.02 (p=0.85) r= 0.007 (p=0.10) r= -0.13 (p=0.001) r= 0.072 (p=0.99) Bray-Curtis r= -0.016 (p=0.19) r= -0.013 (p=0.99) r= -0.23 (p=0.001) r= 0.056 (p=0.99) 75 Table 2.2 (cont’d) Shown are results from multivariate partial mantel tests that evaluated whether host age, kinship, sampling date, or mean monthly prey abundance were correlated with gut microbiome similarity, while accounting for microbiome variation attributable to host individual identity. Partial Mantel tests were based on Jaccard distances, which use bacterial presence or absence data, and Bray-Curtis distances which take into account the abundances of bacterial taxa. Significant p-values (α=0.05) are bolded. A) B) C) Figure 2.3 Host socioecological determinants of gut microbiota similarity in wild spotted hyenas. A) PCoA plots constructed from Bray-Curtis dissimilarity matrices. Each point represents a sample and is color-coded by sample year (left) or a binary sample year category (right; samples before 2008 vs. samples after). Closeness of points indicates high community similarity. The percentage of variance accounted for by each principal-coordinate axis is shown in the axis labels. B) Average Bray-Curtis dissimilarity values as a function of the length of time between sample pairs. Microbiome similarity is highest among samples collected around same time (partial Mantel test, Bray-Curtis r=0.23, p=0.001). C) % variance explained by host socioecological predictors in a PERMANOVA model (all p-values <0.001; see Table S4 for model output). To supplement these analyses, we also constructed a global PERMANOVA model that evaluated the host predictors from above all at once, except for this test, the pairwise values for the host predictors were replaced by single values due to the nature of the statistical test. Host genetic relatedness became host matriline, the length of time between samples was replaced by categorical variables for sample year (e.g., 1993-1995) and 76 sample month (e.g., March) and the annual wildebeest migration variable indicated prey availability. On average, hyena identity explained the most variance in gut microbiota structure among individuals (9.5% variance across distance metrics), followed by sample year (4.2%), sample month (4.4%), host age (1.88%), and the annual wildebeest migration (1.47%) (Fig 2.3C, Table S2.5). An ordination plot showed that gut microbiota samples clustered by year along PCoA Axis 1 (Fig 2.3A; Fig S2.5). Furthermore, because a number of bacterial taxa underwent dramatic shifts in abundance during 2007-2009 (see first section of the Results), we also color-coded the same ordination to indicate which samples were collected prior to 2008 and which were collected after. The ordination shows a demarcation between the gut microbiota of hyenas residing in the reserve from 1993-2007 and those living in the Reserve from 2008-2016 (Fig 2.3A). Overall, beta-diversity analyses demonstrate that gut microbiota structure is strongly associated with local environmental conditions present at the time of sampling, and these effects override those of host age, kinship, and prey abundance. Gut microbiota structure also varied with sample year and shifted significantly during 2007/2008, echoing the earlier patterns observed in gut microbiota composition. Do gut microbiome functional profiles reveal adaptations to a carnivorous diet? To determine whether the hyena gut microbiome was enriched in functional pathways relevant to host digestion and physiology, a subset of samples (N=32) from two mother-daughter pairs were submitted for metagenomic sequencing. The pairs belonged to the matrilines 1 and 3, respectively, and samples from each individual spanned 2 years. A total of 322 bacterial pathways (KEGG Level 3) were represented in the gut microbiomes of adult female hyenas, and 18 of these pathways each constituted > 1% of all annotated predicted open-reading frames (ORFs). Specifically, the gut microbiome was enriched in pathways related to metabolism (23.2% of annotated ORFS), the biosynthesis 77 of antibiotics (tetracycline and vancomycin; 6.05%), secondary metabolites (4.36%), and amino acids (2.40%), trans-membrane transport (3.93%), signal transduction (2.69%), carbon metabolism specifically (2.84%), and quorum sensing (1.28%), among others (Fig 2.4A). Although bacterial pathway abundances appeared to be highly uniform across hyena individuals and matrilines (Fig 2.4A), we tested whether the abundances of specific bacterial pathways shifted with prey abundance or varied among the sampled years. The KEGG (Level 2) bacterial pathways that were evaluated included but were not limited to: carbohydrate metabolism, lipid metabolism, the biosynthesis of secondary metabolites, xenobiotics biodegradation (of pollutants and contaminants), and replication and repair (Table S2.6). Likelihood ratio tests conducted on linear mixed models demonstrated that 3 of the 10 bacterial pathways (biosynthesis of other secondary metabolites, degradation of potentially toxic xenobiotic compounds, metabolism of terpenoids) declined in abundance over time (Table S2.6, Fig 2.4B). The bacterial pathway “replication and repair” increased at higher prey abundances while “xenobiotics biodegradation” decreased at higher prey abundances (Fig 2.4B). At a more granular-level, the gut microbiomes of adult female hyenas contained a total of 6, 270 distinct functional gene orthologs. The most represented genes across samples coded for membrane transport proteins, restriction enzymes (defense against viruses), beta-galactosidase (lactose fermentation), synthetases (protein synthesis), 3- oxoacyl-ACP reductases (fatty-acid biosynthesis), iron transport proteins, DNA gyrases (DNA unwinding) and starch-binding membrane proteins (starch degradation) (Table S2.7). 78 Figure 2.4 Taxonomic and functional profiles of hyena gut metagenomes. A) B) C) D) M1 D1 M3 D3 E) M1 D1 M3 D3 M1 D1 M3 D3 A) Relative abundance of KEGG Level 3 bacterial pathways across samples. B) KEGG Level 2 pathways that vary in their relative abundances with year and mean prey abundance according to linear models. Boxplots are color-coded by generation (purple-mothers, turquoise=daughters). C) Shown are the taxonomic profiles of gut metagenomes from two hyena mother(M)-daughter(D) pairs, from a high-ranking (matriline 1) and low-ranking matriline (matriline 3). Quality-filtered metagenome reads were classified using Metaphlan2. D) Bray-Curtis PCoA ordinations based on Metaphlan2 taxonomic profiles (left), and 79 Figure 2.4 (cont’d) KEGG Orthology abundance profiles (right). E) Boxplots of Jensen-Shannon distances for metagenome taxonomic and functional profiles. Are gut microbiome functional profiles more stable than taxonomic profiles over short time-scales? Here, we evaluate whether gut microbiome functional profiles were more stable than taxonomic profiles over the 2-year period. We also determine whether host predictors including individual identity, matriline, and sample year explained similar amounts of variation in gut microbiome taxonomic vs. functional profiles. Taxonomic profiles of metagenome samples showed that the hyena gut microbiome was dominated by Clostridiales (42.62% mean relative abundance), Lactobacillales (28.03%), Enterobacteriales (6.33%), and Actinomycetales (4.78%), although the abundances of these bacterial orders varied drastically with host matriline (Fig 2.4C). Hyenas from the higher-ranking matriline (matriline 1) harbored greater abundances of Lactobacillales and Actinomycetales than hyenas from the lower-ranking matriline (matriline 3), which were instead enriched in Clostridiales and Enterobacteriales. Overall, individual identity explained the most variation in both gut microbiome taxonomic and functional profiles (16-22%) (Table 2.3). Mean monthly prey abundance explained an additional ~5.5% of the variation in taxonomic profiles, and 9% of variation in functional profiles (Table 2.3), which exceeded the little variation this variable accounted for in amplicon gut microbiota taxonomic profiles. This suggests that the effects of host ecology on gut microbiome composition are enhanced when sampling during short time scales. Host matriline was associated with 5% of the variation in gut microbiome taxonomic profiles, but none of the variation in KEGG ortholog profiles (Table 2.3, Fig 2.4C). In a PCoA ordination, 80 samples appear to cluster by hyena identity along both axes, and by matriline along axis 2 (Fig 2.4D), although there is some dispersion in the samples from each hyena individual. Table 2.3 Gut metagenome taxonomic and functional profiles are individual-specific and vary with host ecology. Taxonomic Profiles (Jaccard) 10.66, p=0.001 5.97, p=0.001 3.19, p=0.32 Taxonomic Profiles (Bray-Curtis) 16.60, p=0.001 4.84, p=0.046 5.49, p=0.023 Functional Profiles (Jaccard) 17.09, p=0.006 3.06, p=0.29 5.05, p=0.065 Functional Profiles (Bray-Curtis) 22.49, p=0.001 1.75, p=0.53 8.86, p=0.013 Host predictors hyena identity matriline (1 vs 3) mean monthly prey abundance year 3.03, p=0.38 3.26, p=0.24 5.25, p=0.054 3.68, p=0.17 Shown are the R2 values (% variance explained) and p-values for marginal PERMANOVA tests (y ~ matriline + prey abundance + hyena identity + year) based on Bray-Curtis and Jaccard distance matrices calculated from Metaphlan2 taxonomic profiles and KEGG Orthology functional profiles. The data come from 32 metagenomes from two mother-daughter pairs. In a marginal PERMANOVA, the variance explained by each term does not depend on the order of explanatory variables. Significant p-values (α=0.05) are bolded. In line with our prediction, gut microbiome functional profiles were more stable than taxonomic profiles (Figure 2.4E, Wilcoxon rank sum test on JSD distances: W=852, p=0.0001). Taxonomic profiles were highly heterogenous within and across individuals (mean JSD 0.75) while functional profiles were very consistent (mean JSD 0.25). DISCUSSION Principal findings of study Here, we used over two decades of data on a wild population of spotted hyenas to conduct the longest longitudinal survey (23 yrs) of the gut microbiota and microbiome in a wild mammal. We assayed temporal variability and stability in the gut microbiota and microbiome over both short and long temporal scales, and are the first to survey the metabolic potential of the gut microbiome in wild spotted hyenas. We found that gut microbiota composition was highly plastic across the two decades of sampling and a significant portion of this variation was attributed to local environmental conditions rather 81 than host age or kinship. These findings illustrate that over long temporal scales, the gut microbiota can shift with changes in host ecology, with much lesser influences from other host predictors. Hyena gut metagenomes harbored genes that coded for protein metabolism (fermentation, fatty-acid biosynthesis, amino acid anabolism), and the synthesis of antibiotics, suggesting direct links between gut microbiome function, host health, and possible adaptations to meat/carrion-eating. Lastly, gut metagenome taxonomic profiles were far more variable than functional profiles, potentially to accommodate changes in the host’s internal and external environment while preserving microbial function. Nonetheless, gut microbiome functional profiles did vary with host individual identity, sample year, and prey abundance, suggesting that they are also flexible to a degree. Gut microbiota composition is highly plastic over the two decades of sampling Gut microbiota composition was highly plastic among individuals over the two decades of sampling, with the abundances of entire bacterial orders and specific ASVs fluctuating over time. Additionally, within individuals, the gut microbiota was not stable and few bacterial ASVs were retained over time. This was in line with our prediction that the composition of the hyena gut microbiota would be highly variable, potentially to accommodate changes in the host’s internal and external environment, such as a change in a food source, the weather, or habitat quality. During the 23 years of sampling, the reserve experienced significant declines in herbivore and lion densities, and increases in livestock grazing and anthropogenic activity related to tourism [36, 39, 40]. At the same time, hyena group size doubled, from 60 individuals in 1990 to 120 individuals in 2013. Thus, over time, individual hyena diets changed from consisting strictly of antelope, to containing notable proportions of livestock. This potentially lead rapid and significant alterations to gut microbiota composition. Furthermore, increasing habitat disturbance due to anthropogenic activity is known to induce shifts in the gut microbiota composition of cercopithecid and 82 strepsirrhine primates, polar bears, coyotes and rodents [82–87], and this could be true of hyenas as well. Interestingly, many bacterial orders exhibited dramatic shifts in abundance during 2006-2008, which coincides with the timing of a severe drought in the reserve [36] and with the sharpest increases in livestock grazing and tourism. These changes collectively must have significantly altered the ecology of the region and affected hyena physiology in some capacity, which cascaded to changes in their gut environments and microbiome compositions. The bacterial taxa that increased over time included Bacteroides, Peptoclostridium, Lachnospiraceae, and Fusobacteria, while those that decreased over time included Paeniclostridium, Peptoniphilus, Clostridium, and Enterococcus. Bacteroides are major constituents of the mammal gut microbiota and are found exclusively in the GI tract of animals [88]. They have the capacity to use a wide range of simple and complex polysaccharides for growth, play a role in carbohydrate fermentation, and thrive in fiber- enriched environments [89]. Thus, it appears that the increase in Bacteroides over time was likely beneficial for the hyenas. Increases in Lachnospiraceae are associated with the consumption of anthropogenic food in urban coyotes [90], and physiological stress in mice [91]. Changes in Lachnospiraceae in hyenas may be a response to the increased anthropogenic disturbance in the reserve. Enteroccous was one of the bacterial taxonomic groups that declined over time in the hyena gut microbiota. Enterococcus is a speciose genus that is ubiquitous and found in soils, seawater, plants, and the guts of vertebrates and invertebrates [92]. It is unclear whether changes in Enterococcus abundances had consequences for host phenotype. Nonetheless, some of the temporal changes in the gut microbiota might simply reflect strategies by microbes to maximize their own fitness. When the hosts ingest a novel food source, microbes that can utilize that resource will increase in abundance while others 83 might be driven to local extinction. Changes in the host’s external or internal environment may open a previously unavailable niche or close an existing niche, causing decreases or increases in bacterial abundances [93]. Additionally, because of the extensive cross-feeding or competitive interactions observed in microbial communities [94], shifts in the abundances of some bacterial species will inevitably cause reductions or increases in others. Lastly, some of the fluctuations in taxa abundances could be random; stochasticity can eliminate low-abundance microbial species unless these are “rescued” by dispersing microbes from outside of the gut community [93]. Local environmental conditions predict gut microbiota variation in wild spotted hyenas Our results showed that local environmental conditions were a stronger predictor of gut microbiota structure than host kinship, age or the wildebeest migration. Hyena gut microbiotas from the same time period were more similar than hyena gut microbiotas from different time periods (r=0.23, p<0.001). That is, samples collected days apart from each other harbored more similar gut microbiotas than samples collected years apart, suggesting that local conditions at the time of sampling (of both the habitat and the hyena itself) are highly predictive of gut microbiota composition. This is consistent with prior studies that report that the environment outweighs host genetics in predicting gut microbiota variation in mammals [95]. A survey of the gut microbiota of 1,046 human individuals reported that there was significant similarity among the microbiotas of genetically unrelated individuals who share a household, but no significant similarity among relatives who did not live together [95]. Furthermore, in red squirrels, mother-offspring pairs had more similar gut microbiotas than father-offspring pairs or unrelated pairs [96]. Environmental factors, including season and year, explained 11 times more variation in gut microbiota structure than host sex, age, and kinship. 84 Nonetheless, the finding that host kinship, age, or the annual wildebeest migration did not predict much variation in the hyena gut microbiota was unexpected. Closely related hyenas typically spend more time together than unrelated hyenas, and they also occupy similar positions in the clan’s hierarchy. This means that related hyenas frequently come into close contact with one another, consume food of equal quality, and occupy similar physical spaces. Thus, we expected greater convergence in the gut microbiotas of maternally related hyenas, but this was not supported by our data. Perhaps in order to accurately measure whether genetic relatedness or other host factors predict gut microbiota variation in hyenas, we must sample a large number of hyena group-mates within a short time period. Sampling over large time-scales might obscure findings because of the high temporal variation observed in microbiota composition. The same could be said for age-related changes in the gut microbiota. Prior studies report that juvenile and adult hyenas from this population differ in the structure and diversity of their rectal microbiotas [97], and in a population of hyenas in the Serengeti, juvenile hyenas harbor less diverse gut microbiotas than adults [98]. Nonetheless, both of these studies restricted their sampling to 3 years and sampled juveniles, whereas our study contained samples that spanned 23 years and only surveyed adults. Given that the gut microbiota undergoes changes in adulthood in rodents [7], domestic dogs [99], and primates [100], we believe age-related changes may be observed in the hyena gut microbiota if sampling is conducted over a much shorter time scale. Lastly, given that host diet is a major environmental filter of the gut microbiota in mammals [101–103], it was surprising that the annual wildebeest migration explained less than 2% of the variation in the gut microbiota. Perhaps the dietary change resulting from the wildebeest migration (e.g. increased consumption of wildebeest and zebra; more meat consumption overall) was not as substantial as was anticipated, and thus, did not induce 85 changes in gut microbiome composition. Adult female hyenas eat protein year-round; during the wildebeest migration, they obtain this protein from wildebeest and zebra, and outside of the migration, they obtain it from gazelles, warthogs, and other antelope [34]. In our study, we did not sample immigrant males, which have poorer diets than adult females [32], and they potentially, would have exhibited differences in their gut microbiota during the wildebeest migration vs. outside of the migration. Gut microbiota composition reflects its host’s carnivorous diet On average, the hyena gut microbiota was dominated by Firmicutes (70%), Actinobacteria (11%), and Bacteroidetes (8%), and closely resembled the composition of black bear, jackal, and mongoose gut microbiotas [104–106]. Other carnivores including brown bears, polar bears, Tasmanian devils, tigers, and domestic dogs [84, 107–110] harbored gut microbiotas with much lower relative abundances of Firmicutes and Actinobacteria, and higher abundances of Proteobacteria (15-40%) than those of spotted hyenas (<3%). It appears that the gut microbiota of hyenas converges with those of closely related hosts (e.g., mongooses), but also overlaps with those of hosts with similar diets (e.g., jackals). This is expected, as both host phylogenetic relatedness and dietary similarity predict gut microbiome variation in a range of mammalian taxa, extending from rodents and artiodactyls, to carnivores and primates [111–114]. The gut microbiotas of hyenas are not completely identical to those of other carnivore hosts potentially due to differences among hosts in their particular diets, gut physiologies, immune components, and habitats, all of which have been shown to contribute to microbiome variation [82, 115]. Hyenas in particular, consume animal carcasses at all stages of decomposition and may contain lower gastric pH than non-scavenging animals [116]. The gut environment and diet of hyenas potentially select for slightly different gut microbes than those of other carnivores. 86 Because hyenas are carnivores, their gut microbiotas likely contain microbes capable of digesting the protein found in the muscle tissue and bone (collagen) of animal carcasses. Such bacterial taxa include Clostridiales, Lactobacillales (7.5%), Bacillales (7.4%), Erysipelotrichales (5%), and Bacteroidales (7.4%), which were all fairly abundant in hyena guts. They metabolize carbohydrates, proteins, and fats in the mammalian intestine, and in doing so, promote nutrient absorption and supply energy to their hosts [117]. Members of the order Bacillales are found in diverse environments [118], from soils and oceans to the gastrointestinal tracts of insects and mammals, and produce a wide range of antimicrobial compounds [119]. The bacterial order Erysipelotrichales is also prevalent in canids and mustelids [120], and may act as a source of the short-chain fatty acid butyrate [121]. Greater abundances of Clostridiales compared to Bacteroidales typically correspond to larger body sizes and fat stores [122]; and this was observed in female hyenas, which are very large, and can weigh up to 64kg. However, hyena gut microbiotas might also contain transient microbes from the skin, guts, and glands of their herbivore prey. Moeller and colleagues [123] surveyed the gut microbiotas of sympatric and allopatric populations of carnivore, artiodactyl, and rodent species in the Americas, and found that the gut microbiotas of sympatric predator–prey host populations were more similar than those of sympatric non-predator–prey hosts. Furthermore, predators and their preferred prey harbored compositionally similar gut microbiotas potentially indicating that predators acquire gut bacteria from their prey [123]. In our study, the gut microbiotas of hyenas harbored similar abundances of Clostridiales (53%) as the gut microbiotas of 11 species of African herbivores (53%) [124] living sympatrically with hyenas. Clostridiales spp. also form part of the vulture gut microbiota, which are carrion-eaters like hyenas [125]. This suggests that hyenas and carnivores in general may be acquiring bacteria from their prey. This might also explain why hyena guts 87 contained high abundances of bacteria that can digest fiber, cellulose, and other plant compounds, which are absent in hyena diets but are major constituents of herbivore diets. Collectively, our findings show that hyena GI tracts may be colonized by resident microbes that are adapted to their host’s meat diet, or by transient microbes that were acquired from their prey. The gut microbiome of hyenas is implicated in host digestion and in host protection from pathogens Analysis of the functional repertoire of the hyena gut microbiome demonstrated that gut microbes were enriched in functions related to metabolizing the components of their herbivore prey (e.g., protein, carbohydrates, and lipids). At a broad level, the gut microbiome was enriched in pathways related to carbon metabolism, purine metabolism, the biosynthesis of amino acids, pyruvate metabolism, and methane metabolism. The gut microbiomes of mongooses, baleen whales, Eurasian lynx, domestic dogs, feral cats, and meerkats, which are all carnivorous, also harbored a large proportion of genes related to carbohydrate and amino acid metabolism [113, 126, 127]. At a more specific level, the most abundant gene orthologs in the hyena gut microbiome coded for enzymes associated with membrane transport, restriction enzymes, DNA gyrases and polymerases, β-galactosidase, fatty-acid biosynthesis, and starch degradation. While some of these orthologs represent essential and conserved functions that would be present in the genomes of most bacteria (e.g., membrane transport, DNA gyrases and polymerases), others were more specialized and attuned to the hyena’s diet and metabolism (fatty-acid biosynthesis). A portion of the microbial functions may also be dedicated to metabolizing the food of their host’s prey; starch-binding outer membrane proteins bind plant-derived glycans for degradation [128]. Overall, it appears that the hyena gut microbiome is enriched in genes critical for bacterial survival and metabolism, but also enriched in genes that regulate energy in the intestine. 88 Interestingly, the 3rd and 4th most abundant bacterial pathways (out of 322 total pathways) in the hyena gut microbiome were related to the biosynthesis of secondary metabolites and antibiotics (tetracycline and vancomycin classes). Hyenas hunt up to 95% of their food, and scavenge as well [29], and synthesizing antibiotics may be a strategy that resident gut microbes employ to compete against the pathogenic bacteria, fungi, and parasites of animal carcasses that may be trying to establish in the gut [129–131]. Thus, it might be adaptive for both the host and gut microbes to outcompete incoming microbes. Indeed, the hyena gastrointestinal tract was not a reservoir of many ruminant, felid, or canid bacterial pathogens (data not shown). Amplicon reads were only recovered for Pasteurella multocida, Staphylococcus (both cause skin infections), and Clostridium perfringens (causes chronic diarrhea). This provides further evidence that the hyena gut microbiome is functioning to keep pathogens from colonizing in the intestine. Furthermore, a recent study that surveyed the gut microbiomes of 180 species of reptiles, birds, and mammals, reported that the gut microbiomes of carnivores relative to herbivores were enriched in ABC transporters for amino acids, the biosynthesis of metabolites with potential antibiotic and cytotoxic activity, and the degradation of xenobiotic compounds (e.g. contaminants and pollutants) [132]. These findings suggest that synthesis of antibiotics by gut microbes is a strategy employed by meat-eating animals in general, not just carrion eaters. The same study experimentally confirmed that the gut microbes of griffon vultures are capable of synthesizing proteases that can metabolize bacterial toxins from food [132]. Gut microbes in hyenas may be employing multiple strategies to fight pathogens: by outcompeting them, by synthesizing antibiotics to eradicate them, or by secreting proteases that neutralize the pathogen’s toxins. 89 Gut microbiome functional profiles are stable, yet flexible among individuals over time As predicted by our hypothesis, gut metagenome taxonomic profiles were much more variable than KEGG functional profiles among individuals and between individuals over a 2-yr time-period. The gut microbiome taxonomic profiles of hyenas had a mean JSD of 0.75, while ortholog profiles had a mean JSD of 0.25. These findings are in accordance with prior studies conducted in humans, where taxonomic gut communities can be on average 35-46 % similar among individuals while KEGG profiles are 90-96 % similar [18, 19]. The high-degree of stability in gut metagenome functional profiles could be due to the extensive functional redundancy observed in environmental-associated and host-associated microbial communities. Nonetheless, hyena gut microbiome functional profiles were also flexible to a degree, and exhibited variation among individuals. In our study, host identity, prey abundance, and sample year accounted for 22%, 9%, and 5% of the variation in gut metagenome functional profiles, respectively. This suggests that gut microbial function may shift in response to changes in host diet (e.g., prey abundance, increased livestock consumption) and host habitat (e.g., sample year), including changes in anthropogenic disturbance and rainfall (drought). Our study changes the narrative that the gut microbiome is redundant among individuals, and shows that mammalian gut microbiomes can exhibit both stability and variability, which may be adaptive for the host. Stability ensures that critical microbial functions are preserved despite disturbances to the gut community, but a small degree of flexibility allows gut microbiomes to adapt to internal and external changes. Similarly, Milani et al. (2020) found that the gut microbiome functional profiles across 24 species of mammals varied with host dietary category (carnivore, piscivore, herbivore) [127]. Carnivore gut microbiomes were enriched in pathways related to degradation of choline, an amine found in meats, whereas those of herbivores were enriched in plant 90 carbohydrate (xylose, sucrose, starch) degradation pathways. Levin et al. (2021) state that across 180 species of reptiles, birds, and mammals, gut microbiome functional profiles are significantly correlated with host diet (carnivore vs. herbivore vs. ominivore), activity hours (diurnal vs. nocturnal), social structure (solitary vs. social), and lifespan (short vs. long) [132]. Other studies also observed shifts in gut microbial function due to habitat variation. Captive panda gut microbiomes exhibit lower level of cellulase activity and are enriched in antibiotic resistance genes and virulence factors compared to their wild counterparts [133]. Limitations of the current study We only examined gut microbiome function in terms of the abundances of broad pathways and gene orthologs, which may or may not be transcriptionally active. To get more accurate representation of microbial metabolisms, future studies should survey microbial gene expression via metatranscriptomics or quantify microbial metabolites with metabolomics. Second, only 38% of our metagenomic reads (open-reading frames, ORFs) could be assigned a KEGG function, meaning that 62% of ORFs were of unknown function and were excluded from analysis. Similarly, Levin et al. (2021) [132] report that only 21% of the putative genes from 180 species of wild animals could be assigned KEGG orthology. It is evident that the microbiome of wild animals is extensively unexplored and uncharacterized. This is a limitation of the field itself; genes from wild animals can code for completely novel proteins or share weak homology with known proteins from reference genomes [134]. But these data can still be useful; for example, if one gene ortholog matches the sequence of a gene ortholog in another study, this can enable comparisons, even if the ortholog does not have a known function. Thus, despite the limitations, our study establishes a baseline understanding of gut microbiome function in a wild carnivorous mammal. Future studies can build from this study and others to uncover the roles of microbes across wild mammals. 91 CONCLUSIONS The gut microbiome is a dynamic ecosystem whose taxonomic composition or metabolic function can be substantially altered by a range of host factors at any time, either temporarily or permanently. Although the number of longitudinal studies assaying gut microbiota composition in conjunction with microbiome function is growing, it’s scarcer for large wild mammals. Here, we capitalized on long-term behavioral data available from the Mara Hyena project to examine variability in gut microbiota composition and microbiome function in three generations of adult female hyenas. We show that gut microbiota composition is temporally variable and responsive to large-scale changes occurring in the host’s external environment. Furthermore, we show that similar to the microbiomes of other carnivores, those of hyenas reveal adaptations to a meat-eating diet, and may also be playing important roles in pathogen defense. We provide novel perspectives, as we demonstrate how the gut microbiome can be both stable and functionally redundant among individuals, but is also flexible enough to accommodate future changes. This study contributes to our understanding of the potential functional contributions of microbes in wild animal systems, which is critical as the gut microbiome of wild mammals remains largely unexplored. 92 APPENDICES 93 Table S2.1 Distribution of samples from each hyena individual included in our longitudinal study. APPENDIX A: SUPPLEMENTAL TABLES Matriline #1 (high rank) M1 (13) D1 (35) G1 (33) Matriline #2 (med-high rank) M2 G2 (17) (33) D2 (24) Matriline #3 (med-low rank) G3 M3 (14) (16) D3 (49) Matriline #4 (low rank) M4 (18) D4 (27) G4 (24) 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Shown are the total number of samples and distribution of those samples across time for 12 adult female hyenas. Light gray cells indicate years that the hyena was alive and dark gray cells indicate the year of their fecal samples. M-mother, D-daughter, and G-granddaughter. 94 KEGG % ORFs L50 L75 ORFS) Table S2.2 Statistics of the metagenome assemblies as determined by QUAST (v5.0.0). KEGG output (# annotated KEGG input Number Total of M1 M1 M1 M1 M1 M1 M1 M1 D1 D1 D1 D1 D1 D1 D1 D1 M3 M3 M3 M3 M3 M3 M3 M3 D3 D3 D3 D3 D3 D3 D3 D3 202951 212832 235252 306589 213236 196377 103877 252197 243407 1455656 230935 267765 192363 196790 348023 154320 484090 233813 428253 196453 160192 261399 272051 382136 374411 371796 164015 147452 138945 347644 453528 219462 contigs 157419 153258 105626 146985 148679 159292 88127 102798 73534 101542 71563 161167 144374 110986 27596 148297 75881 191401 99020 106739 256907 175865 115577 72134 82754 328452 89683 81571 84567 183977 159114 251446 assembly length (bp) 239649058 226611515 152866106 162864880 272518472 257993258 100458702 153495267 111302462 164341142 116928938 309913433 243461158 194664744 48225319 253746786 114191190 337896444 199105890 205757238 257507466 334640232 229059157 98466255 120748979 330251560 115737053 113024299 77707193 262223723 228528374 333154588 Max contig N75 length (bp) N50 951 23705 67945 2069 935 25028 68125 1959 877 15549 45828 1986 1182 645 24559 74324 3092 1145 16717 54878 989 20646 64342 2428 1296 691 16558 44290 870 12796 42584 2164 907 2099 9713 30863 966 11865 40571 2370 2562 981 8557 28118 3371 1219 16918 57292 2720 1004 15805 55298 2774 1073 12217 42045 2848 1047 9956 2650 1063 18278 57831 2082 8863 31346 2882 1085 21025 71821 3781 1252 8362 33013 3681 1200 10137 36246 957 673 64289 1E+05 3220 1222 20476 64603 3671 1290 12617 39972 718 2059 6848 29551 859 10861 35096 2028 677 83001 2E+05 980 767 13984 41899 1550 1763 843 12749 36908 589 19559 48527 811 843 25205 80710 1842 887 25221 71055 1892 1617 808 41782 1E+05 sampleID hyenaID P102 P107 P167 P225 P295 P304 P46 P47 P178 P196 P276 P291 P294 P43 P74 P82 P143 P149 P14 P179 P212 P261 P301 P84 P186 P190 P191 P265 P266 P41 P75 P9 We filtered metagenomic reads with Trimmomatic, removed hyena DNA using HISAT2, and assembled reads into contigs with MEGAHIT. MetageneMark predicted gene open reading frames (ORFs) within contigs, and these were functionally annotated using the Kyoto Encyclopedia of Genes and Genomes (KEGG). Table S2.3 Temporal variation in the mean relative abundances of bacterial orders and ASVs. (# of ORFs) 504311 483587 370692 400325 500499 510889 277889 399915 256030 346296 250398 552309 498402 371468 97095 4850004 247499 641538 354253 366248 368244 586846 419216 250502 277157 560946 308240 262427 230058 606828 510827 803525 annonated 41.6 42.2 38.9 27.3 41 41.2 29.8 42.1 39.1 37.2 40.7 41.5 40.3 40.6 39.8 41.5 39.6 41.5 43.5 43.1 28.5 41.7 43 28.8 40.4 35 40.4 41.5 20.2 40.4 42.6 44 210033 204250 144100 109451 205288 210711 82934 168230 100162 128892 101952 229351 201045 150971 38615 201325 97982 266162 154083 157843 104880 244425 180399 72055 111983 196426 124400 108797 46430 245304 217626 353862 2575 877 Bacterial Order Annual wildebeest migration ChiSq. 3.42 Bacillales 5.92 Bacteroidales 4.02 Bifidobacteriales 6.73 Clostridiales 0.044 Coriobacteriales 0.028 Erysipelotrichales 1.08 Fusobacteriales Lactobacillales 9.13 Methanobacteriales 0.051 p.val 0.064 0.0149 0.04 0.016 0.83 0.86 0.29 0.002 0.81 Year ChiSq. 25.18 62.39 1.03 1.93 18.77 23.68 39.86 20.07 5.61 p.val <0.0001 <0.0001 0.31 0.16 <0.0001 <0.0001 <0.0001 <0.0001 0.0178 95 Table S2.3 (cont’d) Micrococcales Pseudomonadales 2.95 0.029 Bacterial ASV ASV469|Bacteroides ASV272|Micrococcaceae ASV1277|Lachnospiraceae ASV1278|Lachnospiraceae ASV1279|Lachnospiraceae ASV1499|Paeniclostridium ASV1507|Peptoclostridium ASV1512|Peptoniphilus ASV1570|Ruminococcaceae_UCG- 014 ASV852|Clostridiales_unclass ASV1629|Streptococcus ASV1413|Planococcaceae ASV1414|Planococcaceae ASV913|Clostridium_ss7 ASV936|Enterococcus ASV879|Clostridium_ss1 ASV1176|Clostridiaceae_1 ASV1280|Lachnospiraceae ASV1710|Fusobacterium 0.08 0.86 14.83 1.01 0.0001 0.31 Annual wildebeest migration p.val 0.007 0.11 0.0003 0.049 0.09 0.78 0.0001 0.45 0.93 ChiSq. 7.08 2.53 12.72 3.85 2.83 0.07 14.29 0.56 0.006 ChiSq. 42.17 11.46 0.105 6.55 2.17 4.3 13.19 8.32 2.5 0.33 1.14 1.14 6.34 0.09 6.99 14.18 2.33 4.45 1.09 0.56 5.85 0.28 0.24 0.28 17.12 0.011 13.75 0.75 10.03 0.008 28.55 0.00016 3.35 0.12 6.39 0.03 11.95 0.29 21.22 Year p.val <0.0001 <0.0001 0.74 0.01 0.14 0.038 0.0002 0.003 0.11 0.015 0.61 <0.0001 0.0002 0.001 <0.0001 0.066 0.011 0.0005 <0.0001 Shown are the LRT estimates for linear mixed effect models that tested whether the abundances of specific bacterial orders (top) or bacterial ASVs (bottom) varied across the sampled years and during periods of the annual wildebeest migration, when 1.5 million wildebeest join the resident herbivore populations for 4 months. We accounted for variation associated with individual identity by including this term as a random factor in the model. The model was specified as follows: Bacterial order or ASV abundance ~ wildebeest migration (T/F) + year + (1|hyenaID). All bacterial orders with >0.5% relative abundances and all bacterial ASVs with >1% relative abundances across samples were tested. Significant estimates (p<0.05) are bolded. 96 Table S2.4 Number of bacterial ASVs retained over time in the gut microbiota of adult female hyenas. hyena identity M1 D1 G1 M2 D2 G2 M3 D3 G3 M4 D4 G4 total ASVs 789 1192 1282 1451 1063 903 963 1519 940 976 1259 1096 # ASVs retained 49 40 55 40 43 74 38 47 60 41 35 50 % ASVs retained 6.21 3.35 4.29 2.75 4.04 8.19 3.94 3.09 6.38 4.2 2.77 4.56 Shown are the number and percentage of ASVs present in 75% of samples from each hyena individual. M-mother, D-daughter, and G-granddaughter. The numbers indicate matriline and rank, 1-highest ranking matriline and 4-lowest ranking matriline. Table S2.5 Predictors of gut microbiota structure in adult female hyenas (N=301). Jaccard (% variance explained) 8.06 (p=0.001) Bray-Curtis (% variance explained) 10.99 (p=0.001) Average (% variance) Predictor hyena identity (categorical) year (in categories, i.e. 2005-2007) month (categorical) age (yrs) wildebeest migration (T/F) 3.57 (p=0.001) 4.32 (p=0.002) 1.50 (p=0.001) 1.50 (p=0.001) 5.02 (p=0.001) 3.77 (p=0.04) 2.26 (p=0.001) 1.45 (p=0.001) 9.5 4.2 4.04 1.88 1.47 Shown are the R2 values (% variance explained) and p-values for PERMANOVA tests that evaluated whether hyena endogenous and exogenous factors predicted gut microbiota similarity. The models were ran on 2 types of distance matrices and were specified as follows: microbiome similarity ~ hyena identity + wildebeest migration + age + sample month + sample year. Bray-Curtis distances take into consideration the proportions of bacterial taxa, while Jaccard distances take into account only their presence or absence. Significant p-values (α=0.05) are bolded. Terms are ordered by their significance (% variance explained). 97 Table S2.6 The abundances of KEGG Bacterial Pathways vary temporally and with prey availability. KEGG Bacterial Pathway (Level 2) Amino acid metabolism Carbohydrate metabolism Glycan biosynthesis and metabolism Lipid metabolism Metabolism of cofactors and vitamins Biosynthesis of other secondary metabolites Xenobiotics biodegradation and metabolism Metabolism of terpenoids and polyketides Replication and repair Cell growth and death mean year monthly prey abundance ChiSq. p.val ChiSq. p.val 0.81 1.14 0.38 0.073 0.28 0.41 0.11 0.48 0.46 0.79 0.0012 0.001 0.054 3.21 1.15 2.49 0.53 10.44 0.28 0.53 0.52 0.48 0.38 0.99 5.14 0.023 7.79 0.8 0.36 4.96 0.005 0.025 5.37 1.35 0.024 2.37 0.24 0.039 0.12 0.84 Shown are the LRT estimates for linear mixed effect models that tested whether the abundances of specific bacterial pathways (KEGG Level 2) varied with monthly prey abundance, and among the sampled years. We accounted for variation associated with individual identity by including this term as a random factor in the model. The model was specified as follows: Bacterial pathway abundance ~ mean monthly prey abundance + year + (1 | hyena identity). Ten bacterial pathways potentially involved in hyena digestion were tested. Significant estimates (p<0.05) are bolded. Table S2.7 The 45 most abundant KEGG functional genes in the hyena gut microbiome. Mean Rel Abun (%) functional annotation KEGG ortholog K06147 K01990 K02003 K03657 K02004 K03088 K03701 K02337 K01153 K00558 K02469 K03046 K02355 K01190 K21636 K01952 K04759 K02015 K03043 type I restriction enzyme, R subunit 0.540 ATP-binding cassette, subfamily B, bacterial 0.430 ABC-2 type transport system ATP-binding protein 0.345 putative ABC transport system ATP-binding protein 0.339 DNA helicase II / ATP-dependent DNA helicase 0.332 putative ABC transport system permease protein 0.288 RNA polymerase sigma-70 factor 0.262 excinuclease ABC subunit A 0.257 DNA polymerase III subunit alpha 0.249 0.246 DNA (cytosine-5)-methyltransferase 1 0.243 DNA gyrase subunit A 0.235 DNA-directed RNA polymerase subunit beta' 0.230 elongation factor G 0.228 0.222 0.220 phosphoribosylformylglycinamidine synthase 0.217 0.216 0.215 DNA-directed RNA polymerase subunit beta ferrous iron transport protein B iron complex transport system permease protein lacZ; beta-galactosidase ribonucleoside-triphosphate reductase (formate) 98 0.215 DNA gyrase subunit B 0.211 alanyl-tRNA synthetase 0.207 0.206 0.206 peptide/nickel transport system substrate-binding type I restriction enzyme, S subunit type I restriction enzyme M protein protein 0.203 preprotein translocase subunit SecA rRNA pseudouridine1911/1915/1917 synthase 0.196 0.194 carbamoyl-phosphate synthase large subunit 0.193 alanine or glycine:cation symporter, 0.193 transcription-repair coupling factor (superfamily II helicase) isoleucyl-tRNA synthetase transketolase 0.193 0.193 DNA polymerase I 0.192 0.191 cell division protease FtsH 0.185 long-chain acyl-CoA synthetase 0.184 chromosome partitioning protein 0.183 DNA topoisomerase III 0.179 3-oxoacyl-[acyl-carrier protein] reductase 0.179 starch-binding outer membrane protein 0.177 P-type Ca2+ transporter type 2C 0.177 ATP-dependent Lon protease 0.176 putative protease 0.173 LacI family transcriptional regulator 0.172 DNA ligase (NAD+) 0.171 leucyl-tRNA synthetase 0.171 Xaa-Pro aminopeptidase Gene open-reading frames were inferred from metagenome assemblies and were functionally annotated against the KEGG database using the GHOSTKOALA online platform. Shown are the mean relative abundances of the 45 most abundant KEGG orthologs (out of 6,270) across samples (N=32). Table S2.7 (cont’d) K02470 K01872 K01154 K03427 K02035 K03070 K06180 K01955 K03310 K03723 K01870 K02335 K00615 K03798 K01897 K03496 K03169 K00059 K21572 K01537 K01338 K08303 K02529 K01972 K01869 K01262 99 APPENDIX B: SUPPLEMENTAL FIGURES Figure S2.1 Relative abundances of ASVs from DNA extraction kit control samples. Heatmap showing ASVs with average proportional abundances >0.007 (i.e. 0.7%) in control samples. Darker colors in the heatmap indicate higher proportional abundances. ASVs boxed in red were removed from the dataset after being identified as contaminants by the R decontam package and our additional criteria (e.g. must have been present in at least 50% of control samples at proportional abundances >0.01). Proportional Abundance 0 0.1 0.2 0.3 0.4 ASV ASV1797|Delftia ASV276|Micrococcaceae ASV1979|Strenotrophomonas ASV1412|Planococcaceae Figure S2.2 Rarefaction curves of gut microbiota ASV richness. Plotted are the number of ASVs (ASV Richness) that are recovered given the number of sequences, after subsampling to 2,900 sequences/sample. Each curve represents a unique sample and are color-coded by hyena identity and ordered by matriline (1-highest ranking, 4-lowest ranking) and generation (M-mother, D-daughter, G- granddaughter). The y=x line shows what a line would look if with each read came a new ASV. s V S A f o r e b m u N 1500 1000 500 0 0 500 M1 D1 G1 M2 D2 G2 M3 D3 G3 M4 D4 G4 Y=X 2000 2500 3000 1000 1500 Number of Reads 100 Figure S2.3 Predominant bacterial families of the hyena gut microbiome. Stacked bar plots showing the relative frequency of 16S rRNA gene sequences assigned to each bacterial family across samples (N=301). Samples ordered by year, and each color represents a bacterial genus. Only bacterial families with average relative abundances >1% are plotted, and all others are summarized by “Other.” Bacterial Family Bacteroidaceae Carnobacteriaceae Clostridiaceae_1 Clostridiales_unclass Clostridiales_XI Clostridiales_XIII Coriobacteriales_unclass Eggerthellaceae Enterococcaceae Erysipelotrichaceae Fusobacteriaceae Lachnospiraceae Micrococcaceae Other Peptostreptococcaceae Planococcaceae Prevotellaceae Ruminococcaceae Streptococcaceae 1993 2000 2005 2010 2015 Samples (N=303) ) % ( e c n a d n u b A e v i t a l e R 100 75 50 25 0 101 Figure S2.4 The relative abundances of bacterial ASVs vary temporally. The abundances of these bacterial ASVs varied with sample year (α=0.05) as determined by linear mixed models (see Table S3 for values). Only ASVs with >1% relative abundance across samples were tested. Shaded line represents a best fit lm model line. ) ASV469|Bacteroides % ) ASV1280|Lachnospiraceae % ) ASV1507|Peptoclostridium % ) ASV1710|Fusobacterium % ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ●● ●●● ● ●● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 1993 1997 2001 2005 2009 2013 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ●● ● ●● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 1993 1997 2001 2005 2009 2013 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 1993 1997 2001 2005 2009 2013 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ( e c n a d n u b A e v i t a l e R V S A ● ● ● ● ● ● ● 40 20 0 ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ●●● ● ● ● ●● ● ●● ● ● ●● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● 1993 1997 2001 2005 2009 2013 ) ASV272|Micrococcaceae % ) ASV1499|Paeniclostridium % ) ASV1512|Peptoniphilus % ) ASV852|Clostridiales_unclass % ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ●● ●● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 1993 1997 2001 2005 2009 2013 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 1993 1997 2001 2005 2009 2013 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 1993 1997 2001 2005 2009 2013 ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ( e c n a d n u b A e v i t a l e R V S A 75 50 25 0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● 1993 1997 2001 2005 2009 2013 ) ASV1413|Planococcaceae % ) ASV1414|Planococcaceae % ) ASV913|Clostridium_ss7 % ) ASV936|Enterococcus % 60 40 20 0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● 1993 1997 2001 2005 2009 2013 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ●●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ●● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● 1993 1997 2001 2005 2009 2013 ( e c n a d n u b A e v i t a l e R V S A 15 10 5 0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 40 30 20 10 0 ( e c n a d n u b A e v i t a l e R V S A ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● 1993 1997 2001 2005 2009 2013 1993 1997 2001 2005 2009 2013 15 10 5 0 ( e c n a d n u b A e v i t a l e R V S A 60 40 20 0 ( e c n a d n u b A e v i t a l e R V S A ( e c n a d n u b A e v i t a l e R V S A 15 ( e c n a d n u b A e v i t a l e R V S A 10 5 0 ( e c n a d n u b A e v i t a l e R V S A 40 30 20 10 0 60 ( e c n a d n u b A e v i t a l e R V S A 40 20 0 ( e c n a d n u b A e v i t a l e R V S A 20 10 0 40 30 20 10 0 ( e c n a d n u b A e v i t a l e R V S A ) ASV1176|Clostridiaceae_1 % 20 15 10 5 0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● 1993 1997 2001 2005 2009 2013 ( e c n a d n u b A e v i t a l e R V S A 102 Figure S2.5 Gut microbiome similarity varies with year along PCoA axis 1. Grey dots represent samples, and the y-axis are gut microbiota similarity values estimated from Bray-Curtis distances. The values are from the first axis of a Principal Coordinate Analysis. Output of a generalized linear model regressing values from Axis 1 and sample year is shown below. i 1 s x A A o C P 0.4 0.2 0.0 -0.2 1993 1997 2001 2005 2009 2013 GLM β=0.02 ± 0.001, p<0.001 103 REFERENCES 104 REFERENCES 1. Trosvik P, Muinck EJ De, Rueness EK, Fashing PJ, Beierschmitt EC, Callingham KR, et al. Multilevel social structure and diet shape the gut microbiota of the gelada monkey , the only grazing primate. Microbiome. 2018;6:1–18. 2. Zhu L, Wu Q, Deng C, Zhang M, Zhang C, Chen H, et al. Adaptive evolution to a high purine and fat diet of carnivorans revealed by gut microbiomes and host genomes. Environ Microbiol. 2018;20:1711–22. 3. Daniel H, Gholami AM, Berry D, Desmarchelier C, Hahne H, Loh G, et al. High-fat diet alters gut microbiota physiology in mice. ISME J. 2014;8:295–308. doi:10.1038/ismej.2013.155. 4. Miyazaki T, Nishimura T, Yamashita T, Miyazaki M. Olfactory discrimination of anal sac secretions in the domestic cat and the chemical profiles of the volatile compounds. J Ethol. 2018;36:99–105. 5. Theis KR, Venkataraman A, Dycus JA, Koonter KD, Schmitt-Matzen EN, Wagner AP, et al. Symbiotic bacteria appear to mediate hyena social odors. PNAS. 2013;110:19832–19837. doi:10.1073/pnas.1306477110. 6. Gaona O, Cerqueda-García D, Moya A, Neri-Barrios X, Falcón LI. Geographical separation and physiology drive differentiation of microbial communities of two discrete populations of the bat Leptonycteris yerbabuenae. Microbiol Open. 2020;00:1–15. doi:10.1002/mbo3.1022. 7. Langille MGI, Meehan CJ, Koenig JE, Dhanani AS, Rose RA, Howlett SE, et al. Microbial shifts in the aging mouse gut. Microbiome. 2014;2:1–12. http://www.microbiomejournal.com/content/2/1/50. 8. Sabey K, Song S, Jolles A, Knight R, Ezenwa V. Coinfection and infection duration shape how pathogens affect the African buffalo gut microbiota. ISME J. 2020. 9. Baniel A, Amato KR, Beehner JC, Bergman TJ, Mercer A, Perlman RF, et al. Seasonal shifts in the gut microbiome indicate plastic responses to diet in wild geladas. Microbiome. 2020;:1– 20. 10. Perofsky AC, Lewis RJ, Abondano LA, Di Fiore A, Meyers LA. Hierarchical social networks shape gut microbial composition in wild Verreaux’s sifaka. Proceedings Biol Sci. 2017;284. 11. Kolodny O, Schulenburg H. Microbiome-mediated plasticity directs host evolution along several distinct time-scales. Philos Trans R Soc B2. 2020;375:20190589. 12. Gomez A, Sharma AK, Mallott EK, Petrzelkova KJ, Jost Robinson CA, Yeoman CJ, et al. Plasticity in the Human Gut Microbiome Defies Evolutionary Constraints. mSphere. 2019;4:1– 14. 13. Deng F, Mcclure M, Rorie R, Wang X, Chai J, Wei X. The vaginal and fecal microbiomes are related to pregnancy status in beef heifers. J Anim Sci Biotechnol. 2019;10:1–13. 14. Clooney AG, Eckenberger J, Laserna-Mendieta E, Sexton KA, Bernstein MT, Vagianos K, et al. Ranking microbiome variance in inflammatory bowel disease: a large longitudinal intercontinental study. Gut. 2021;70:499–510. doi:10.1136/gutjnl-2020-321106. 105 15. Zhang D, Hu Q, Liu X, Liu X, Gao F, Liang Y, et al. A longitudinal study reveals the alterations of the Microtus fortis colonic microbiota during the natural resistance to Schistosoma japonicum infection. Exp Parasitol. 2020;219. 16. Zhuang Y, Chai J, Cui K, Bi Y, Diao Q, Huang W. Longitudinal Investigation of the Gut Microbiota in Goat Kids from Birth to Postweaning. microorga. 2020;8:1–18. 17. Ren T, Grieneisen LE, Alberts SC, Archie EA, Wu M. Development, diet and dynamism: longitudinal and cross-sectional predictors of gut microbial communities in wild baboons. Environ Microbiol. 2016;18:1312–25. 18. Hollister EB, Riehle K, Luna RA, Weidler EM, Rubio-Gonzales M, Mistretta TA, et al. Structure and function of the healthy pre-adolescent pediatric gut microbiome. Microbiome. 2015;3. doi:10.1186/s40168-015-0101-x. 19. Huttenhower C, Gevers D, Knight R, Abubucker S, Badger JH, Chinwalla AT, et al. Structure, function and diversity of the healthy human microbiome. Nature. 2012;486:207–14. doi:10.1038/nature11234. 20. Louca S, Polz MF, Mazel F, Albright MBN, Huber JA, O’Connor MI, et al. Function and functional redundancy in microbial systems. Nat Ecol Evol. 2018;2:936–43. doi:10.1038/s41559- 018-0519-1. 21. Hird SM. Evolutionary Biology Needs Wild Microbiomes. Front Microbiol. 2017;8. doi:10.3389/fmicb.2017.00725. 22. Suzuki TA. Links between Natural Variation in the Microbiome and Host Fitness in Wild Mammals. Integr Comp Biol. 2017;57:1–14. doi:10.1093/icb/icx104. 23. Moeller AH, Sanders JG. Roles of the gut microbiota in the adaptive evolution of mammalian species. Philos Trans R Soc B2. 2020;375:20190597. 24. Ma JE, Jiang HY, Li LM, Zhang XJ, Li GY, Li HM, et al. The fecal metagenomics of Malayan pangolins identifies an extensive adaptation to myrmecophagy. Front Microbiol. 2018;9:1–14. 25. Kohl KD, Oakeson KF, Orr TJ, Miller AW, Sorensen J, Phillips CD, et al. Metagenomic sequencing provides insights into the location of microbial detoxification in the gut of a small mammalian herbivore. bioRxiv. 2018;:1–3. 26. Louis P, Flint HJ. Formation of propionate and butyrate by the human colonic microbiota. Environ Microbiol. 2017;19:29–41. 27. LeBlanc JG, Chain F, Martín R, Bermúdez-Humarán LG, Courau S, Langella P. Beneficial effects on host energy metabolism of short-chain fatty acids and vitamins produced by commensal and probiotic bacteria. Microb Cell Fact. 2017;16:1–10. 28. Frank LG. Social organization of the spotted hyaena Crocuta crocuta. II. Dominance and reproduction. Anim Behav. 1986;34:1510–27. 29. Kruuk H. The spotted hyena : a study of predation and social behavior. Chicago: University of Chicago Press; 1972. 106 30. Strauss ED, Holekamp KE. Inferring longitudinal hierarchies: Framework and methods for studying the dynamics of dominance. J Anim Ecol. 2019;88:521–36. doi:10.1111/1365- 2656.12951. 31. Frank LG. Social organization of the spotted hyaena (Crocuta crocuta). I. Demography. Anim Behav. 1986;34:1500–9. 32. Holekamp KE, Smith JE, Strelioff CC, Van Horn RC, Watts HE. Society, demography and genetic structure in the spotted hyena. Mol Ecol. 2012;21:613–32. 33. Smith JE, Kolowski JM, Graham KE, Dawes SE, Holekamp KE. Social and ecological determinants of fission-fusion dynamics in the spotted hyaena. Anim Behav. 2008;76:619–36. 34. Cooper SM, Holekamp KE, Smale L. A seasonal feast: long-term analysis of feeding behaviour in the spotted hyaena ( Crocuta crocuta ). Afr J Ecol. 1999;37:149–60. doi:10.1046/j.1365-2028.1999.00161.x. 35. Blumstein DT, Rangchi TN, Briggs T, De Andrade FS, Natterson-Horowitz B. A Systematic Review of Carrion Eaters’ Adaptations to Avoid Sickness. J Wildl Dis. 2017;53:577–81. doi:10.7589/2016-07-162. 36. Green DS, Zipkin EF, Incorvaia DC, Holekamp KE. Long-term ecological changes influence herbivore diversity and abundance inside a protected area in the Mara-Serengeti ecosystem. Glob Ecol Conserv. 2019;20:e00697. doi:10.1016/j.gecco.2019.e00697. 37. Ogutu JO, Piepho HP, Dublin HT, Bhola N, Reid RS. Rainfall influences on ungulate population abundance in the Mara-Serengeti ecosystem. J Anim Ecol. 2008;77:814–29. 38. Holekamp KE, Smale L, Berg R, Cooper SM. Hunting rates and hunting success in the spotted hyena (Crocuta crocuta). J Zool. 1997;242:1–15. 39. Green DS. ANTHROPOGENIC DISTURBANCE , ECOLOGICAL CHANGE , AND WILDLIFE CONSERVATION AT THE EDGE OF THE MARA-SERENGETI ECOSYSTEM By Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Zoology – Doctor of Philosophy Eco. 2015. 40. Green DS, Johnson-Ulrich L, Couraud HE, Holekamp KE. Anthropogenic disturbance induces opposing population trends in spotted hyenas and African lions. Biodivers Conserv. 2018;27:871–89. 41. Rosenberg E, Zilber-rosenberg I. The hologenome concept of evolution after 10 years. Microbiome. 2018;6:1–14. 42. Bell RHV. A Grazing Ecosystem in the Serengeti. Sci Am. 1971;225:86–93. 43. Maddock L. The “Migration” and Grazing Succession. In: Sinclair ARE, Norton-Griffiths M, editors. Serengeti: Dynamics of an Ecosystem. Chicago: The University of Chicago Press; 1979. p. 104–29. 44. Riggio J, Jacobson A, Dollar L, Bauer H, Becker M, Dickman A, et al. The size of savannah Africa: A lion’s (Panthera leo) view. Biodivers Conserv. 2013;22:17–35. doi:10.1007/s10531- 107 012-0381-4. 45. Spagnuolo OSB, Jarvey JC, Battaglia MJ, Laubach ZM, Miller ME, Holekamp KE, et al. Mapping Kenyan Grassland Heights Across Large Spatial Scales with Combined Optical and Radar Satellite Imagery. Remote Sens. 2020;12. 46. Western D, Russell S, Cuthill I. The Status of Wildlife in Protected Areas Compared to Non- Protected Areas of Kenya. PLoS One. 2009;4:e6140. doi:10.1371/journal.pone.0006140. 47. Drea CM, Frank LG. The social complexity of spotted hyenas. In: Waal FBM de, Tyack PL, editors. Animal Social Complexity: Intelligence, Culture, and Individualized Societies. Harvard University Press; 2003. p. 121–48. 48. Holekamp KE, Sakai ST, Lundrigan BL. Social intelligence in the spotted hyena (Crocuta crocuta). Phil Trans R Soc B. 2007;362:523–38. doi:10.1098/rstb.2006.1993. 49. Frank LG, Holekamp KE SL. Dominance, demography, and reproductive success of female spotted hyenas. In: Sinclair ARE AP, editor. Serengeti II : dynamics, management, and conservation of an ecosystem. Chicago: University of Chicago Press; 1995. p. 364–84. https://ostromworkshop.indiana.edu/library/node/57709. Accessed 28 Aug 2019. 50. Holekamp KE, Smale L. Ontogeny of dominance in free-living spotted hyaenas: juvenile rank relations with other immature individuals. Anim Behav. 1993;46:451–66. 51. Holekamp KE, Smale L. Dominance Acquisition during Mammalian Social Development: The "Inheritance" of Maternal Rank. Am Zool. 1991;31:306–17. doi:10.2307/3883408. 52. Engh AL, Funk SM, Horn RC Van, Scribner KT, Bruford MW, Libants S, et al. Reproductive skew among males in a female- dominated mammalian society. Behav Ecol. 2002;13. 53. Höner OP, Wachter B, Hofer H, Wilhelm K, Thierer D, Trillmich F, et al. The fitness of dispersing spotted hyaena sons is influenced by maternal social status. Nat Commun. 2010;1:1–7. doi:10.1038/ncomms1059. 54. Höner OP, Wachter B, East ML, Streich WJ, Wilhelm K, Burke T, et al. Female mate-choice drives the evolution of male-biased dispersal in a social mammal. Nature. 2007;448:798–801. doi:10.1038/nature06040. 55. Smith JE, Kolowski JM, Graham KE, Dawes SE, Holekamp KE. Social and ecological determinants of fission–fusion dynamics in the spotted hyaena. Anim Behav. 2008;76:619–36. doi:10.1016/J.ANBEHAV.2008.05.001. 56. Hofer H, East M. Population dynamics, population size, and the commuting system of Serengeti spotted hyaenas. In: Sinclair ARE, Arcese P, editors. Serengeti II : dynamics, management, and conservation of an ecosystem. Chicago: University of Chicago Press; 1995. p. 665. https://ostromworkshop.indiana.edu/library/node/59434. Accessed 28 Aug 2019. 57. Frank LG, Glickman SE, Powch I. Sexual dimorphism in the spotted hyaena (Crocuta crocuta). J Zool. 1990;221:308–13. doi:10.1111/j.1469-7998.1990.tb04001.x. 108 58. Holekamp KE, Smale L, Szykman M. Rank and reproduction in the female spotted hyaena. J Reprod Fertil. 1996;108:229–37. 59. Holekamp KE, Szykman M, Boydston EE, Smale L. Association of seasonal reproductive patterns with changing food availability in an equatorial carnivore, the spotted hyaena (Crocuta crocuta). J Reprod Fertil. 1999;116:87–93. 60. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, et al. Ultra-high- throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 2012;6:1621–4. doi:10.1038/ismej.2012.8. 61. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. Development of a Dual- Index Sequencing Strategy and Curation Pipeline for Analyzing Amplicon Sequence Data on the MiSeq Illumina Sequencing Platform. Appl Environ Microbiol. 2013;79:5112–20. doi:10.1128/AEM.01043-13. 62. R Core Team. R: a language and environment for statistical computing. In: R Foundation for Statistical Computing. Vienna, Austria; 2019. https://www.gbif.org/tool/81287/r-a-language-and- environment-for-statistical-computing. Accessed 28 Aug 2019. 63. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: High- resolution sample inference from Illumina amplicon data. Nat Methods. 2016;13:581–3. 64. Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 2013;41:D590-6. doi:10.1093/nar/gks1219. 65. Davis NM, Proctor DiM, Holmes SP, Relman DA, Callahan BJ. Simple statistical identification and removal of contaminant sequences in marker-gene and metagenomics data. Microbiome. 2018;6:226. doi:10.1186/s40168-018-0605-2. 66. Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol. 2009;75:7537–41. doi:10.1128/AEM.01541-09. 67. Wickham H. ggplot2: Elegant Graphics for Dat Analysis. New York, NY: Springer-Verlag New York; 2009. doi:10.1007/978-0-387-98141-3. 68. Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw. 2015;67:1–48. doi:10.18637/jss.v067.i01. 69. Oksanen J, Blanchet F, Friendly M, Kindt R, Legendre P, McGlinn D, et al. vegan: Community Ecology Package. R Packag version 24-6. 2018. https://cran.r- project.org/web/packages/vegan/index.html. 70. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30:2114–20. doi:10.1093/bioinformatics/btu170. 71. Kim D, Paggi JM, Park C, Bennett C, Salzberg SL. Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype. Nat Biotechnol. 2019;37:907–15. 109 doi:10.1038/s41587-019-0201-4. 72. Zhang J, Kobert K, Flouri T, Stamatakis A. PEAR: A fast and accurate Illumina Paired-End reAd mergeR. Bioinformatics. 2014;30:614–20. 73. Segata N, Waldron L, Ballarini A, Narasimhan V, Jousson O, Huttenhower C. Metagenomic microbial community profiling using unique clade-specific marker genes. Nat Methods. 2012;9:811–4. doi:10.1038/nmeth.2066. 74. Li D, Liu CM, Luo R, Sadakane K, Lam TW. MEGAHIT: An ultra-fast single-node solution for large and complex metagenomics assembly via succinct de Bruijn graph. Bioinformatics. 2015;31:1674–6. 75. Gurevich A, Saveliev V, Vyahhi N, Tesler G. QUAST: quality assessment tool for genome assemblies. Bioinformatics. 2013;29:1072–5. doi:10.1093/bioinformatics/btt086. 76. Zhu W, Lomsadze A, Borodovsky M. Ab initio gene identification in metagenomic sequences. Nucleic Acids Res. 2010;38:e132. doi:10.1093/nar/gkq275. 77. Fu L, Niu B, Zhu Z, Wu S, Li W. CD-HIT: Accelerated for clustering the next-generation sequencing data. Bioinformatics. 2012;28:3150–2. 78. Kanehisa M, Sato Y, Morishima K. BlastKOALA and GhostKOALA: KEGG Tools for Functional Characterization of Genome and Metagenome Sequences. J Mol Biol. 2016;428:726–31. doi:10.1016/j.jmb.2015.11.006. 79. Kanehisa M, Sato Y, Kawashima M, Furumichi M, Tanabe M. KEGG as a reference resource for gene and protein annotation. Nucleic Acids Res. 2016;44:D457–62. 80. Kim S. ppcor: An R Package for a Fast Calculation to Semi-partial Correlation Coefficients. Commun Stat Appl Methods. 2015;22:665–74. 81. Sikes RS, Gannon WL. Guidelines of the American Society of Mammalogists for the use of wild mammals in research. J Mammal. 2011;92:235–53. doi:10.1644/10-MAMM-F-355.1. 82. Bendová B, Piálek J, Ď Ľ, Schmiedová L, Dagmar Č, Martin J, et al. How being synanthropic affects the gut bacteriome and mycobiome : comparison of two mouse species with contrasting ecologies. BMC Microbiol. 2020;20:1–13. 83. Barelli C, Albanese D, Stumpf RM, Asangba AE, Donati C, Rovero F, et al. The Gut Microbiota Communities of Wild Arboreal and Ground-Feeding Tropical Primates Are Affected Differently by Habitat Disturbance. mSystems. 2020;5:1–18. 84. Watson SE, Hauffe HC, Bull M, McKinney MA, Atwood TA, Perkins SE. Global change- driven use of onshore habitat impacts polar bear faecal microbiota. ISME J. 2019. doi:10.1038/s41396-019-0480-2. 85. Sugden S, Sanderson D, Ford K, Stein LY, St. Clair CC. An altered microbiome in urban coyotes mediates relationships between anthropogenic diet and poor health. Sci Rep. 2020;10:1–14. doi:10.1038/s41598-020-78891-1. 110 variability in the gut microbiome of a highly - specialized and critically endangered lemur species 86. Donohue ME, Weisrock DW, Asangba AE, Ralainirina J, Stumpf RM, Wright PC. Extensive across sites. Am J Primatol. 2019;e23046:1–12. 87. Richardson JB, Dancy BCR, Horton CL, Lee YS, Madejczyk MS, Xu ZZ, et al. Exposure to toxic metals triggers unique responses from the rat gut microbiota. Sci Rep. 2018;8:1–12. doi:10.1038/s41598-018-24931-w. 88. Rosenberg E, Delong EF, Thompson F. The Prokaryotes-other Major Lineages. 89. Wexler HM. Bacteroides: The good, the bad, and the nitty-gritty. Clinical Microbiology Reviews. 2007;20:593–621. doi:10.1128/CMR.00008-07. 90. Sugden S, Sanderson D, Ford K, Stein LY, St. Clair CC. An altered microbiome in urban coyotes mediates relationships between anthropogenic diet and poor health. Sci Rep. 2020;10:1–14. doi:10.1038/s41598-020-78891-1. 91. Li S, Wang Z, Yang Y, Yang S, Yao C, Liu K, et al. Lachnospiraceae shift in the microbial community of mice faecal sample effects on water immersion restraint stress. AMB Express. 2017;7:1–11. doi:10.1186/s13568-017-0383-4. 92. Devriese L, Baele M, Butaye P. The Genus Enterococcus. In: The Prokaryotes. Springer US; 2006. p. 163–74. doi:10.1007/0-387-30744-3_5. 93. Costello E, Stagaman K, Dethlefsen L, Bohannan B, Relman D. The Application of Ecological Theory Toward an Understanding of the Human Microbiome. Science (80- ). 2012;336:1255–62. http://science.sciencemag.org/content/sci/336/6086/1255.full.pdf. Accessed 28 Jul 2017. 94. Seth EC, Taga ME. Nutrient cross-feeding in the microbial world. Front Microbiol. 2014;5:1– 6. 95. Rothschild D, Weissbrod O, Barkan E, Kurilshikov A, Korem T, Zeevi D, et al. Environment dominates over host genetics in shaping human gut microbiota. Nature. 2018;555:210. https://doi.org/10.1038/nature25973. 96. Ren T, Boutin S, Humphries MM, Dantzer B, Gorrell JC, Coltman DW, et al. Seasonal, spatial, and maternal effects on gut microbiome in wild red squirrels. Microbiome. 2017;5:163. doi:10.1186/s40168-017-0382-3. 97. Rojas CA, Holekamp KE, Winters AD, Theis KR. Body-site specific microbiota reflect sex and age-class among wild spotted hyenas. FEMS Microbiol Ecol. 2020; January:1–14. 98. Heitlinger E, Ferreira SCM, Thierer D, Hofer H, East ML. The Intestinal Eukaryotic and Bacterial Biome of Spotted Hyenas: The Impact of Social Status and Age on Diversity and Composition. Front Cell Infect Microbiol. 2017;7. 99. Mizukami K, Uchiyama J, Igarashi H, Murakami H, Shima A, Ishiahra G, et al. Age-related analysis of the gut microbiome in a purebred dog colony. FEMS Microbiol Lett. 2019;366. 100. Reveles KR, Patel S, Forney L, Ross CN. Age-related changes in the marmoset gut 111 microbiome. Am J Primatol. 2019;81. 101. van Leeuwen P, Mykytczuk N, Mastromonaco GF, Schulte-Hostedde AI. Effects of captivity, diet, and relocation on the gut bacterial communities of white-footed mice. Ecol Evol. 2020;10:4677–90. 102. Huang G, Wang X, Hu Y, Wu Q, Nie Y, Dong J. Diet drives convergent evolution of gut microbiomes in bamboo-eating species. Sci China Life Sci. 2020;64. 103. Ley RE, Hamady M, Lozupone C, Turnbaugh PJ, Ramey RR, Bircher JS, et al. Evolution of Mammals and Their Gut Microbes. Science (80- ). 2008;320:1647. 104. Gillman SJ, McKenney EA, Lafferty DJR. Wild black bears harbor simple gut microbial communities with little difference between the jejunum and colon. Sci Rep. 2020;10:1–11. doi:10.1038/s41598-020-77282-w. 105. Menke S, Meier M, Mfune JKE, Melzheimer J, Wachter B, Sommer S. Effects of host traits and land-use changes on the gut microbiota of the Namibian black-backed jackal (Canis mesomelas). FEMS Microbiol Ecol. 2017;93:1–16. 106. Pereira AC, Bandeira V, Fonseca C, Cunha M V. Egyptian Mongoose Gut Microbiota : Taxonomical and Functional Differences across Sex and Age Classes. 2020; February:1–18. 107. Sommer F, Stå M, Ilkayeva O, Newgard CB, Frö O, BäCkhed Correspondence F. The Gut Microbiota Modulates Energy Metabolism in the Hibernating Brown Bear Ursus arctos. Cell Rep. 2016;14:1655–61. 108. Cheng Y, Fox S, Pemberton D, Hogg C, Papenfuss AT, Belov K. The Tasmanian devil microbiome-implications for conservation and management. Microbiome. 2015;3:76. doi:10.1186/s40168-015-0143-0. 109. He F, Liu D, Zhang L, Zhai J, Ma Y, Xu Y, et al. Metagenomic analysis of captive Amur tiger faecal microbiome. BMC Vet Res. 2018;14:1–8. 110. Handl S, German AJ, Holden SL, Dowd SE, Steiner JM, Heilmann RM, et al. Faecal microbiota in lean and obese dogs. FEMS Microbiol Ecol. 2013;84:332–43. 111. Youngblut ND, Reischer GH, Walters W, Schuster N, Walzer C, Stalder G, et al. Host diet and evolutionary history explain different aspects of gut microbiome diversity among vertebrate clades. Nat Commun. 2019;10:1–15. doi:10.1038/s41467-019-10191-3. 112. Li J, Zhan S, Liu X, Lin Q, Jiang J, Li X. Divergence of Fecal Microbiota and Their Associations With Host Phylogeny in Cervinae. Front Microbiol. 2018;9:1–11. doi:10.3389/fmicb.2018.01823. 113. Sanders JG, Beichman AC, Roman J, Scott JJ, Emerson D, Mccarthy JJ, et al. Baleen whales host a unique gut microbiome with similarities to both carnivores and herbivores. Nat Commun. 2015;6:8286. doi:10.1038/ncomms9285. 114. Brooks AW, Kohl KD, Brucker RM, Van Opstal EJ, Bordenstein SR. Phylosymbiosis: Relationships and Functional Effects of Microbial Communities across Host Evolutionary 112 History. PLoS Biol. 2016;14:1–19. doi:10.1371/journal.pbio.2000225. 115. Liu G, Bou G, Su S, Xing J, Qu H, Zhang X, et al. Microbial diversity within the digestive tract contents of Dezhou donkeys. PLoS One. 2019;14:1–15. doi:10.1371/journal.pone.0226186. 116. Chung O, Jin S, Cho YS, Lim J, Kim H, Jho S, et al. The first whole genome and transcriptome of the cinereous vulture reveals adaptation in the gastric and immune defense systems and possible convergent evolution between the Old and New World vultures. Genome Biol. 2015;16:1–11. doi:10.1186/s13059-015-0780-4. 117. Rosenberg E, Delong EF, Lory S, Stackebrandt E, Thompson F. The Prokaryotes- Firmicutes Tenericutes. 4th edition. Springer-Verlag; 2014. 118. Nicholson WL. Roles of Bacillus endospores in the environment. Cellular and Molecular Life Sciences. 2002;59:410–6. doi:10.1007/s00018-002-8433-7. 119. Zhao X, Kuipers OP. Identification and classification of known and putative antimicrobial compounds produced by a wide variety of Bacillales species. BMC Genomics. 2016;17:1–18. doi:10.1186/s12864-016-3224-y. 120. Finlayson-Trick ECL, Getz LJ, Slaine PD, Thornbury M, Lamoureux E, Cook J, et al. Taxonomic differences of gut microbiomes drive cellulolytic enzymatic potential within hind-gut fermenting mammals. PLoS One. 2017;12:1–22. 121. Estaki M, Pither J, Baumeister P, Little JP, Gill SK, Ghosh S, et al. Cardiorespiratory fitness as a predictor of intestinal microbial diversity and distinct metagenomic functions. Microbiome. 2016;4. doi:10.1186/s40168-016-0189-7. 122. Koliada A, Syzenko G, Moseiko V, Budovska L, Puchkov K, Perederiy V, et al. Association between body mass index and Firmicutes/Bacteroidetes ratio in an adult Ukrainian population. BMC Microbiol. 2017;17:120. doi:10.1186/s12866-017-1027-1. 123. Moeller AH, Suzuki TA, Lin D, Lacey EA, Wasser SK, Nachman MW. Dispersal limitation promotes the diversification of the mammalian gut microbiota. PNAS. 2017;114:13768–73. doi:10.1073/pnas.1700122114. 124. Rojas CA, Ramirez-Barahona S, Holekamp KE, Theis KR. Host phylogeny and host ecology structure the mammalian gut microbiota at different taxonomic scales. bioRxiv. 2020;:2020.09.23.299727. doi:10.1101/2020.09.23.299727. 125. Roggenbuck M, Bærholm Schnell I, Blom N, Bælum J, Bertelsen MF, Pontén TS, et al. The microbiome of New World vultures. Nat Commun. 2014;5:1–8. 126. Becker AAMJ, Hill KC, Butaye P. Unraveling the Gut Microbiome of the Invasive Small Indian Mongoose ( Urva auropunctata ) in the Caribbean. Microorganisms. 2021;9:1–14. 127. Milani C, Alessandri G, Mancabelli L, Mangifesta M, Lugli GA, Viappiani A, et al. Multi- omics Approaches To Decipher the Impact of Diet and Host Physiology on the Mammalian Gut Microbiome. Appl Environ Microbiol. 2020;86:1–21. 113 128. Tuson HH, Foley MH, Koropatkin NM, Biteen JS. The Starch Utilization System Assembles around Stationary Starch-Binding Proteins. Biophysj. 2018;115:242–50. doi:10.1016/j.bpj.2017.12.015. 129. Crippen T, Benbow M, Pechal J. Microbial Interactions during Carrion Decomposition. In: Benbow E, Tomberlin J, Taronem A, editors. Carrion Ecology, Evolution, and Their Applications. Boca Raton, FL: CRC Press; 2016. p. 31–64. 130. Hyde ER, Haarmann DP, Lynne AM, Bucheli SR, Petrosino JF. The Living Dead: Bacterial Community Structure of a Cadaver at the Onset and End of the Bloat Stage of Decomposition. PLoS One. 2013;8:e77733. doi:10.1371/journal.pone.0077733. 131. Harrison L, Kooienga E, Speights C, Tomberlin J, Lashley M, Barton B, et al. Microbial succession from a subsequent secondary death event following mass mortality. 2020;:1–11. 132. Levin D, Raab N, Pinto Y, Rothschild D, Zanir G, Godneva A, et al. Diversity and functional landscapes in the microbiota of animals in the wild. Science (80- ). 2021;372. doi:10.1126/science.abb5352. 133. Services A, Search A, Overview A, Versions A, Article E, Links RI, et al. Comparative Study of Gut Microbiota in Wild and Captive Giant Pandas. Genes (Basel). 2019;10:1–22. 134. Prakash T, Taylor TD. Functional assignment of metagenomic data: Challenges and applications. Brief Bioinform. 2012;13:711–27. doi:10.1093/bib/bbs033. 114 ABSTRACT DO HOST SOCIAL INTERACTIONS PREDICT GUT MICROBIOTA SIMILARITY AND DIVERSITY IN A FISSION-FUSION MAMMALIAN SOCIETY? CHAPTER THREE: Connie A. Rojas, Julie W. Turner, Kevin. R. Theis, and Kay E. Holekamp Group living facilitates the spread of pathogens and disease but may also promote the sharing of beneficial microbes among group-mates and the establishment of gut bacterial communities (termed the gut ‘microbiota’). Research primarily conducted in primates demonstrates that social relationships may impact gut microbiota structure, as individuals with stronger social bonds show more convergence in their gut microbiotas than individuals with weaker social ties. Nonetheless, little is known about the impact of social interactions on the gut microbiota of individuals that form less cohesive groups, such as those from fission-fusion societies. Here, we use 16S rRNA gene sequencing and social network analyses to determine whether host social interactions are associated with gut microbiota variation in a social group of wild spotted hyenas, which are gregarious large carnivores living in fission-fusion societies. Fecal samples were collected from adult female hyenas during several 6-month time-periods. We assess whether hyena dyads with stronger social ties have more similar gut microbiotas than dyads with weaker social ties. We also investigate whether individuals that are more socially connected have more diverse gut microbiotas than less connected individuals. Contrary to expectations, hyenas that interacted more frequently had less similar gut microbiotas than individuals that interacted less frequently. Furthermore, hyenas that were more socially connected harbored less rich gut microbiotas than less connected individuals. Due to fission-fusion dynamics, hyenas may have fewer social partners or less physical contact than primates living in cohesive societies. Additionally, our findings may be bacteria-specific, as some bacteria (non-spore 115 formers) may be more reliant on social interactions for their transmission than others. INTRODUCTION In social mammals, interactions among group-mates shape an individual’s fitness. Group-mates may forage or hunt together, defend one another against aggression from other group-mates, and cooperatively rear and provision offspring [1]. Social interactions and physical interactions shape hierarchical relationships among group-mates and help prevent the escalation of future physical altercations [2, 3]. Yet, group living mammals must also compete with one another, and social interactions can result in physical injury, infanticide, and eviction from the group [1]. It is also widely known that social interactions, particularly physical interactions can spread pathogens and disease among group-mates [4, 5]. However, social interactions can also facilitate the exchange and general acquisition of beneficial and commensal microbes [5, 6], as microbes can be acquired and transmitted among members via social contact (e.g. grooming, huddling, provisioning, mating) or via shared diets and environments. In primates [7–9], equids [10], and rodents [11], there is evidence that individuals with strong social bonds tend to harbor similar symbiotic gut bacterial communities, hereafter termed gut microbiotas. The gut microbiota also varies with host social group such that changes in social group membership, like those experienced by individuals at dispersal [12] or after group fission events [13] are often accompanied by changes in gut microbiota composition so that it converges with that of the host’s new group. Furthermore, social interactions may also be important in influencing gut microbiota diversity; in humans and wood mice, individuals that engage in higher levels of social interactions [14] or have denser social networks [11, 15] tend to have more diverse gut microbiotas. Cumulatively, this research suggests that the host’s social environment is an important predictor of the gut microbiota in gregarious mammals. However, the extent to which social interactions 116 influence the gut microbiota of individuals that form less cohesive groups, such as those from fission-fusion societies is unknown. Spotted hyenas (Crocuta crocuta) are highly social carnivores that offer an excellent model system for studying whether host social interactions are associated with variation in the gut microbiota of individuals from fission-fusion societies. Hyenas inhabit much of sub- Saharan Africa and live in large, mixed-sex groups that can contain over 90 individuals [16]. These groups are composed of overlapping generations of females and their offspring, along with immigrant males, and are characterized by low mean relatedness among clan members [17]. Hyena societies are matrilineal [15], and are structured by linear dominance hierarchies, in which an individual’s position in the hierarchy determines its priority of access to resources and its subsequent fitness [18–20]. Females exhibit dominance over males [16] and are philopatric, whereas males disperse to neighboring clans upon reaching reproductive maturity [18, 21, 22]. Nonetheless, as mentioned, hyenas live in fission-fusion societies, where individuals mainly travel, rest and forage alone or in small subgroups that change in composition throughout the day and may come together to form larger groups [18, 19, 21, 22]. In their daily interactions, hyenas exhibit a range of cooperative and aggressive behaviors, although here we focus on greeting ceremonies and aggressive encounters, as these behaviors may provide direct pathways for microbe transmission between dyads. During greeting ceremonies, two or more hyenas stand parallel to one another and reciprocally sniff one another’s anogenital region [23]. This occurs when individuals want to reconcile after aggressive interactions [24], although greetings also occur in contexts unrelated to aggression, and may function to maintain social bonds [23]. Aggression in this species occurs in several different contexts (e.g. competition over food, for defense of 117 offspring) and varies in intensity [18], from low-level threats like head waves to lunging attacks and chases to injurious biting. Prior research on this species suggests that the host’s social environment may be an important predictor of the gut microbiota. In adult females, host social rank predicts the diversity of gut eukaryotic communities [25], and social group membership correlates with scent gland microbiota composition [26]. Furthermore, a recent study demonstrated that the majority of the variation in the gut microbiota across two decades and multiple generations of hyenas cannot be accounted for by host predictors such as age, maternal relatedness, or prey abundance (Rojas, 2021). However, this variation may be potentially explained by host social interactions. Here, we couple behavioral observations of wild spotted hyenas in the Masai Mara National Reserve, Kenya (Table 3.1), with 16S rRNA gene sequencing and social network analyses to inquire whether host social interactions predict gut microbiota variation in adult females from a single social group. Specifically, we predict that 1) individuals who associate frequently and have stronger social bonds should have more similar gut microbiotas than individuals who rarely encounter each other, and 2) individuals that are more socially connected should have more diverse gut microbiotas than less connected individuals. Findings from this study have implications for identifying the host factors potentially structuring the gut microbiota at a group level in a gregarious large carnivore. The gut microbiota in other mammalian species affects host digestion [27–30], immune activity [31–34]. and behavior [35]. If this is also true among spotted hyenas, then our study should elucidate at least some of the mechanisms via which functionally-relevant, symbiotic microbes are transmitted in this species. 118 Year 2000 2000 2011 2013 2015 2001-2002 2005-2006 2014-2015 1 2 3 4 5 6 7 8 39 37 36 39 30 29 33 25 17 18 19 11 21 19 20 19 Table 3.1 Breakdown of the samples included in this study. Time period Number of fecal samples Number of hyena focal individuals Fecal samples were collected from adult female hyenas from a single social group during eight distinct time-periods, which each spanned 6 months. METHODS Study location and sampling Fecal samples were collected opportunistically from adult female wild spotted hyenas residing in the Masai Mara National Reserve (MMNR), Kenya between 2000-2006 and 2011-2015. The MMNR is a savanna ecosystem that boasts high densities of carnivores and herbivores and undergoes two rainy seasons [36–42]. The fecal samples (N=268) came from 81 unique adult females from a single clan, and were partitioned into eight categorical time-periods that were each six months long (Table 3.1, Table S3.1). These time-periods were created to minimize the temporal variation in the gut microbiota. In the field, hyenas are identified individually by their unique spot patterns, sexed based on the morphology of their erect phallus [43], and aged according to their appearance as cubs when first seen [20]. Individuals reach reproductive maturity at 24 months, but reproduction is often delayed [44, 45]. The individuals in our study ranged from 2 to 17 years of age. Social dominance ranks, which are relatively stable over an individual’s lifespan, were determined based on the outcomes of dyadic aggressive interactions as described by Strauss et al. [46]. 119 To collect behavioral and demographic data, research assistants monitored hyena study clans daily during morning (6 to 10 AM) and evening (4 to 8 PM) observation periods. An observation “session” was initiated when a subgroup of one or more hyenas was encountered, and all hyenas within 200 m of that subgroup were considered to be present in that session. During each session, all instances of aggressive and affiliative interactions among hyenas were recorded using all-occurrence sampling [47], and every 15–20-mins, the identity of all individuals present was noted in scan samples. Sessions ended when social interactions ceased because hyenas were out of sight or resting. Construction of social networks We constructed three types of social networks: association, affiliation and aggression networks, which were based on session co-occurrence, greetings, and aggressive interactions between hyena individuals, respectively. Networks were constructed using the R asnipe and spatsoc packages [48–50] for each individual hyena included in this study during each time-period, as documented previously [51, 52]. We used behavioral data from the year preceding and including fecal sample collection for each time-period to generate networks (Table S3.1). For example, the first time-period included fecal samples that were collected from 2-Jan-2000 to 11-Jun-2000, thus behavioral data from 11-Jun-1999 to 11-Jun-2000 was used to construct the social networks for hyena focal individuals (Table S3.1). The focal individual had to be seen at least 10 times during the 365-day observation period, and each of its partners also had to be seen at least ten times to be included in the network. We believed one year of behavioral data would be sufficient to encapsulate social interactions among group-mates. Prior microbiome studies done in social primates [8, 53, 54], including in baboons which have similar social structures as hyenas [8], also used the year preceding and including fecal sample collection to quantify host social interactions. 120 In each network, individual hyenas represented nodes and the strength of the relationship between dyads were edges. For association networks, the edges represented simple ratio association indices (SRIs; AIs) [55], which are robust indicators of social bond strength in this species. AIs were calculated as the number of sessions in which individuals A and B were seen together divided by the total number of sessions in which A and B were seen together plus the total number of sessions in which A was seen without B plus the total number of sessions in which B was seen without A [55]. For affiliation networks, we first calculated interaction rates as the number of greetings between the focal individual and its partners, divided by the duration of the behavioral observation period (i.e. 365 days). Then, we conducted a regression between AIs and interaction rates to control for individual variation in gregariousness, and used the residuals from the regression to represent the network edges [56, 57]. Similarly, aggression network edges were calculated as the residuals from the regression modelling the number of aggressive acts the focal individual initiated or received within each dyad, weighted by the intensity of the aggression (levels 1- 3) [58], divided by 365. Since residuals can be positive or negative but network ties need to be positive, we rescaled the residuals so that they were bounded by 0 and 1. Calculating hyena ‘sociability’ For analyses relating gut microbiota similarity to the strength of the relationship between hyena dyads, we used network edges as our independent variable. For analyses relating gut microbiota alpha-diversity with an individual’s degree of ‘sociability’, we used three metrics of social network position as our independent variables: degree, betweenness, and eigenvector centrality. An individual’s degree centrality was their number of connections, ‘betweenness’ centrality was the number of shortest paths between members of any dyad in the network that run through the focal individual, and eigenvector centrality examined the connectedness of the focal individual as well as the connectedness 121 of its connections [59]. These three metrics were chosen because they represented different dimensions of sociality and each may differentially affect microbial transmission: degree indicated which individuals are social hubs, betweenness indicated which individuals act as bridges between otherwise unconnected individuals, and eigenvector centrality measured an individual’s social capital [59]. We predicted that these three metrics of social network position would be positively correlated with gut microbiota alpha-diversity. DNA extractions and 16S rRNA gene sequencing of fecal samples We extracted DNA from fecal samples using QIAGEN DNeasy PowerSoil Kits. All samples (N=268) were submitted for 16S rRNA gene sequencing (V4 region) on the Illumina MiSeq v2 platform (250 bp, paired-end reads, pooled) at the Michigan State University Genomics Core. Sequencing, library preparation, and preliminary quality filtering were completed according to Caporaso et al. 2012 [60] and Kozich et al. 2013 [61]. Upon receiving bacterial gene sequences, they were processed, filtered for quality and classified into amplicon sequence variants (ASVs) using the Divisive Amplicon Denoising Algorithm (DADA2; v1.36.1) [62] in R. Briefly, we trimmed forward and reverse reads to 240 and 230 bp, respectively. After calculating sequencing errors, reads were clustered into ASVs, and forward and reverse reads were merged. Chimeric sequences were removed, and the remaining sequences were assigned a taxonomy using the Silva reference database (v128) [63]. ASVs classified as Eukarya, Chloroplast, and Mitochondria, or were of unknown Kingdom, were removed from the dataset. The end result was a table of 2,500 total ASVs and their counts for each sample. Samples had an average of 28,956 sequences (± 18,450). Six samples did not amplify well (<400 sequences after processing) and were filtered from all analyses. We exported the final ASV table, ASV taxonomy table, and sample metadata into R (v3.6.2) [48] and these files can be viewed on the GitHub repository for this project (https://github.com/rojascon/SocialNetworks_HyenaGutMicrobiome). 122 Before proceeding with statistical analyses, contaminant bacterial ASVs were identified and removed using the R decontam package [64]. We had 10 extraction kit control samples, which yielded an average of 764 sequences [range 92 – 1635]. ASVs were deemed to be contaminants if they were present in at least 50% of control samples at relative abundances >1% and had decontamination scores below our specified threshold of 0.5. Three ASVs fulfilled these criteria and were removed from the dataset: ASV 538 unclassified Clostridiales, ASV 668 Clostridium senso stricto 1, and ASV 927 Escherichia shigella. Characterizing gut microbiota profiles The ASV abundance table output by DADA2 served as input for all statistical analyses in R. First, gut microbiota composition was visualized using stacked bar plots in ggplot2 [65] depicting the relative abundances of bacterial phyla and families. Only bacterial phyla with average relative abundances >1% across samples and bacterial families with average relative abundances >2% were visualized; all other taxa were clumped into ‘Other.’ To calculate gut microbiota similarity between any combination of hyena pairs, Jaccard, and Bray-Curtis distance matrices were computed using the R vegan package [66]. When computing similarity, Jaccard distances use the presence/absence data of ASVs, while Bray-Curtis distances additionally take into account the relative abundances of these ASVs. For this manuscript, we also generated a series of plots using ggplot2: a PCoA ordination showed the clustering of samples based on Jaccard distances, scatterplots showed gut microbiota similarity (Jaccard/Bray-Curtis) ~ strength of dyadic relationship (association/affiliation/aggression networks ), and other scatterplots depicted gut microbiota richness ~ individual ‘sociability’ (degree, betweenness, or eigenvector centrality). Networks that depicted interactions among focal hyenas were constructed for each time-period using 123 the igraph R package [67]. Individuals represented nodes, and dyadic association indices (SRIs) represented edges. Combined gut microbiota and social network analyses In this study, we constructed two types of generalized linear models. Model 1 tested the relationship between gut microbiota similarity ~ strength of the dyadic relationship, based on association indices, or rates of affiliation or aggression interactions. Model 2 evaluated the relationship between gut microbiota alpha-diversity and an individual’s social connectedness, as indicated by three metrics of social network position (degree, betweenness, and eigenvector centrality). Before proceeding with model 1, we ran a Permutational Multivariate Analysis of Variance (PERMANOVA) test to identify potential covariates to include in the generalized model. The PERMANOVA model included Jaccard or Bray-Curtis distances as the dependent variable, and host social rank (0-1) + age (yrs) + hyena lineage (name of oldest known descendant) + individual identity + month category (e.g. April-June) + time period (1- 8) as predictors. The R vegan package calculated the R2 and p-values for the terms in our PERMANOVA model using 999 permutations. After, we constructed generalized linear models using the R stats package [48] to evaluate the association between gut microbiota similarity and bond strength within dyads. Based on output from the PERMANOVA model, the generalized linear model included gut microbiota similarity (0-1; Jaccard or Bray-Curtis) as the dependent variable, the strength of the dyadic social relationship (0-1; association, affiliation, or aggression indices) as the independent variable, maternal relatedness (0-1) as a covariate, and only included comparisons from the same time-period. Maternal relatedness was assigned as follows: 0.5 for mom-daughter pairs or sisters, 0.25 for grandmother-granddaughter pairs, 0.125 for aunts-niece pairs or cousins, and 0 for unrelated hyenas. Because some hyenas had more than one fecal sample for a given time- 124 period, pairwise microbiota similarity values were averaged so that each hyena dyad only appeared once for that time-period. Model fit was assessed from residuals and variance inflation factors, which were within the norm. Before proceeding with model 2, we first controlled for sequencing effort by subsampling samples to 11,000 reads each using mothur 2 [68]. Four samples had fewer reads than this cutoff and were excluded from alpha-diversity analyses. Good’s coverage values were high (99.14 ± 0.42), indicating that the sequence cutoff was appropriate, and we could proceed with statistical analyses. Two metrics of gut microbiota alpha-diversity were calculated: Chao 1 richness and Shannon diversity. To identify potential covariates of microbiota alpha-diversity to control for in our model, we conducted backwards model selection using the dredge function from the R MuMIn package [69]. The model terms evaluated were hyena social network position + hyena identity + mother’s identity + age (yrs), and a random effect for time-period (1-8). The best fitting model only included social network position as a predictor, and thus our final model was: microbiota alpha-diversity (richness or evenness) ~ hyena sociability (degree, betweenness, or eigenvector centrality). Both metrics of microbiota alpha-diversity were log-transformed, and hyena sociability was scaled. Model fit was assessed from residuals, which were within the norm. Data availability All 16S rRNA sequence data will be uploaded to the NCBI Sequence Read Archive (SRA) prior to journal submission. All R code for analyzing gut microbiota data, generating social networks, conducting statistics, and building figures is available on GitHub (https://github.com/rojascon/SocialNetworks_HyenaGutMicrobiome). 125 Ethical approval Our research and procedures were most recently approved January 8, 2020 (IACUC approval no. PROTO201900126) and comply with the ethical standards of Michigan State University and Kenya. RESULTS Gut microbiota composition across the eight sampled time-periods Across our dataset, which included fecal samples from eight distinct time periods (Table 3.1, Table S3.1), the hyena gut microbiota mostly consisted of Firmicutes (77.42%), Bacteroidetes (7.80%), Actinobacteria (6.35%), Fusobacteria (5.68%), and Proteobacteria (1.21%); all other bacterial phyla were found at abundances <1% (Fig S3.1). Additionally, gut microbiota composition was temporally variable, and samples from 2000-2006 harbored greater abundances of Actinobacteria than more recent samples, while samples from 2011- 2015 contained higher proportions of Bacteroidetes and Fusobacteria, compared to the older samples (Fig S3.1). Similar findings were observed when surveying the gut microbiota of adult female hyenas over a longer timescale of 23 years. In our study, highly-abundant bacterial families of the gut microbiota included Clostridiaceae (17%), Lachnospiraceae (15%), Peptostreptococcaceae (10%), Planococcaceae (7%), Enterococcaceae (6%), Fusobacteriaceae (5%), and Erysipelotrichaceae (4%), among others (Fig 3.1A). Hyenas residing in the clan prior to 2007 harbored greater proportions of Enterococcaceae, Planococcaceae, and Micrococcaceae compared to the other sampled hyenas, while hyenas residing in the clan 2011-2015 contained more Bacteroidaceae, Erysipelotrichaceae, and Fusobacteriaceae in their gut microbiotas than other adult females (Fig 3.1A). 126 A B Figure 3.1 Temporal variation in the gut microbiota composition and structure of wild spotted hyenas. A)Stacked bar plots of gut microbiota composition, specifically of the relative abundance of 16S rRNA gene sequences assigned to each bacterial family. Only bacterial families with average relative abundances >2% across samples are shown, and all others are represented by “Other”. Each colors represents a unique bacterial family and bars are the averages of the samples for that particular time period. B) PCoA ordination of gut microbiota structure based on Jaccard distances; samples are color coded by year (from dark to light colors). Overall, gut microbiota structured varied with host individual identity, which explained, on average 28.9% of the variation among samples. Gut microbiota structure also varied with hyena lineage (5.64% variance explained), the sample time period (4.31%), and sample month (1.61%) (Table S3.2, Fig 3.1B). Hyena social rank and age accounted for little to none of the variation. Thus, when testing whether the strength of social relationships correlated with gut microbiota similarity, we restricted comparisons to within a time period and excluded within-individual comparisons. Additionally, we included maternal relatedness between hyena dyads as a covariate to control for variation associated with genetic relatedness. 127 Effects of social interactions on gut microbiota similarity The aim of this study was to determine whether the strength of dyadic social relationships predicted gut microbiota similarity in adult female hyenas from a single social group. We quantified social bonds between hyena dyads from association (Fig 3.2), affiliation, and aggression networks. The average microbiota similarity (range 0-1) among hyena dyads was 0.30 ± 0.005 (Table S3.3), indicating that overall, gut microbiotas were highly distinct among individuals. The mean strength of dyadic social relationships (range 0- 1) was 0.04 for association indices, 0.15 for affiliative bonds, and 0.42 for aggressive interactions (Table S3.4). 3 6 1 4 7 2 5 8 Figure 3.2 Association networks of the adult female hyenas included in this study. Networks were made separately for each time period; nodes represent adult females and edges (e.g. line thickness) represent the strength of the association between hyena dyads. Only females with fecal samples were included in the network. 128 Behavior type Association (N=1696) Microbiota similarity Jaccard dyad strength relatedness Bray-Curtis dyad strength relatedness Affiliation (N=370) Jaccard dyad strength relatedness Bray-Curtis dyad strength relatedness dyad strength relatedness Jaccard Aggression (N=388) Bray-Curtis dyad strength relatedness 0.024 -0.01 -0.04 -0.02 -0.11 0.02 -0.14 -0.02 0.068 0.06 0.48 0.01 0.08 0.59 0.025 0.36 0.03 0.02 0.05 0.03 0.001 0.38 0.009 0.48 0.01 0.001 0.03 0.03 0.02 0.02 0.0001 0.04 0.94 0.199 0.35 0.99 Contrary to our prediction, hyena dyads that greeted more frequently harbored less similar gut microbiotas than hyena dyads with that greeted less frequently, although the effect sizes were modest (Jaccard and Bray-Curtis; Table 3.2, Figure 3.3). Additionally, no statistically significant relationships were observed between gut microbiota similarity and the frequency of associations between hyena dyads, or between gut microbiota similarity and the frequency and intensity of aggressive interactions between hyena dyads (Table 3.2). Table 3.2 Do social bonds between hyena dyads predict their gut microbiota similarity? Term β SE p.value Coefficient estimates, standard error, and p-values of generalized linear models relating gut microbiota similarity (Jaccard or Bray-Curtis) and dyadic social bond strength based on association, affiliation, or aggression networks. The maternal relatedness between hyena pairs was included as a covariate to control for variation due to genetic relatedness. N represents the number of dyadic comparisons that were included in the analysis. Significant p-values (α<0.05) are bolded. 129 0.4 ) 1 − 0 ( y t i r a l i i m s d r a c c a J 0.2 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ●● ● 0.0 0.00 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ) 1 − 0 ( y t i r a l i i m s s i t r u C − y a r B 0.6 0.4 0.2 0.0 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ●●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 Strength of Social Bond (Affiliation) Strength of Social Bond (Affiliation) ● 1.00 Figure 3.3 Hyenas that greet more frequently do not harbor more similar gut microbiotas than dyads that do not greet as frequently. A)Scatterplot of gut microbiota similarity (Jaccard index) against dyadic social bond strength (affiliation indices). Hyena pairs with larger social bond values greet more frequently than those with lower values. Jaccard distances capture the amount of shared bacterial types between two gut microbiotas. B) Scatterplot of gut microbiota similarity (Bray-Curtis index) against dyadic social bond strength (affiliation indices). Bray-Curtis distances take into account the relative abundances of bacterial types when calculating microbiota similarity. The dark lines and 95% CI represent the relationship between x and y estimated as a linear regression. See Table 3.2 for associated statistics. Effects of individual sociability on gut microbiota alpha-diversity We predicted that hyenas that interacted with a greater number of individuals and were more socially connected would have more diverse gut microbiotas. Three metrics of network position that each captured different aspects of individual sociability were tested: degree (# of social partners), eigenvector centrality (connectedness of the individual as well as the connectedness of its partners), and betweenness (# of times individual acts as a bridge between otherwise unconnected individuals). Mean gut microbiota diversity values or mean sociability values across hyena individuals are listed on Tables S3.5-S3.6. Contrary to our predictions, adult female hyenas that were more socially connected; that is, they had a greater number of social partners, had more connected social partners, and acted as bridges harbored less rich gut microbiotas than less socially connected individuals (Table 3.3, Figure 3.4). Although the effect sizes were modest. Furthermore, gut microbiota 130 evenness, another metric of microbiota alpha-diversity, was only significantly associated with hyena betweenness centrality (Table 3.3); hyenas that bridged unconnected individuals harbored slightly more even gut microbiotas than less gregarious hyenas. In summary, it appears that more social hyenas harbor less rich gut bacterial communities than more isolated hyenas, and gut microbiota evenness generally does not correlate with individual sociability. Table 3.3 Do more social hyenas harbor more diverse gut microbiotas? Behavior type Network Centrality Metric Microbiota α-diversity β SE Association Affiliation Aggression Degree Betweenness Eigenvector Centrality Degree Betweenness Eigenvector Centrality Degree Betweenness Eigenvector Centrality metric Richness Evenness Richness Evenness Richness Evenness Richness Evenness Richness Evenness Richness Evenness Richness Evenness Richness Evenness Richness Evenness p.value 0.028 0.09 0.13 0.056 0.004 0.66 0.07 0.15 0.005 0.33 0.19 0.51 -0.06 0.027 -0.04 0.03 -0.08 -0.006 -0.05 -0.02 -0.07 0.015 -0.03 -0.01 0.02 0.01 0.02 0.01 0.02 0.01 0.02 0.16 0.02 0.01 0.02 0.01 -0.06 0.01 -0.002 0.02 -0.016 -0.008 0.03 0.02 0.42 0.01 0.02 0.93 0.015 0.09 0.02 0.56 0.016 0.61 Coefficient estimates, standard error, and p-values of generalized linear models relating gut microbiota alpha-diversity (Chao1 Richness or Shannon diversity) and individual sociability, as quantified by one of three metrics of network centrality: degree, betweenness, or eigenvector centrality. Three types of networks were examined (association, affiliation, or aggression networks). 131 ) g o l ( s s e n h c R 1 o a h C i ● ● ●● ● ● ● ● ● ● ● ● 6 5 4 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 6 5 4 ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 6 5 4 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 25 50 75 0.00 0.25 0.50 0.75 1.00 10 20 30 Degree Betweenness Figure 3.4 Gut microbiota richness as a function of hyena sociability. Scatterplots of gut microbiota Chao1 richness regressed against three measures of social network centrality: degree, betweenness, and eigenvector centrality. The networks were based on association data. The dark line and 95% CI represent the relationship between x and y estimated as a linear regression. See Table 3.3 for associated statistics. DISCUSSION Gut microbiota similarity Eig.vect Centrality The main aim of this study was to determine whether host social interactions were associated with variation in the gut microbiota in a social group of wild spotted hyenas. We predicted that hyena dyads that had stronger social ties would harbor more similar gut microbiotas than hyenas with weaker social ties. Contrary to our prediction, hyenas that engaged in more affiliative interactions, i.e. greeted more, harbored less similar gut microbiotas than hyenas that did not greet as frequently; although the effect sizes were modest. Furthermore, we found no support for the prediction that hyena dyads that associated more frequently or interacted more within an aggressive context harbored more similar gut microbiotas than hyena pairs that rarely engaged. Collectively, our findings show no support or contradictory support for our predictions, which was unexpected, given that the strength of social relationships predicts gut microbiota variation in wild mice[11], Welsh ponies [10], red-bellied lemurs [9], Verreaux’s sifaka [53], colobus monkeys [70], black howler monkeys [54], baboons [8], chimpanzees [71], and humans [14]. One potential explanation is that bacterial taxa may differ in their preferred transmission routes, and more fine-scale analyses are needed [72]. Our analyses treated all 132 resident gut microbes as a unit, and perhaps, some bacterial taxa are more reliant on social interactions for their transmission compared to other bacterial taxa. In a study of wild wood mice [11], the effects of host social interactions on gut microbiota similarity depended on the bacterial families that were included in the analysis. Exclusion of anaerobic, non-spore forming bacterial families weakened the effects of social bond strength on the gut microbiota, but the exclusion of spore-forming bacteria strengthened the effects. In black howler monkeys [54], only half of the gut microbial genera were correlated with time spent in contact and/or close proximity among dyads. Future work should evaluate whether dyadic bond strength predicts the abundances of bacteria that cannot disperse or form spores (e.g. Bacteroidales, Enterococcus). Analyses should also see whether bacteria that are primarily acquired from the hosts’ diet (animal carcasses) are less likely to be socially transmitted than other bacterial types, since these bacteria may not as reliant on host social behavior for their dispersal [5]. Additionally, findings may also be body-site specific. Although greetings and aggressive interactions can result in physical contact among dyads, their effect might be more apparent on skin or scent gland bacterial communities, rather than gut communities. Captive colonies of Egyptian fruit bats [73] show convergence in their fur communities, such that samples from different individuals collected on the same date are more similar to one another than samples from the same individual collected at different time points. Similarly, the skin microbiotas of fur seals vary with host colony [74], and in humans, cohabitating couples share similar skin bacterial profiles [75]. Hyenas that associate or interact more frequently may have more similar skin or scent gland communities than those that rarely interact. One of the limitations of working with terrestrial wild carnivores with large home ranges and crepuscular activity is that we were unable to collect fecal samples from all of 133 the individuals in the clan. Though 90-120 adult individuals formed part of the clan at any point in time, fecal samples were obtained from 21 unique hyenas for each time-period. That is, only 16%-22% of the clan was represented in our study, and the lack of data may have contributed to our non-significant or marginal findings. Furthermore, gut microbiota profiles were highly dissimilar among individuals within time-periods (mean: 0.30 out of 1.00), suggesting that the 6-month time periods may have been too long to detect any signals of host social interactions on the microbiota. Future studies should sample a clan of hyenas within a 1-month period and obtain fecal samples from a greater number of individuals. Gut microbiota diversity In accordance with prior findings, we predicted that hyenas that were more socially connected would possess more diverse gut microbiotas than hyenas that were less socially connected. Contrary to our prediction, hyenas that had a greater number of social partners, were more socially connected, and had greater potential to act as bridges between otherwise unconnected individuals had less rich gut microbiotas than less social hyenas. Importantly, this was only true for affiliation network data, our effect sizes were modest, and no relationship was observed between gut microbiota evenness and social network position. The fission-fusion dynamics typical of hyena societies could have influenced findings. Unlike some primate species, hyenas are not all together throughout the day, and many spend their time in small subgroups of 7 of fewer individuals, which may fusion into larger subgroups later in the day [22]. This may mean that hyena individuals may have fewer social partners than animals in non-fission fusion societies, and their social relationships may be more transient. Hyenas are also less physical with each other than some primates that groom each other frequently and are more tactile with their hands. Gut microbiota diversity was also negatively correlated or not significantly correlated with host social 134 centrality in red-bellied lemurs [9] and humans from agrarian villages [72], and individuals in these societies may also have smaller social groups and fewer social partners like hyenas. Our findings could also indicate that greater social connectedness causes excessive disturbance to the gut microbiota, and may prevent the establishment of the diverse array of microbes that hyena individuals encounter. The “intermediate disturbance hypothesis” posits that species diversity should be highest at moderate levels of disturbance [76], meaning at intermediate levels of host sociability. Furthermore, host stress is known to reduce gut microbiota alpha diversity in humans and laboratory animals [77, 78]. Although it is unknown whether more social hyenas experience higher levels of physiological stress than less social hyenas, stress could potentially be mediating the effects of individual sociability on gut microbiota diversity. Lastly, in meat- and carrion-eaters like hyenas, there might be additional selective pressures that restrict gut microbiota diversity. A recent study surveyed the metabolic functions of gut microbes in hyenas and reports that one of the most abundant bacterial gene pathways was the synthesis of antibiotics (Rojas, 2021). Synthesizing antibiotics may be a strategy that resident gut microbes employ to compete against the pathogenic bacteria, fungi, and parasites from animal carcasses that may be trying to establish in the gut. These mechanisms might limit which socially or environmentally transmitted microbes can colonize the hyena gut. Hyena individuals might be encountering a diversity of microbes from the environment and other hyenas, but few can ultimately establish in the gastrointestinal tract. CONCLUSIONS Social interactions dictate everyday life in gregarious species, and shape individual fitness. Interactions among group-mates also facilitate the spread of pathogenic bacteria, 135 and potentially of beneficial bacteria. Contrary to what was documented for rodents, equids, and primates the strength of social relationships generally did not positively correlate with gut microbiota similarity in wild spotted hyenas. These findings suggest in hyenas, the effects of host social behavior on gut microbiota composition is bacterial-specific. Some bacterial types may be more responsive to dispersal via social interactions than other bacterial types. Additionally, spotted hyenas live in fission-fusion societies where individuals are mainly found in small subgroups throughout the day, and hyenas may have fewer social partners and exhibit less physical contact with one another than individuals in some primate species. It is important to note that our sample sizes were modest, and only 20% of the clan was represented in our study, as sampling of both behavioral data and fecal material close in space and time is challenging in this species. Thus, our findings do not necessarily indicate that host social interactions are not important drivers of the gut microbiota in wild spotted hyenas, and future studies should sample hyena clans more comprehensively over shorter temporal scales to determine whether gut microbes are socially transmitted in this species. 136 APPENDICES 137 APPENDIX A: SUPPLEMENTAL TABLES Table S3.1 Samples included in this study and their relevant meta data. Behavioral Behavioral fecal hyena samples individuals Fecal Fecal sample sample end date start date 11-Jun-00 2-Jan-00 31-Dec-00 9-Jul-00 9-Oct-01 30-Mar-02 15-Nov-05 21-Apr-06 15-May-11 31-Oct-11 26-Jul-13 6-Feb-13 28-Mar-15 10-Oct-14 8-May-15 19-Oct-15 data start date 11-Jun-99 31-Dec-99 30-Mar-01 21-Apr-05 31-Oct-10 26-Jul-12 28-Mar-14 19-Oct-14 data end date 11-Jun-00 31-Dec-00 30-Mar-02 21-Apr-06 31-Oct-11 26-Jul-13 28-Mar-15 19-Oct-15 (N) 39 37 36 39 30 29 33 25 (N) 17 18 19 11 21 19 20 19 Table S3.2 Results of PERMANOVA model relating host factors to gut microbiota similarity. Predictor Hyena identity Lineage Time-period (8 total) Month category (4 total) Social rank (0-1) Age (yrs) Unexplained variation Jaccard (% variance explained) 27.27, p=0.001 4.69, p=0.001 4.04, p=0.001 1.49, p=0.026 0.63, p=0.018 0.51, p=0.85 61.34 Bray-Curtis (% variance explained) 30.57, p=0.001 6.59, p=0.001 4.59, p=0.001 1.81, p=0.004 0.57, p=0.065 0.44, p=0.158 55.40 Mean variance explained (%) 28.92 5.64 4.31 1.65 0.6 0.47 58.37 Shown are the R2 values (% variance explained) and p-values for host predictors of a PERMANOVA model relating gut microbiota similarity ~ host social rank (0-1) + age (yrs) + hyena lineage + individual identity + month category + time period. Two metrics of gut microbiota similarity were specified: Jaccard distances, which use bacterial taxa presence/absence data and Bray-Curtis distances, which take into account the proportions of bacterial taxa. In the table, host predictors are listed from most to least predictive. Table S3.3 Mean±SE of the two microbiota similarity metrics calculated for the dyadic comparisons included in this study. Microbiota Similarity Metric (0-1) Jaccard Bray-Curtis Mean SE 0.28 0.30 0.003 0.005 Gut microbiotas were compared among dyads within time periods, and not across time periods. 138 Table S3.4 Mean±SE of social bond strength between hyena dyads included in this study. Behavior type (# of dyadic comparisons) Association (N=1696) Affiliation (N=370) Aggression (N=388) Strength of dyadic relationship Mean SE Strength of dyadic association Strength of dyadic affiliative bond 0.048 0.001 0.15 0.006 Strength of dyadic aggressive relationship 0.42 0.013 The strength of dyadic relationships were calculated from association, affiliation, and aggression networks. Hyena individuals had to be seen at least 10 times during the observation period, and each of its partners also had to be seen at least ten times to be included in the network. Because affiliative and aggressive interactions are observed less frequently than hyena co-occurrence, the number of dyadic comparisons for affiliations and aggressions are much less than for associations. Table S3.5 Mean±SE of the two gut microbiota α-diversity metrics calculated for hyena individuals included in this study. Microbiota α-diversity metric Mean SE (N=158 samples) Chao 1 Richness Shannon diversity 390.37 3.38 9.65 0.06 Microbiota α-diversity metrics were calculated for each fecal sample after reducing the number of reads per sample to 11,000 to minimize biases due to sequencing coverage. Table S3.6 Mean±SE of three network centrality metrics calculated for hyena individuals included in this study. Behavior type Association (N=114) Affiliation (N=111) Aggression (N=113) Degree Degree Betweenness Eigenvector centrality 0.49 Centrality Metric Mean SE 25.40 0.62 13.22 1.89 0.02 10.63 0.53 26.99 4.41 Eigenvector centrality 0.41 0.02 27.45 1.83 23.04 3.03 Eigenvector centrality 0.29 0.02 Betweenness Betweenness Degree Network centrality metrics quantified the degree of ‘sociality’ or sociability for each hyena individual. An individual’s degree centrality was their number of connections, ‘betweenness’ centrality was the number of shortest paths between members of any dyad in the network that run through the focal individual, and eigenvector centrality examined the connectedness of the focal individual as well as the connectedness of its connections. The number of hyena samples (N=158) does not equal the number of hyena individuals (N=114) because some hyena individuals had more than 1 fecal sample for that time period, and not all hyena individuals had sufficient behavior data to quantify their sociability. 139 APPENDIX B: SUPPLEMENTAL FIGURES Figure S3.1 Predominant bacterial phyla in the hyena gut microbiota. Stacked bar plots showing the relative frequencies of 16S rRNA gene sequences assigned to each bacterial phylum across samples (N=268). Samples were ordered by year, and each color represents a bacterial phylum. Only bacterial phyla with average relative abundances >1% are plotted, and all others are summarized by “Other.” ) % ( e c n a d n u b A e v i t a e R l 100 75 50 25 0 Bacterial Phylum Actinobacteria Bacteroidetes Firmicutes Fusobacteria Other Proteobacteria 2000 2005 2011 2015 140 REFERENCES 141 REFERENCES 1. Clutton-Brock T. Mammal Societies. Chichester, West Sussex, UK: John Wiley & Sons, Inc; 2016. 2. Candland DK, Conklyn DH. Social dominance and learning in the domestic chicken. Psychon Sci. 1968;11:247–8. 3. Drews C. The concept and definition of dominance in animal behaviour. Behaviour. 1993;125:283–313. doi:10.1163/156853993X00290. 4. VanderWaal KL, Ezenwa VO. Heterogeneity in pathogen transmission: mechanisms and methodology. Funct Ecol. 2016;30:1606–22. doi:10.1111/1365-2435.12645. 5. Lombardo MP. Access to mutualistic endosymbiotic microbes: An underappreciated benefit of group living. Behav Ecol Sociobiol. 2008;62:479–97. 6. Sarkar A, Harty S, Johnson KVA, Moeller AH, Archie EA, Schell LD, et al. Microbial transmission in animal social networks and the social microbiome. Nat Ecol Evol. 2020;4:1020– 1035. 7. Perofsky AC, Lewis RJ, Abondano LA, Di Fiore A, Meyers LA. Hierarchical social networks shape gut microbial composition in wild Verreaux’s sifaka. Proceedings Biol Sci. 2017;284. 8. Tung J, Barreiro LB, Burns MB, Grenier JC, Lynch J, Grieneisen LE, et al. Social networks predict gut microbiome composition in wild baboons. Elife. 2015;2015:1–18. 9. Raulo A, Ruokolainen L, Lane A, Amato K, Knight R, Leigh S, et al. Social behaviour and gut microbiota in red-bellied lemurs ( Eulemur rubriventer ): In search of the role of immunity in the evolution of sociality. J Anim Ecol. 2017;87:1–12. doi:10.1111/1365-2656.12781. 10. Antwis RE, Lea JMD, Unwin B, Shultz S. Gut microbiome composition is associated with spatial structuring and social interactions in semi-feral Welsh Mountain ponies. Microbiome. 2018;6:1–11. 11. Raulo A, Allen B, Troitsky T, Husby A, Firth JA, Coulson T, et al. Social networks strongly predict the gut microbiota of wild mice. ISME J. 2020. doi:10.1101/2020.09.24.311357. 12. Grieneisen LE, Livermore J, Alberts S, Tung J, Archie EA. Group Living and Male Dispersal Predict the Core Gut Microbiome in Wild Baboons. Integr Comp Biol. 2017;57:770–85. 13. Goodfellow CK, Whitney T, Christie DM, Sicotte P, Wikberg EC, Ting N. Divergence in gut microbial communities mirrors a social group fission event in a black-and-white colobus monkey (Colobus vellerosus). Am J Primatol. 2019;81. doi:10.1002/ajp.22966. 14. Dill-McFarland KA, Tang Z-Z, Kemis JH, Kerby RL, Chen G, Palloni A, et al. Close social relationships correlate with human gut microbiota composition. Sci Rep. 2019;9:703. doi:10.1038/s41598-018-37298-9. 15. Johnson KVA. Gut microbiome composition and diversity are related to human personality 142 traits. Hum Microbiome J. 2020;15:100069. doi:10.1016/j.humic.2019.100069. 16. Green DS, Johnson-Ulrich L, Couraud HE, Holekamp KE. Anthropogenic disturbance induces opposing population trends in spotted hyenas and African lions. Biodivers Conserv. 2018;27:871–89. 17. Van Horn RC, Engh AL, Scribner KT, Funk SM, Holekamp KE. Behavioural structuring of relatedness in the spotted hyena (Crocuta crocuta) suggests direct fitness benefits of clan-level cooperation. Mol Ecol. 2004;13:449–58. doi:10.1046/j.1365-294X.2003.02071.x. 18. Kruuk H. The spotted hyena : a study of predation and social behavior. Chicago: University of Chicago Press; 1972. 19. Frank LG. Social organization of the spotted hyaena Crocuta crocuta. II. Dominance and reproduction. Anim Behav. 1986;34:1510–27. 20. Holekamp KE, Smale L, Szykman M. Rank and reproduction in the female spotted hyaena. J Reprod Fertil. 1996;108:229–37. 21. Höner OP, Wachter B, East ML, Streich WJ, Wilhelm K, Burke T, et al. Female mate-choice drives the evolution of male-biased dispersal in a social mammal. Nature. 2007;448:798–801. doi:10.1038/nature06040. 22. Smith JE, Kolowski JM, Graham KE, Dawes SE, Holekamp KE. Social and ecological determinants of fission–fusion dynamics in the spotted hyaena. Anim Behav. 2008;76:619–36. doi:10.1016/J.ANBEHAV.2008.05.001. 23. Smith JE, Powning KS, Dawes SE, Estrada JR, Hopper AL, Piotrowski SL, et al. Greetings promote cooperation and reinforce social bonds among spotted hyaenas. Anim Behav. 2011;81:401–15. doi:10.1016/j.anbehav.2010.11.007. 24. Wahaj SA, Guse KR, Holekamp KE. Reconciliation in the Spotted Hyena (Crocuta crocuta). Ethology. 2001;107:1057–74. doi:10.1046/j.1439-0310.2001.00717.x. 25. Heitlinger E, Ferreira SCM, Thierer D, Hofer H, East ML. The Intestinal Eukaryotic and Bacterial Biome of Spotted Hyenas: The Impact of Social Status and Age on Diversity and Composition. Front Cell Infect Microbiol. 2017;7. 26. Theis KR, Schmidt TM, Holekamp KE. Evidence for a bacterial mechanism for group- specific social odors among hyenas. Sci Rep. 2012;2:615. doi:10.1038/srep00615. 27. Gill SR, Pop M, DeBoy RT, Eckburg PB, Turnbaugh PJ, Samuel BS, et al. Metagenomic Analysis of the Human Distal Gut Microbiome. Science (80- ). 2006;312:1355–9. 28. Ungerfeld E, Leigh M, Forster R, Barboza P. Influence of Season and Diet on Fiber Digestion and Bacterial Community Structure in the Rumen of Muskoxen (Ovibos moschatus). Microorganisms. 2018;6:89. doi:10.3390/microorganisms6030089. 29. Bäckhed F, Ding H, Wang T, Hooper L V., Gou YK, Nagy A, et al. The gut microbiota as an environmental factor that regulates fat storage. Proc Natl Acad Sci U S A. 2004;101:15718–23. 143 30. den Besten G, van Eunen K, Groen A, Venema K, Reijngoud D, Bakker B. The role of short- chain fatty acids in the between diet, gut microbiota, and host energy metabolism. J Lipid Res. 2013;54:2325–40. doi:10.1194/jlr.R036012. 31. Thaiss CA, Zmora N, Levy M, Elinav E. The microbiome and innate immunity. Nature. 2016;535:65–74. 32. Honda K, Littman DR. The microbiota in adaptive immune homeostasis and disease. Nature. 2016;535:75–84. 33. Gerardo NM, Hoang KL, Stoy KS. Evolution of animal Immunity in the light of beneficial symbioses. Philos Trans R Soc B2. 2020;375:20190601. 34. McLaren MR, Callahan BJ. Pathogen resistance may be the principal evolutionary advantage provided by the microbiome. Philos Trans R Soc B2. 2020;375:20190592. 35. Münger E, Montiel-Castro AJ, Langhans W, Pacheco-López G. Reciprocal Interactions Between Gut Microbiota and Host Social Behavior. Front Integr Neurosci. 2018;12:1–14. doi:10.3389/fnint.2018.00021. 36. Bell RHV. A Grazing Ecosystem in the Serengeti. Sci Am. 1971;225:86–93. 37. Maddock L. The “Migration” and Grazing Succession. In: Sinclair ARE, Norton-Griffiths M, editors. Serengeti: Dynamics of an Ecosystem. Chicago: The University of Chicago Press; 1979. p. 104–29. 38. Riggio J, Jacobson A, Dollar L, Bauer H, Becker M, Dickman A, et al. The size of savannah Africa: A lion’s (Panthera leo) view. Biodivers Conserv. 2013;22:17–35. doi:10.1007/s10531- 012-0381-4. 39. Spagnuolo OSB, Jarvey JC, Battaglia MJ, Laubach ZM, Miller ME, Holekamp KE, et al. Mapping Kenyan Grassland Heights Across Large Spatial Scales with Combined Optical and Radar Satellite Imagery. Remote Sens. 2020;12. 40. Western D, Russell S, Cuthill I. The Status of Wildlife in Protected Areas Compared to Non- Protected Areas of Kenya. PLoS One. 2009;4:e6140. doi:10.1371/journal.pone.0006140. 41. Green DS, Zipkin EF, Incorvaia DC, Holekamp KE. Long-term ecological changes influence herbivore diversity and abundance inside a protected area in the Mara-Serengeti ecosystem. Glob Ecol Conserv. 2019;20:e00697. doi:10.1016/j.gecco.2019.e00697. 42. Ogutu JO, Piepho HP, Dublin HT, Bhola N, Reid RS. Rainfall influences on ungulate population abundance in the Mara-Serengeti ecosystem. J Anim Ecol. 2008;77:814–29. 43. Frank LG, Glickman SE, Powch I. Sexual dimorphism in the spotted hyaena (Crocuta crocuta). J Zool. 1990;221:308–13. doi:10.1111/j.1469-7998.1990.tb04001.x. 44. Hofer H, East M. Population dynamics, population size, and the commuting system of Serengeti spotted hyaenas. In: Sinclair ARE, Arcese P, editors. Serengeti II : dynamics, management, and conservation of an ecosystem. Chicago: University of Chicago Press; 1995. p. 665. 144 45. Holekamp KE, Smale L, Szykman M. Rank and reproduction in the female spotted hyaena. Reproduction. 1996;108:229–37. doi:10.1530/jrf.0.1080229. 46. Strauss ED, Holekamp KE. Inferring longitudinal hierarchies: Framework and methods for studying the dynamics of dominance. J Anim Ecol. 2019;88:521–36. doi:10.1111/1365- 2656.12951. 47. Altmann J. Observational Study of Behavior: Sampling Methods. Behaviour. 1974;49:227– 66. doi:10.1163/156853974X00534. 48. R Core Team. R: a language and environment for statistical computing. In: R Foundation for Statistical Computing. Vienna, Austria; 2019. https://www.gbif.org/tool/81287/r-a-language-and- environment-for-statistical-computing. Accessed 28 Aug 2019. 49. Robitaille AL, Webber QMR, Vander Wal E. Conducting social network analysis with animal telemetry data: Applications and methods using spatsoc. Methods Ecol Evol. 2019;10:1203–11. doi:10.1111/2041-210X.13215. 50. Farine DR. Animal social network inference and permutations for ecologists in R using asnipe. Methods Ecol Evol. 2013;4:1187–94. doi:10.1111/2041-210X.12121. 51. Turner JW, Robitaille AL, Bills PS, Holekamp KE. Early-life relationships matter: Social position during early life predicts fitness among female spotted hyenas. J Anim Ecol. 2021;90:183–96. 52. Turner JW, Bills PS, Holekamp KE. Ontogenetic change in determinants of social network position in the spotted hyena. Behav Ecol Sociobiol. 2018;72:10. http://link.springer.com/10.1007/s00265-017-2426-x. Accessed 22 Jan 2018. 53. Perofsky AC, Lewis RJ, Abondano LA, Difiore A, Meyers LA. Hierarchical social networks shape gut microbial composition in wild verreaux’s sifaka. Proc R Soc B Biol Sci. 2017;284. 54. Amato KR, Belle S Van, Fiore A Di, Estrada A, Stumpf R, White B, et al. Patterns in Gut Microbiota Similarity Associated with Degree of Sociality among Sex Classes of a Neotropical Primate. Microb Ecol. 2017;74:250–8. doi:10.1007/s00248-017-0938-6. 55. Cairns SJ, Schwager SJ. A comparison of association indices. Anim Behav. 1987;35:1454– 69. 56. Whitehead H, James R. Generalized affiliation indices extract affiliations from social network data. Methods Ecol Evol. 2015;6:836–44. doi:10.1111/2041-210X.12383. 57. Godde S, Humbert L, Côté SD, Réale D, Whitehead H. Correcting for the impact of gregariousness in social network analyses. Anim Behav. 2013;85:553–8. 58. Szykman M, Engh AL, Van Horn RC, Boydston EE, Scribner KT, Holekamp KE. Rare Male Aggression Directed Toward Females in a Female-Dominated Society: Baiting Behavior in the Spotted Hyena. Aggress Behav. 2003;29:457–74. 59. Sosa S, Sueur C, Puga-Gonzalez I. Network measures in animal social network analysis: Their strengths, limits, interpretations and uses. Methods Ecol Evol. 2021;12:10–21. 145 doi:10.1111/2041-210X.13366. 60. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, et al. Ultra-high- throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 2012;6:1621–4. doi:10.1038/ismej.2012.8. 61. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. Development of a Dual- Index Sequencing Strategy and Curation Pipeline for Analyzing Amplicon Sequence Data on the MiSeq Illumina Sequencing Platform. Appl Environ Microbiol. 2013;79:5112–20. doi:10.1128/AEM.01043-13. 62. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: High- resolution sample inference from Illumina amplicon data. Nat Methods. 2016;13:581–3. 63. Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 2013;41:D590-6. doi:10.1093/nar/gks1219. 64. Davis NM, Proctor DiM, Holmes SP, Relman DA, Callahan BJ. Simple statistical identification and removal of contaminant sequences in marker-gene and metagenomics data. Microbiome. 2018;6:226. doi:10.1186/s40168-018-0605-2. 65. Wickham H. ggplot2: Elegant Graphics for Dat Analysis. New York, NY: Springer-Verlag New York; 2009. doi:10.1007/978-0-387-98141-3. 66. Oksanen J, Blanchet F, Friendly M, Kindt R, Legendre P, McGlinn D, et al. vegan: Community Ecology Package. R Packag version 24-6. 2018. https://cran.r- project.org/web/packages/vegan/index.html. 67. Csardi G, Nepusz T. The Igraph Software Package for Complex Network Research. InterJournal, Complex Syst. 2005;:1695–705. https://www.researchgate.net/publication/221995787_The_Igraph_Software_Package_for_Com plex_Network_Research/citations. Accessed 13 Apr 2021. 68. Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol. 2009;75:7537–41. doi:10.1128/AEM.01541-09. 69. Barton K. MuMIn: Multi-Model Inference. R package version 1.43.17. https://CRANR- project.org. 2020. 70. Wikberg EC, Christie D, Sicotte P, Ting N. Interactions between social groups of colobus monkeys (Colobus vellerosus) explain similarities in their gut microbiomes. Anim Behav. 2020;163:17–31. doi:10.1016/j.anbehav.2020.02.011. 71. Moeller AHA, Foerster S, Wilson ML, Pusey AE, Hahn BH, Ochman H. Social behavior shapes the chimpanzee pan-microbiome. Sci Adv. 2016;2:e1500997–e1500997. doi:10.1126/sciadv.1500997. 72. Brito IL, Gurry T, Zhao S, Huang K, Young SK, Shea TP, et al. Transmission of human- 146 associated microbiota along family and social networks. Nature Microbiology. 2019;4:964–71. doi:10.1038/s41564-019-0409-6. 73. Kolodny O, Weinberg M, Reshef L, Harten L, Hefetz A, Gophna U, et al. Coordinated change at the colony level in fruit bat fur microbiomes through time. Nat Ecol Evol. 2019;3. doi:10.1038/s41559-018-0731-z. 74. Grosser S, Sauer J, Paijmans AJ, Caspers BA, Forcada J, Wolf JBW, et al. Fur seal microbiota are shaped by the social and physical environment, show mother–offspring similarities and are associated with host genetic quality. Mol Ecol. 2019;28:2406–22. 75. Ross AA, Doxey AC, Neufeld JD. The Skin Microbiome of Cohabiting Couples. mSystems. 2017;2:1–15. 76. Roxburgh SH, Shea K, Wilson JB. The intermediate disturbance hypothesis: patch dynamics and mechanisms of species coexistence. Ecology. 2004;85:359–71. doi:10.1890/03- 0266. 77. Knowles SR, Nelson EA, Palombo EA. Investigating the role of perceived stress on bacterial flora activity and salivary cortisol secretion: A possible mechanism underlying susceptibility to illness. Biol Psychol. 2008;77:132–7. doi:10.1016/j.biopsycho.2007.09.010. 78. Partrick KA, Chassaing B, Beach LQ, McCann KE, Gewirtz AT, Huhman KL. Acute and repeated exposure to social stress reduces gut microbiota diversity in Syrian hamsters. Behav Brain Res. 2018;345:39–48. doi:10.1016/j.bbr.2018.02.005. 147 CHAPTER FOUR: PUBLISHED AS: HOST PHYLOGENY AND HOST ECOLOGY STRUCTURE THE GUT MICROBIOTA AT DIFFERENT TAXONOMIC SCALES Rojas, C.A., S. Ramírez-Barahona, K.E. Holekamp, and K.R. Theis. 2021. Host phylogeny and host ecology structure the mammalian gut microbiota at different taxonomic scales. Animal Microbiome 3(33). https://doi.org/10.1186/s42523-021-00094-4 ABSTRACT The gut microbiota is critical for host function. Among mammals, host phylogenetic relatedness and diet are strong drivers of gut microbiota structure, but one factor may be more influential than the other. Here, we used 16S rRNA gene sequencing to determine the relative contributions of host phylogeny and host diet in structuring the gut microbiotas of 11 herbivore species from 5 families living sympatrically in southwest Kenya. Herbivore species were classified as grazers, browsers, or mixed-feeders and dietary data (% C4 grasses in diet) were compiled from previously published sources. We found that herbivore gut microbiotas were highly species-specific, and that host taxonomy accounted for more variation in the gut microbiota (30%) than did host dietary guild (10%) or sample month (8%). Overall, similarity in the gut microbiota increased with host phylogenetic relatedness (r=0.74) across the 11 species of herbivores, but among 7 closely related Bovid species, dietary %C4 grass values more strongly predicted gut microbiota structure (r=0.64). Additionally, within bovids, host dietary guild explained more of the variation in the gut microbiota (17%) than did host species (12%). Lastly, while we found that the gut microbiotas of herbivores residing in southwest Kenya converge with those of distinct populations of conspecifics from central Kenya, fine-scale differences in the abundances of bacterial amplicon sequence variants (ASVs) between individuals from the two regions were also observed. Overall, our findings suggest that host phylogeny and taxonomy strongly structure the gut microbiota across broad host taxonomic scales, but these gut microbiotas 148 can be furthered modified by host ecology (i.e., diet, geography), especially among closely related host species. INTRODUCTION The gut microbiota, which is the collection of microbes inhabiting the gastrointestinal tract, is essential to host functioning. In mammals, resident gut microbes promote the digestive efficiency of their hosts by synthesizing vitamins, breaking down fiber, and supplementing the host with energy released from fermentation [1–4]. The gut microbiota also interacts with the host immune system, and may also modulate behavior [5–7]. Due to the critical importance of the gut microbiota for host performance, research has focused on determining the forces that shape its assembly and composition. Decades of research show that across vertebrate hosts, the gut microbiota is predominantly shaped by host phylogeny and ecology. Closely related host species tend to have more similar gut microbiotas than more distantly related host species [8–12] and this congruence between host phylogenetic relatedness and gut microbiota similarity is termed “phylosymbiosis” [13–15]. However, gut microbiotas can also be further shaped by their host’s ecology, including their host’s diet, habitat, and geographic location [16–19]. Thus, although both of these host factors may shape the gut microbiota, their relative contributions might be influenced by a variety of variables including the taxonomic breadth of the host species surveyed, and the diversity of host habitats, diets, and gut physiologies represented. If phylosymbiosis is observed, both host phylogenetic relatedness and ecology could be contributing to the pattern. Several studies have disentangled the effects of these two factors and have shown that phylosymbiosis can be observed among hosts that share habitats or diets, and among hosts that reside in different habitats and consume different diets. For example, in mice, voles, and shrews, gut microbiotas tend to be more similar 149 among closely related host species, despite these animals occupying different habitats and consuming different diets [20]. In populations of American pikas (Ochotona princeps) from different mountain ranges, a cladogram of gut microbiota similarity was congruent with a phylogeny of host genetic similarity [21]. Among hosts with overlapping diets, gut microbiotas still exhibit patterns consistent with phylosymbiosis, as has been documented for folivorous primates [9]. Furthermore, in 33 species of sympatric herbivores from the Laikipia region in central Kenya, host phylogenetic relatedness strongly predicted gut microbiota composition (r = 0.91), which was weakly correlated with host diet (r = 0.28) [22], suggesting that convergence of gut microbiotas among closely related hosts was not primarily due to similarities in their diet. Here, we build upon this work and use 16S rRNA gene sequencing to determine the relative influences of host phylogenetic relatedness and host ecology in structuring the gut microbiotas of 11 species of herbivores living sympatrically in the Masai Mara National Reserve (henceforth the Masai Mara) in southwestern Kenya. We survey the gut microbiotas of African buffalo, domestic cattle, common eland, impala, Kirk’s dik-dik, Thompson’s gazelle, topi, Masai giraffe, common warthog, plains zebra, and African elephant (Table 4.1). These species represent 5 mammalian families (Bovidae, Elephantidae, Equidae, Giraffidae, and Suidae) and three dietary guilds: grazers, browsers, and mixed-feeders. Furthermore, we compare the gut microbiotas of conspecific herbivores from the Masai Mara and Laikipia to determine the extent to which host geography and/or local habitat influence gut microbiota composition and patterns of phylosymbiosis. The two regions differ in their altitude, soils, rainfall, vegetation, mammal densities, and degree of human disturbance [23–27], any of which could potentially affect the gut microbiota compositions of their resident herbivores. Specifically, our study aims were to survey the gut microbiotas of 11 species of herbivores and 1) determine whether host phylogenetic 150 relatedness or diet more strongly predict gut microbiota similarity among these hosts at broad taxonomic scales (i.e. among all study species) and lesser taxonomic scales (i.e. among 7 closely related Bovid species), 2) evaluate the influences of host taxonomy (family and species) and host dietary guild on gut microbiota composition and diversity, and 3) examine the amount of variance in the gut microbiota explained by host phylogeny and ecology (i.e., diet, geography) in conspecific hosts from the Masai Mara (this study) and Laikipia [22]. Collectively, our findings elucidate the factors shaping the gut microbiota of hosts at greater and lesser taxonomic scales. Table 4.1 List of host study species and their associated metadata. Order Family Species Dietary Guild Total Samples (N) 18 14 8 20 37 14 19 25 9 5 12 Analyzed samples (N) 17 13 8 20 31 14 19 18 8 5 12 Cetartiodactyla Bovidae Cetartiodactyla Bovidae Cetartiodactyla Bovidae Cetartiodactyla Bovidae Cetartiodactyla Bovidae Cetartiodactyla Bovidae Cetartiodactyla Bovidae Cetartiodactyla Giraffidae Cetartiodactyla Suidae Perissodactyla Equidae Proboscidea grazer grazer African buffalo Domestic cattle Common eland mixed feeder mixed feeder Impala browser Kirk’s dik dik Thomson’s mixed feeder gazelle Topi Masai giraffe Warthog Plains zebra grazer browser mixed feeder grazer Elephantidae African elephant mixed feeder METHODS Study location and sampling. Fecal samples (N=181) were collected opportunistically from 11 species of herbivores permanently residing in the Talek and Mara Triangle regions of the Masai Mara (1°22′19″S, 34° 56′17″E) from March-June 2018 (Table 4.1). This Reserve is covered by open rolling grassland interspersed with seasonal watercourses and riparian vegetation. It has two rainy seasons (March-May and November-December, with annual rainfall 151 >1000mm) [28], and 81% of our sampling took place during the rainy months, particularly during the month of March (Figure S4.4). Although the Masai Mara is home to small resident populations of zebra and wildebeest, millions of these individuals migrate into the Reserve from July-October every year. Because our sampling occurred before July, samples from wildebeest and zebras were limited. For fecal sample collection, we either observed animals defecating or identified species-of-origin based on the size, shape, and consistency of fresh dung, following Kartzinel et al. [22]. Samples were then placed in sterile cryogenic vials and stored in liquid nitrogen until they were transported on dry ice to Michigan State University, where they remained frozen at -80°C until nucleic acid extraction. For a list of samples and their associated metadata, see the Github repository for this project (https://github.com/rojascon/Rojas_et_al_2020_African_herbivores_gut_microbiome). While we did not directly collect diet data from the surveyed herbivores, we used Kingdon’s East African Mammals [29–32] to classify our study species into grazers, browsers, and mixed-feeders. To obtain more fine-scale data on host diet, we also compiled dietary C4 (%) data for these herbivores from previously published studies (Table S4.1) [33–36]. Percent C4 values reflect the proportion of monocotyledon grasses consumed relative to trees, shrubs, and forbs. DNA extraction and 16S rRNA gene sequencing. Fecal samples were sent to the University of Illinois Chicago (UIC) Sequencing Core for automated DNA extractions using QIAGEN DNeasy PowerSoil kits (Valencia, CA, USA). DNA concentrations of the fecal sample extracts were quantified using Qubit. The V4 region of the 16S rRNA gene was targeted for sequencing on the Illumina MiSeq platform at the Michigan State University Genomics Core, using published protocols by Caporaso et al. 2012 [37] and Kozich et al. 2013 [38]. 152 Sequence processing and bioinformatics. Sequences from Masai Mara herbivore gut microbiotas were processed in R (v.3.6.2) [39] using the Divisive Amplicon Denoising Algorithm (DADA2) pipeline (v.1.14.1) [40] to infer amplicon sequence variants (ASVs). Briefly, reads were filtered for quality, allowing for 2 and 3 errors per forward and reverse read, respectively (trimLeft = c(10, 10), maxN = 0, maxEE = 2, truncQ = 2). Forward reads were trimmed to 240bp and reverse reads to 200bp; these paired-end reads were merged. Sequences were then dereplicated to remove redundancy and ASVs were inferred by pooling reads from all samples. Prior to creating the ASV abundance table, chimeras were removed and ASVs were taxonomically classified using the SILVA rRNA gene reference database (v.132) [41] with an 80% confidence threshold. ASVs taxonomically assigned as Eukarya, Chloroplasts, or Mitochondria were removed from the dataset, as were those of unknown Kingdom origin; 12,938 total ASVs remained. The resulting ASV table and the taxonomic designations of the ASVs are available on GitHub. On average, samples retained over 70% (± 11%) of their total sequences after processing in DADA2. Nineteen samples did not amplify well (<400 sequences after processing) and were removed from the dataset. Most of these samples belonged to browser species (giraffes and dik-diks), suggesting that there may have been PCR inhibitors in their fecal samples (e.g., humic acid, tannins) that prevented successful extraction of DNA or library preparation. Table 4.1 has the sample sizes (N) for each study species before and after this filtering. Microbiota composition analyses Statistical analyses and data visualization were completed in R unless otherwise stated. To visualize microbiota composition, stacked barplots were constructed in ggplot2 (v.3.3.2) [42]. These plots showed the bacterial phyla, families, and genera with average relative abundances greater than 1% across samples. We also identified the ASVs (N = 10) 153 that were present in >90% of samples across all host species, and the relative abundances of these ASVs were displayed as heatmaps using the R pheatmap package (v.1.0.12) [43]. Sequences from the 10 ASVs were blasted against the National Center for Biotechnology Information (NCBI) Nucleotide database [44] to find similar biological sequences from known bacterial taxa. Furthermore, we also identified ASVs that were biased towards particular host species; these were ASVs that were present in >75% of the samples for a particular host species (e.g., giraffes) and absent in 97% of samples from the other host species. To detect the bacterial taxa strongly associated with particular host families or dietary guilds, we used the R indicspecies package (v.1.7.9) [45], which calculates an indicator value for each bacterial taxon based on its prevalence in a given group and absence in others. A table of bacterial family relative abundances was used as input, and significance was assessed via permutation tests using 999 permutations (α=0.05). Bacterial families with indicator values >0.4 were plotted in ggplot2. Microbiota α-diversity statistical analyses Prior to alpha-diversity analyses, we controlled for the potential influences of sequencing depth by subsampling all samples to 17,000 sequences using the mothur (v.1.42.3) [46] sub.sample command. Four fecal samples did not meet this sequence cutoff criterion and were excluded from all alpha-diversity analyses. Mothur was used to construct rarefaction curves of ASV richness vs. sequencing depth (Figure S4.5) and Good’s coverage values averaged 97.78 ± 0.91 across all samples, indicating that sample coverage was high and appropriate for characterizing fecal microbiota profiles. These values are comparable to those typically reported in other mammalian gut microbiota studies [8, 47, 48]. 154 Microbiota alpha-diversity was estimated using Chao1 Richness, Shannon diversity, and Faith’s Phylogenetic Diversity (PD) in R. Chao1 Richness and Shannon indices were calculated using the phyloseq package (v.1.33.0) [49]. To obtain measures of Faith’s PD, we constructed a phylogenetic tree of ASV sequences using phangorn (v. 2.5.5) [50] and calculated PD using the picante package (v.1.8.1) [51]. The effects of predictor variables on each measure of alpha-diversity across all samples were evaluated via linear mixed models (LMMs) using the lme4 package (v.1.1.23) [52], specifying host dietary guild and host family as fixed variables and sample month as a random effect. A similar model that included host species as a predictor in lieu of host family was also evaluated. A third model was built for bovid samples only, which included host species and dietary guild as predictors. The significance of each predictor variable was determined by calculating likelihood ratio χ2 test statistics (α=0.05) on the full models using the car package (v.3.0.7) [53]. These tests were followed by TukeyHSD post-hoc tests with Benjamini-Hochberg adjustments to control for multiple comparisons. Boxplots of microbiota alpha-diversity were generated in ggplot2. To further quantify the influence of host diet on the three metrics of gut microbiota alpha-diversity, we conducted partial Mantel tests with 999 permutations using the R vegan package (v.2.5.7) [54]. Specifically, we evaluated whether similarity in gut microbiota alpha- diversity was associated with similarity in dietary C4(%) after accounting for variation due to host phylogenetic relatedness. The 3 matrices used as input were i) a dissimilarity matrix of gut microbiota alpha-diversity, ii) a dissimilarity matrix of host %C4, and iii) a matrix of host divergence times. Microbiota β-diversity analyses and testing for phylosymbiosis In order to determine the relative contributions and amount of variance explained by host predictor variables, permutational multivariate analyses of variance (PERMANOVA) tests based on Bray-Curtis, Jaccard, and Unifrac distance matrices were run using vegan. 155 Bray-Curtis/Jaccard distances were estimated using vegan, whereas weighted and unweighted Unifrac distances were estimated using phyloseq. Bray-Curtis and weighted Unifrac distances take into account the abundances of bacterial taxa while Jaccard and unweighted Unifrac metrics only consider their presence or absence. Both UniFrac metrics utilize information on the phylogenetic diversity of bacterial members when calculating microbiota similarity. PERMANOVA model #1 included sample month, host dietary guild, and host family as predictors (in this order) and included all 11 host species. Model 2 was identical to Model 1, except it included host species in lieu of host family. Model 3 was similar to Model 2, except it was restricted to the Bovid dataset (7 host species). Microbiota similarity and groupings across samples were visualized via Principal Coordinates Analysis (PCoA) plots. To test for phylosymbiosis, i.e., the congruence between host phylogenetic relatedness and gut microbiota similarity, mean divergence times (mya) were calculated between every pair of host species in R. First, we retrieved 1000 phylogenetic trees that included all species of Artiodactyla and African elephants (Loxodonta Africana) from Upham’s et al. (2019) Mammalian supertree [55]. The trees were randomly sampled from the posterior distribution of Upham’s supertrees (Mammals birth-death tip-dated DNA-only trees) using the VertLife online resource (http://vertlife.org/). Each tree was pruned to include only the species in this study, and branch lengths (i.e. divergence times between each pair of host species) were extracted using the R ape package (v.5.4.1) [56]. All 1000 trees showed the same phylogenetic relationships among the study species and matrices of mean divergence times were estimated from those trees. To determine the strength of the phylosymbiosis signal relative to influences attributable to host diet, we conducted partial correlation tests using Spearman correlations with the R ppcor package [57]. These tests correlated i) gut microbiota dissimilarity with host phylogenetic distance (divergence times), 156 while controlling for dietary similarity (%C4), or ii) correlated gut microbiota dissimilarity with dietary dissimilarity, while controlling for host phylogenetic distance. We visualized the phylosymbiosis findings by plotting gut microbiota similarity (0-1) against host phylogenetic divergence time (mya) in ggplot2. We added a trendline to this plot for plotting purposes; the trendline represented the best fit line of a linear model regressing Bray-Curtis dissimilarity with host phylogenetic distance. We also constructed a consensus phylogeny of our host species and compared it against a dendrogram of gut microbiota dissimilarity, which was calculated using hierarchical clustering with the R stats package [39] and plotted using the ape package. Comparisons of Masai Mara and Laikipia herbivores. In order to compare the gut microbiotas of Masai Mara (1°22′19″S, 34° 56′17″E) herbivores to the gut microbiotas of their conspecifics in Laikipia (0°17′33″N, 36° 53′55″E) (> 300 km from the Masai Mara), we downloaded all publicly available sequences from Kartzinel et al. [22], and combined them with the raw 16S rRNA gene sequences from this study (Masai Mara herbivores). The sequences from both studies were then processed together in DADA2. A total of eight herbivore species overlapped between the two studies: African buffalo, domestic cattle, common eland, impala, giraffe, warthog, plains zebra, and African elephant. 96% of samples from Kartzinel et al. [22] and 81% of samples from our study were collected during the wet seasons in their respective regions (Table S4.10), although, in general, Laikipia is more arid than the Masai Mara, with only 300-600mm precipitation annually [26, 58, 59]. For a list of all samples (N=305), and their associated metadata, see the Availability of data and materials section. The bioinformatics processing and statistical analyses were performed as described above, with a few exceptions. In DADA2, forward and reverse reads were trimmed to 240bp and 150bp, respectively, to better account for sequence quality. Up to 2 errors were allowed 157 per forward read and up to 4 errors per reverse read. To identify the strongest predictors of gut microbiota structure, we constructed a PERMANOVA model that included sample month, host geographic region, host dietary guild, and host species as variables (in this order). PCoA ordinations and testing for phylosymbiosis were conducted as described earlier. To visualize gut microbiota compositions between Masai Mara and Laikipia herbivores, a heatmap of the 32 most abundant bacterial ASVs was constructed using R pheatmap. We furthered compared the gut microbiotas of conspecific hosts by conducting Linear discriminant analysis Effect Size (LEfSe) [60] in the Galaxy platform [61] using default parameters. Only ASVs >0.01% average relative abundance across samples were included in the data frame uploaded to Galaxy. ASVs that were enriched in hosts from one geographic region relative to the other were visualized via diverging dot plots in R with the ggplot2 package. Ethical Approval Our research and procedures were most recently approved on January 8, 2020 (IACUC approval no. PROTO201900126) and comply with the ethical standards of Michigan State University and Kenya. Availability of data and materials The 16S rRNA gene sequence data from this study were deposited in NCBI’s Sequence Read Archive under BioProject PRJNA656793 and accession numbers SAMN15803511- SAMN15803691. Sample metadata, data output by DADA2 (ASV table & ASV taxonomic classifications), data obtained from LEfSe, and R scripts for all analyses and figures included in this manuscript are available on Github (https://github.com/rojascon/Rojas_et_al_2021_African_herbivores_gut_microbiome). 158 RESULTS Aim 1: Determine the strongest predictor of gut microbiota similarity among herbivore hosts at greater and lesser taxonomic scales We conducted partial correlation coefficient tests to determine the relative contributions of host phylogenetic relatedness and diet in predicting gut microbiota structure. These tests evaluated the strength of the relationship between host phylogenetic relatedness and gut microbiota similarity (phylosymbiosis), while controlling for dietary similarity, and assessed the relationship between host dietary similarity and gut microbiota similarity, while controlling for phylogenetic relatedness. Phylogenetic relatedness was based on divergence times between host species and diet was quantified by %C4 grass values in the diet (e.g. proportion of monocotyledon grasses consumed relative to trees and shrubs) previously published for these host species [33–36] (Table S4.1). Although a dendrogram of gut microbiota similarity did not closely reflect host phylogeny (Figure 4.1A), partial correlation coefficient tests indicated that gut microbiota similarity increased with host phylogenetic relatedness even after controlling for dietary similarity (Table 4.2). Across the 11 herbivore species, the gut microbiotas were generally more similar among closely related host taxa (e.g., buffalo and cattle) than among distantly related host taxa (e.g., impala and elephant), and the average strength of the phylosymbiosis signal across microbiota similarity metrics was 0.74 (Figure 4.1B). No relationship was observed between host dietary similarity (%C4) and gut microbiota similarity at this broad taxonomic scale (Table 4.2). Importantly, at a lesser host taxonomic scale, among 7 closely related Bovid species, we observed the opposite patterns. Gut microbiota similarity did not covary with host phylogenetic relatedness after adjusting for dietary similarity among these bovid species (Figure 4.1B), but we did find a significant relationship between dietary similarity and gut microbiota similarity (average r=0.64) (Table 159 4.2). In summary, across broad host taxonomic scales, host phylogenetic relatedness strongly predicted gut microbiota similarity, but host diet more strongly predicted gut microbiota structure at a lesser host taxonomic scale. Table 4.2. The relative contributions of host phylogenetic relatedness and diet in predicting gut microbiota similarity. Model Metric Across all host study species (11 sp.) (N=165) Across bovids (7 sp.) (N=122) Bray-Curtis Jaccard Unifrac (weighted) Unifrac (unweighted) Bray-Curtis Jaccard Unifrac (weighted) Unifrac (unweighted) Phylogenetic Relatedness Z stat 8.90 7.50 8.51 p.val <0.0001 <0.0001 <0.0001 R 0.77 0.72 0.76 Dietary Similarity (% C4 grasses) R 0.15 0.03 0.10 Z stat 1.09 0.27 0.73 p.val 0.27 0.78 0.46 0.71 7.47 <0.0001 0.02 0.152 0.87 0.43 0.35 0.31 2.07 1.63 1.41 0.29 1.29 0.05 0.12 0.17 0.21 0.68 0.69 0.57 0.63 3.98 4.05 2.94 3.52 <0.001 <0.001 <0.001 0.002 Shown are the rho, test statistic, and p-values associated with partial correlation coefficient tests that evaluated the correlation between 2 variables (e.g. gut microbiome similarity and phylogenetic relatedness, while controlling for a third (e.g. dietary similarity). The tests were conducted on 4 types of gut microbiome distance metrics, and significant p-values are bolded. 160 Figure 4.1. African herbivore gut microbiotas exhibit patterns consistent with phylosymbiosis. Madoqua kirkii Eudorcas thomsonii Damaliscus lunatus Aepyceros melampus Syncerus caffer Bos taurus Tragelaphus oryx Giraffa camelopardalis Phacochoerus africanus Equus quagga Loxodonta africana A) B) C) D) Table S1.3 (cont’d) 161 Figure 4.1 (cont’d) A) Phylogenetic tree of host species (left) obtained from pruning Upham’s et al. 2019 Mammalian supertree, compared against a dendogram (right) of gut microbiota similarity using hierarchical clustering. B) Scatterplot of pairwise host divergence times (in millions of years) vs. gut microbiota similarity (Bray- Curtis distances) across all sampled herbivores (left) and within the single host family Bovidae (right). The plot on the left has a trendline representing the best fit line of a linear model regressing Bray-Curtis dissimilarity with host phylogenetic distance, which was added for plotting purposes. C) PCoA plots constructed from Bray-Curtis dissimilarity matrices. Each point represents a sample and is color-coded by host family (left) or host dietary guild (right). Closeness of points indicates high community similarity. The percentage of variance accounted for by each principal-coordinate axis is shown in the axis labels. D) PCoA plots constructed from Bray-Curtis dissimilarity matrices of bovid species only. Each point is color- coded by host species (left) or host dietary guild (right). PERMANOVA analyses that included categorical variables for host taxonomy (family or species), dietary guild (grazer, browser, or mixed feeder) and sample month echoed the findings described above. Across the surveyed herbivores, host family explained on average ~23% of the variation in gut microbiota structure, followed by host dietary guild (10%), and sample month (8%) (Table 4.3). Regardless of whether distance matrices took into account the presence/absence of bacterial taxa, their proportional abundances, or their phylogenetic relatedness, the percent variation explained by each host factor was consistent. Therefore, for brevity, we only present PCoA ordination plots using the Bray- Curtis index. These plots show that gut microbiotas primarily partition by host family, and also secondarily by host dietary guild (Figure 4.1C). Additionally, we conducted the same PERMANOVA statistics from above but specified host species in lieu of the host family term; here, host species explained on average 30.92% of the variation across distance metrics, host diet explained 10.44% of the variation and sample month accounted for 7.93% of the variance (Table S4.2). Lastly, within Bovidae, host dietary guild was a slightly stronger predictor of the gut microbiota than host species or sample month. On average, host dietary guild accounted for 17.3% of the variation, whereas host species explained 12.2% of the variation, and sample month contributed to 7.3% of the variation (Table 4.3). PCoA ordinations showed that the gut microbiotas of bovids clustered by host dietary guild 162 and host species (Figure 4.1D). Collectively, our findings suggest that host phylogenetic relatedness and taxonomy predict gut microbiota structure across the studied herbivores, but among closely related host species, host diet is the more influential predictor. Table 4.3. Host taxonomy and dietary guild shape the gut microbiotas of African herbivores. Analysis Host factors Bray-Curtis (% variance explained) Jaccard (% variance explained) Weighted Unifrac (% variance explained) 25.90, p=0.001 10.04, p=0.001 p=0.001 18.35, p=0.001 8.23, p=0.001 7.59, p=0.001 Unweighted Unifrac (% variance explained) 24.29, p=0.001 10.30, p=0.001 7.68, p=0.001 16.77, p=0.001 12.16, p=0.001 7.17, p=0.001 20.62, p=0.001 10.17, p=0.001 6.78, p=0.001 9.90, 15.91, p=0.001 13.38, p=0.001 7.03, p=0.001 Across all host study species (11 sp.) (N=165) Across bovids (7 sp.) (N=122) Host family Host dietary guild sample month 22.34, p=0.001 11.20, p=0.001 7.39, p=0.001 Host dietary guild Host species sample month 18.26, p=0.001 15.16, p=0.001 7.56, p=0.001 Shown are the R2 values (% variance explained) and p-values for PERMANOVA tests (y ~ sample month + host dietary guild + host taxonomy) based on 4 types of distance matrices. Bray-Curtis and Weighted Unifrac distance matrices take into consideration the proportions of bacterial taxa, while Jaccard and unweighted Unifrac take into account only their presence or absence. Both Unifrac distances account for phylogenetic relatedness among bacterial types. Significant p-values (α=0.05) are bolded. Aim 2: Evaluate the influences of host taxonomy and host dietary guild on gut microbiota composition and diversity Here, we compared gut microbiota taxonomic composition among the different hosts to identify the bacterial taxa that were characteristic of particular host families or dietary guilds. We also examine the extent to which host taxonomy (family or species) and dietary guild were associated with gut microbiota alpha-diversity. Microbiota composition. Our analyses showed that some bacterial taxa were widely shared among host species and dietary guilds, whereas others were abundant only in particular host species. All herbivore gut microbiotas were dominated by two bacterial 163 phyla, Firmicutes (51% average relative abundance across samples), and Bacteroidetes (32%) (Figure S4.1). The most abundant bacterial families were Ruminococcaceae (30.8%), Rikenellaceae (11.4%), Lachnospiraceae (10.9%), and Prevotellaceae (8%) (Figure 4.2A). Prevalent bacterial genera included Alistipes, Bacteroides, Ruminococcus, and Treponema (Figure S4.2). Only 10 out of 11,930 (0.08%) Amplicon Sequence Variants (ASVs) were present in 90% of samples pooled across all host species; 7 were assigned to the family Ruminococcaceae, 1 to Peptococcaceae, and 2 to Lachnospiraceae (Agathobacter). According to a blast search against the NCBI nucleotide database, sequences from the 7 Ruminococcaceae ASVs were highly similar to sequences from uncultured Ruminococcaceae strains, uncultured rumen bacteria, and uncultured anaerobic bacteria. Of these 10 ASVs, only two were abundant across samples (e.g., ASV10543 Ruminococcaceae), five were modestly abundant in specific host species (e.g., ASV71 Agathobacter in elephants), and 3 were present at very low abundances in all samples (e.g., ASV7824 Peptococcaceae) (Figure S4.3). The latter 3 ASVs do not appear to represent contamination introduced during DNA extraction and sequencing, as these sequences are highly similar to those found in rumen and fecal samples. Nonetheless, variation in gut microbiota compositions among host families and dietary guilds was evident. Indicator species analysis showed that the gut microbiotas of elephants were significantly associated with Endomicrobiaceae and Desulfobulbaceae, those of zebras with Helicobacteraceae and Deltaproteobacteria, and those of warthogs with Myxococcales and Coxiellaceae (Figure 4.2B). Giraffe gut microbiotas were highly associated with Enterobacteriaceae, Bifidobacteriaceae, and Bacillaceae. No bacterial taxa were strongly and specifically associated with Bovid hosts. Furthermore, there were bacterial types that were indicative of specific dietary guilds. Grazer gut microbiotas were characterized by Sphingobacteriaceae, Flavobacteriaceae, Neisseriaceae, and 164 Lentisphaeria (Figure 4.2C). Browser gut microbiotas had 11 indicator bacterial taxa, including Bacillaceae, Coriobacteriales, Methanomicrobia and Rubrobacteriaceae. Lastly, the gut microbiotas of mixed feeders were highly associated with Synergistaceae, Succinivibrionaceae, and Bacteroidales, among other bacteria. More fine-scale analysis of the presence and absence of bacterial ASVs also revealed that the gut microbiotas of our studied herbivores contained microbes that were biased towards particular host species. These were bacterial ASVs that were present in 75% of samples for that host species, and absent in 97% of samples from other hosts. The gut microbiotas of buffalo, cattle, topi, and impala mostly contained ASVs that were present in other herbivores, as <3% of their ASVs were biased towards any of these host species. Between 4% and 8% of ASVs comprising the gut microbiota of dik-diks, eland, elephant, Thompson’s gazelle, and giraffe were biased towards these host species. Warthogs and zebras however, harbored more unusual microbiotas, as 70-77% of their ASVs were rarely detected in the guts of the other surveyed African mammals. 165 A) B) C) Figure 4.2 Gut microbiota composition of African herbivores. A) Stacked bar plots showing the relative frequency of 16S rRNA gene sequences assigned to each bacterial family (or order, if a family- level classification could not be assigned) across samples. Samples are grouped by host species, and each color represents a bacterial family. B) Bacterial families significantly associated with particular herbivore families as determined by indicator species analysis. Differences in these taxa abundances can explain differences in the microbiota among the different groups. Note how no bacterial taxa were associated with Bovid hosts. C) Bacterial families significantly associated with herbivores from different dietary guilds as determined by indicator species analysis. Microbiota alpha-diversity : Gut microbiota richness, evenness, and phylogenetic diversity also varied with host taxonomy (family and species) and dietary guild (Table 4.4, Figure 4.3). Post-hoc comparisons revealed that hosts from the Suidae and Elephantidae 166 families generally harbored less rich and less even gut microbiotas than the other surveyed host families (Figure 4.3A; Table S4.3). Moreover, equids harbored more phylogenetically diverse (PD) gut communities than all other herbivores (Table S4.3). Across the three alpha-diversity metrics, browsers had less diverse gut microbiotas than grazers or mixed- feeders (Figure 4.3A, Table S4.4). Similar tests that included host species in lieu of host family indicated that host species was a strong predictor of gut microbiota alpha-diversity (Chao1 c2=134.42, Shannon diversity c2=45.86, PD c2=74.31). Host dietary guild was also associated with microbiota alpha-diversity, but with a lower effect size (Chao1 c2=16.86, Shannon diversity c2=14.26, PD c2=5.85; all p<0.001). Post-hoc comparisons indicated that warthogs, giraffes, and elephants harbored less rich and phylogenetically diverse gut microbiotas than most bovids, whereas zebras harbored the richest gut microbiotas of the host species surveyed (Table S4.5). Not as many species-specific differences were observed when examining microbiota evenness (Table S4.5). Within a single host family (i.e. Bovidae), the gut microbiotas of buffalo, dikdiks, and gazelles were less rich, even, and phylogenetically diverse than those of all other sampled bovids (Figure 4.3B, Table S4.6). In these bovid hosts, browsers also had less diverse gut microbiotas than grazers or mixed-feeders (Figure 4.3B, Table S4.4). Additionally, we determined whether similarity in gut microbiota alpha-diversity was correlated with dietary similarity (%C4 grasses in diet), while accounting for variation in the gut microbiota associated with host phylogenetic relatedness. Gut microbiota evenness (Shannon diversity) increased with host dietary similarity, but this was not true of gut microbiota richness or phylogenetic diversity (partial Mantel Chao1 r=0.03, p=0.34; Shannon r=0.35, p=0.028, PD r=0.097, p=0.25). 167 Model Factor Across all study sp. (11 sp.) (N=161) Within bovids (7 sp.) (N=118) Host family Host dietary guild Host species Host dietary guild Chao 1 Richness c2=53.58, p<0.001 c2=79.02 p<0.001 c2=57.03 p<0.001 c2=19.55 p<0.0001 Shannon diversity c2=33.45 p<0.001 c2=73.72 p<0.001 c2=10.81 p=0.02 c2=33.91 p<0.0001 Phylogenetic diversity c2=18.31 p=0.001 c2=52.61 P<0.001 c2=50.53 p<0.001 c2=10.67 p<0.01 Table 4.4. Microbiota richness, evenness, and phylogenetic diversity vary with host taxonomy and dietary guild. Shown are the likelihood ratio c2 test statistics obtained for linear mixed effects models specifying host dietary guild and host family as predictor variables, sample date as a random effect, and an alpha- diversity metric as a dependent variable. A similar model restricted to bovids was also constructed; it specified host species instead of host family. Significant p-values (α=0.05) are bolded. A) B) A C B AC B A B A AC AC A AC A C B A B A Figure 4.3 Host taxonomy and dietary guild are associated with gut microbiota diversity in African herbivores. A) Boxplots of microbiota evenness (Shannon diversity) among host families and dietary guilds across all studied herbivores, and B) among host species and dietary guilds within the family Bovidae. Boxes that do not share any letters represent statistically significant comparisons; see Tables S4.3-S4.6 for all post-hoc comparisons. Thicker dots represent outlier values. 168 Aim 3: Examine the amount of variance in the gut microbiota explained by geographic region among conspecific hosts Lastly, we further analyzed the influences of host phylogenetic relatedness and host ecology (i.e., diet and geography) on the gut microbiota of 8 herbivore species from two distinct populations in the Masai Mara (this study) and Laikipia (Kartzinel et al. [22]). The eight herbivore species overlapping both studies were African buffalo, domestic cattle, common eland, impala, giraffe, plains zebra, common warthog, and African elephant. Four of these 8 species were bovids (buffalo, cattle, eland, and impala). We found that gut microbiota structure differed little between conspecific hosts from the two geographic regions of Kenya, as this factor accounted for <3% of the variation in the gut microbiota, on average (PERMANOVA analyses, Table S4.7). The gut microbiotas were primarily structured by host species and host dietary guild, which explained on average, 38% and 11% of the variation, respectively (PERMANOVA analyses, Table S4.7); sample month explained an additional 7% of the variation. Ordination plots confirm these findings and demonstrate that samples primarily cluster by host species (Figure 4.4A), although some separation of samples based on host geographic region is also observed, particularly among cattle, impala, and giraffe. Patterns consistent with phylosymbiosis were also observed in this combined dataset, despite herbivore hosts occupying habitats in Kenya separated by over 300 km, and representing distinct populations. Gut microbiota similarity increased with host phylogenetic relatedness even after accounting for variation attributable to host diet (%C4 grasses) (Table S4.8), although the strength of the phylosymbiosis signal (r=0.62) was less than that obtained for Masai Mara herbivores only (r=0.74), or the value that was previously reported for the Laikipia herbivores (r=0.91) [22]. No relationship was observed between gut microbiota similarity and dietary similarity after controlling for variation due to host phylogeny. Among the four species of bovids that overlapped between 169 the two studies, neither host phylogenetic relatedness nor host diet (%C4) significantly predicted gut microbiota similarity (Table S4.8). The latter findings should be interpreted with caution, as only 4 host species were sampled, whereas the same analyses had a larger sample size (7 host species) when conducted solely on our Masai Mara dataset. In this combined dataset, only 3 of 18,039 (0.01%) ASVs were present across 90% of samples; two were classified as Ruminococcaceae, and 1 as Lachnospiraceae. All 3 ASVs were among the 10 ASVs also present across 90% of Masai Mara samples. To further compare the gut microbiotas of conspecific hosts, we visualized the relative abundances of the 32 most abundant ASVs in the dataset. A heatmap of these 32 ASVs demonstrate that there were ASVs that were found in comparable proportions in conspecific hosts, as well as ASVs that were differentially abundant among conspecific hosts (Figure 4.4B). For example, ASV4033 Prevotellaceae is similarly abundant between buffalo, cattle, and giraffe in the Masai Mara and Laikipia, but ASV15828 Kiritimatiellae appears to be enriched in cattle from Laikipia compared to cattle from the Masai Mara. 170 A) C) Buffalo Cattle Eland Impala Giraffe Warthog Zebra Elephant B) Figure 4.4 The gut microbiotas of conspecific African herbivores broadly converge, but also exhibit differences in their ASV abundances. We compared the gut microbiotas of eight species of herbivores residing in both the Masai Mara (this study) and Laikipia (Kartzinel et al. 2019) regions in Kenya. A) PCoA plot constructed from Bray-Curtis dissimilarity matrices. Each point represents a sample and is color-coded by host species; shape shading indicates geographic region. B) Heatmap of the 32 most abundant bacterial ASVs residing in the gut microbiotas of Masai Mara and Laikipia herbivores. Samples are grouped by host species, and are color-coded by host geographic region. C) ASVs enriched in Masai Mara vs. Laikipia herbivores as determined by LEfSe. Each dot represents a unique ASV and is color-coded by host geographic region. A total of 212 ASVs are displayed (those with LDA >3.2) and their family or genus level classification are on the x-axis. To further extend these analyses and identify additional ASVs that may be differentially abundant among conspecifics (e.g., Masai Mara elephants vs. Laikipia elephants), we also conducted Linear Discriminant Analysis Effect Size (LEfSe) for each 171 host species. Roughly 30% of the ASVs in the gut microbiotas of each host species were enriched in hosts from one population relative to the other (Table S4.9). ASVs that were typically enriched were classified as Ruminococcaceae, Lachnospiraceae, Rikenellaceae, Clostridiales, Bacteroidales, and Kiritimatiellae (Figure 4.4C). Herbivore species from Laikipia tended to be enriched in Methanocorpusculum, Clostridiales, and Kiritimatiellae ASVs relative to Masai Mara herbivores (Figure 4.4C). Masai Mara gut microbiotas were overrepresented by Lachnospiraceae and Treponema ASVs. Interestingly, hosts from both geographic regions could be enriched in taxonomically similar ASVs. For example, eland in Laikipia were enriched in 3 ASVs classified as Prevotellaceae, Rikenellaceae, and Ruminococcaceae, respectively, and eland from the Masai Mara were enriched in 3 different ASVs that were also classified as Prevotellaceae, Rikenellaceae, and Ruminococcaceae (Figure 4.4C). These findings suggest that variation in the gut microbiotas of these herbivore conspecifics is observable at the level of specificity of bacterial ASVs. Discussion Principal findings of study The primary purpose of this study was to determine the relative contributions of host phylogenetic relatedness and dietary guild in structuring the gut microbiotas of 11 species of sympatric African herbivores. We also compared the gut microbiotas of herbivores from the Masai Mara to herbivores from Laikipia, Kenya to determine the extent to which two distinct populations of identical herbivore species varied in their gut microbiotas. We found that gut microbiotas were highly species-specific, but also varied with host ecology, including host diet and sample month, particularly among closely related Bovid species. Furthermore, gut microbiota similarity increased with host phylogenetic relatedness at a relatively broad host 172 taxonomic scale, but at a lesser taxonomic scale, host diet (%C4 grasses) was the strongest predictor of gut microbiota similarity. Lastly, although the gut microbiotas of conspecific herbivore hosts converged and primarily clustered by host species, variation among conspecifics in the relative abundances of their bacterial ASVs were also observed. Collectively, our findings suggest that mammalian gut microbiotas are strongly shaped by host phylogenetic relatedness and taxonomy, but they can be further modified by host ecology, including host diet and geography. Aim 1: Determine the strongest predictor of gut microbiota similarity among herbivore hosts at greater and lesser taxonomic scales Our results showed that phylosymbiosis was observed across the relatively broad host taxonomic scale encompassing multiple herbivore families, i.e. among 11 species of herbivores living sympatrically in the Masai Mara. Patterns of phylosymbiosis have been documented extensively in many vertebrate groups, including primates, rodents, ruminants, carnivores, reptiles, and insects [9–12, 22, 62, 63]. Evidence of phylosymbiosis among host species living in sympatry specifically, has been previously documented in seven species of deer mice [64], six species of Malagasy mammals [65], twelve species of lemurs [66], and nine species of diurnal, non-human primates [67]. The mechanisms and processes that yield patterns of phylosymbiosis have not yet been elucidated, but host ecological and phenotypic traits are likely acting as filters and thus shaping microbial community assembly. Closely related hosts are potentially colonized by taxonomically similar microbes due to similarities in their morphology, anatomy, digestive physiologies, and immune system components [68–70]. Specifically, related hosts may possess similar antimicrobial peptides and toll-like receptors that serve to filter the same bacterial clades from the environment [71, 72]. Closely related hosts may further develop 173 immune tolerance via adaptive immunity to the same symbiotic, commensal, and transient microbes [71, 72]. Lastly, some closely related hosts may also possess similar social group structures and pathways for transmitting microbes among group-mates, thereby contributing to patterns of phylosymbiosis [73–76]. Overall, accumulation of differences in traits as hosts diverged from one another could potentially provide enough niche differentiation in the gut to promote the divergence of their symbiotic bacterial communities. At a lower host taxonomic scale, within our sampled group of closely related Bovid species in the Masai Mara, variation in the gut microbiota was more strongly associated with host diet than host species, and we did not detect a pattern congruent with phylosymbiosis with this dataset. Similarly, other studies report that among closely related hosts, host ecology more strongly predicts the structure of the gut microbiota than host relatedness. For example, in lemurs (Eulemur spp., Propithecus spp.), phylosymbiosis was observed across but not within two host lineages, and within host lineages, host habitat (dry forest vs. rainforest) was significantly correlated with gut microbiota diversity [66]. In populations of yellow (Papio cynecephalus) and anubis baboons (Papio anubis), gut microbiota dissimilarity did not increase with host genetic distance, but did vary with their habitat’s soil chemistry [77]. Because the bovids surveyed here are closely related, their gut microbiotas are already very similar, and variation can result from fine-scale differences in diet (proportions of grass vs. shrubs vs. trees consumed) [78–81]. Nonetheless, some of the variation in the gut microbiota of bovids is not attributable to host diet and can be explained by host phylogenetic relatedness. Thus, even among closely related host species, both ecological and evolutionary forces shape gut microbial communities. 174 Aim 2: Evaluate the influences of host taxonomy and host dietary guild on gut microbiota composition and diversity Across the surveyed herbivores, gut microbiota composition, diversity, and structure varied with host taxonomy. Species-specificity of the gut microbiota is widespread, and is commonly reported in the majority of comparative gut microbiome studies [11, 82–84]. Host species may vary in their body size, behavior, neuroendocrine system, immune function, and metabolism, any of which could potentially influence the structure of their gut microbiotas [69, 70, 85, 86]. When comparing gut microbiota alpha-diversity, results showed that warthogs and elephants harbored less diverse gut communities than did the other sampled herbivores. Due to their omnivory, warthogs have a greater dietary breadth than the other studied herbivores, yet they harbored less diverse microbiotas. This is in accordance with prior findings, which report that the most diverse diets do not always correlate with the most diverse gut microbiotas [22, 87, 88]. Furthermore, analyses showed that browsers had the least diverse gut microbiotas, potentially because they consume vegetation that has a higher lignin content and a lower fiber digestibility than grass [80]. Specialized bacterial metabolisms may be required to digest this tougher plant material. Additionally, group size has been shown to correlate with gut microbiota diversity [89, 90], and the browsers in our study (giraffes, dikdiks) typically forage in smaller groups than do grazers (buffalo, zebras) and mixed-feeders (gazelles, impala), which forage in herds. Frequent social interactions and interactions with a greater number of individuals is known to promote species richness in individual gut microbiotas [90, 91]. Similar to findings from a plethora of microbiome studies, the gut microbiota structure of the studied herbivores also varied with host diet. Hosts from each dietary guild consume food sources that vary in their structure, chemistry, and nutrition quality; these require morphological, physiological and behavioral adaptations [92, 93]. For example, 175 grazers mostly feed on grasses, which have thicker cell walls, a lower protein content, and use the C4 photosynthetic pathway compared to the leaves, shrubs, and woody vegetation consumed by browsers, which have a higher protein content, and use C3 photosynthesis [92, 93]. To efficiently extract energy from these different food sources, browsers and grazers evolved adaptations in their salivary chemistry, tooth morphology, gut structure, and speed of digestion [94, 95]. These adaptations, along with the actual nutrients hosts are providing to their microbes, potentially contribute to gut microbiota divergence among hosts from different dietary guilds. Despite differences in the gut microbiota among host species and dietary guilds, there were some features of the gut microbiota that were shared across individuals from multiple species. Across our surveyed herbivores, the most abundant bacterial taxa in the gut microbiota were Ruminococcaceae, Rikenellaceae, Lachnospiraceae, and Prevotellaceae which represent core taxa previously found in the gut microbiotas of many ruminants and herbivores in general, including cervids and bovids [8, 96], equids [97], elephants [98], and giraffes [99]. Ruminococcaceae and Lachnospiraceae have also been found in the guts of folivorous primates [3] and in domestic pigs [100, 101]. Members of these bacterial families are responsible for digesting the cellulose, hemicellulose, lignin, and protein found in vegetation, and fermenting these into short-chain fatty acids (SCFAs) [102]. SCFAs represent usable forms of energy for the hosts [103] and contribute to host colonocyte growth, communication, immune defense, and anti-inflammatory responses [1]. These bacterial taxa also possess fiber-degrading capabilities and can provide their hosts with protection against ingested toxic plant secondary metabolites [104]. Interestingly, 7 of the 10 ASVs that were present in 90% of Masai Mara herbivores were classified as Ruminococcaceae and were sequences highly similar to uncultured Ruminococcaceae strains extracted from bovine, ovine, and caprine rumens [44], suggesting that these “core” 176 microbes may be functionally important for the host, and/or are easily acquired from the environment. Aim 3: Examine the amount of variance in the gut microbiota explained by geographic region among conspecific hosts While gut microbiota structure was primarily associated with host species and phylogeny in the combined Masai Mara and Laikipia dataset, differences in gut microbiota composition between conspecific hosts from the two populations were also evident. Herbivores of the same species may possess similar evolutionary trajectories, physiologies, and behaviors, and thus may be providing microbes with similar niches for colonization, which is why at a broad level their gut microbiotas converge. However, the two geographic regions do vary in their climate, soil geochemistry, plant communities, and resident herbivore species [23–27, 105], and potentially in their bacterial species pools, which could lead to the fine-scale microbiota differences among conspecifics. This finding was supported by our data; according to LEfSe analyses, over 30% of ASVs were differentially enriched between Masai Mara and Laikipia hosts. Laikipia herbivores for example, tended to be enriched in ASVs classified as Methanocorpusculum, Clostridiales, and Kiritimatiellae, while Masai Mara herbivores had an overrepresentation of ASVs belonging to Lachnospiraceae and Treponema. Abundances of the methanogenic Methanocorpusculum are related to forage type and geographic location in cattle [106], and in the mammalian gut, Lachnospiraceae are associated with a high-fat diet [107]. Furthermore, in the bovine rumen, Treponema degrade hemicellulose and their growth increases in the presence of pectin [108, 109], a carbohydrate abundant in non-woody plants. This suggests that differences in the ASV abundances between conspecific hosts likely reflect fine-scale differences in their diets and habitats. Interestingly, both Clostridiales and Lachnospiraceae 177 are major microbial taxa of the mammalian gut and comprise fermentative bacteria that synthesize SCFAs from the hydrolysis of starches and sugars [102]; thus, conspecific hosts can be enriched in taxonomically distinct microbes that perform similar functions. Future studies should examine whether phylosymbiosis is evident at the functional level in the gut metagenomes of African herbivores and in metazoan taxa in general. A recent study by Milani and colleagues reports that gut microbiome functional profiles varied with host dietary category (carnivore, piscivore, herbivore) across 24 species of mammals [110]. These gut microbiome functional profiles might also vary with host phylogeny. Such studies will be necessary to further our understanding of the processes and mechanisms potentially underlying patterns of phylosymbiosis. Lastly, we found that phylosymbiosis was also evident among conspecific African herbivores living in allopatry, although the strength of the phylosymbiotic signal was slightly reduced compared to that observed for either sympatric population considered in isolation. Overlap in gut microbiota structure is thought to be lower in allopatric animal populations than in sympatric animals due to variation introduced by habitat, dietary differences, and the spatial limits of bacterial dispersal [12]. It is important to note that differences between the gut microbiotas of Laikipia and Masai Mara conspecifics could also be potentially attributable to differences in sampling, DNA extraction, and sequencing protocols between the two studies [111–113]. Collectively, our findings show that mammalian gut microbiotas converge among closely related host species and among conspecifics, but can be differentiated with variation introduced by the host’s ecology. Conclusions Our study showed that among 11 species of African herbivores living in sympatry, gut microbiotas are highly species-specific and exhibit patterns congruent with 178 phylosymbiosis. However, these gut microbiotas are also shaped by their host’s ecology, and within closely related bovid host species, gut microbiota similarity is strongly predicted by host diet (%C4 grasses in diet and dietary guild) and is not associated with host phylogenetic relatedness. Furthermore, among eight species of herbivores residing in two geographic regions in Kenya, gut microbiotas were similar among hosts of the same species, but also exhibited fine scale differences in the abundances of their bacterial ASVs. Overall, these findings suggest that related hosts are providing microbes with similar niches for colonization, but these microbial niches are further shaped by host diet, geography, and local environmental conditions. 179 APPENDICES 180 APPENDIX A: SUPPLEMENTAL TABLES Table S4.1 Dietary C4 (%) data for the 11 herbivore species included in this study. As no single study had %C4 data for all of our herbivore species, we compiled values from multiple studies. These studies used stable isotope analysis of bones, teeth, and hair (Cerling et al. 2003), or feces (all other studies) to estimate dietary %C4. These data were averaged and then used to conduct a series of multivariate Mantel tests (e.g. partial Mantel tests). Although none of the studies estimated %C4 intake for Masai Mara herbivores specifically, the %C4 values for the host species appear consistent across geographic regions. Host species Buffalo Cattle Eland Impala DikDik Gazelle Topi Giraffe Zebra Warthog Elephant East Africa (EA) Cerling et al. 2003 Kartzinel et al. 2015 South Africa (SA) Codron et al. 2007 Codron et. al. 2006 mean %C4 92 87 11 55 2 68 97 5 92 91 41 NA NA NA 65 NA NA NA NA NA NA 50 100 NA 18 52 0 68 100 NA NA NA NA 87 87 NA 41 4 NA NA NA 92 NA 31 88 NA 3 60 NA NA 94 5 92 91 NA Table S4.2 Influences of host species and diet on the gut microbiota across 11 species of herbivores. Shown are the R2 values (% variance explained) and p-values for PERMANOVA tests that tested the influence of sample month, host dietary guild, and host species (in this order) on the gut microbiota of 11 species of herbivores (e.g. across "broad" host taxonomic scales"). The PERMANOVA tests were based on 4 types of distance matrices. Bray-Curtis and Weighted Unifrac distance matrices take into consideration the proportions of bacterial taxa, and both Unifrac distances account for phylogenetic relatedness among bacterial types. Analysis Host factors Across all host study species (11 sp) (N=165) Host species Host diet guild sample month Average (% variance explained) 30.92 10.44 7.93 Bray- Curtis (% variance explained) 31.75, p=0.001 11.20, p=0.001 Jaccard Weighted Unifrac Unweighted Unifrac (% variance explained) 29.23, p=0.001 10.17, p=0.001 (% variance explained) 30.50, p=0.001 10.04, p=0.001 (% variance explained) 31.53, p=0.001 10.36, p=0.001 7.39, p=0.001 6.78, p=0.001 9.90, p=0.001 7.68, p=0.001 181 Table S4.3 Multiple-comparison testing of gut microbiota alpha-diversity among host families. Shown are the estimate, std. error, and Benjamini-Hochberg adjusted p-values of multiple comparison testing to identify which host families varied from one another in terms of their gut microbiota diversity. A)Chao Richness, B)Shannon diversity index, C)Faith’s phylogenetic diversity. Comparisons are read from left to right. A) Chao 1 Richness Giraffidae Suidae Equidae 182 Bovidae Giraffidae Suidae Equidae B) Shannon Diversity Bovidae Giraffidae Suidae 1007.27 -123.22 132.40 166.00 (p<0.0001)*** (p=0.44) 1130.49 211.33 (p<0.0001)*** Giraffidae Suidae -0.32 0.15 (p=0.05)* 0.43 0.19 (p=0.045)* 0.75 0.25 (p=0.006)** Giraffidae Suidae Equidae C) Faith’s Phylogenetic Diversity Bovidae Giraffidae Suidae -2.20 2.83 (p=0.43) 8.08 3.55 (p=0.03)* 10.29 4.52 (p=0.03)* Equidae -67.19 197.62 (p=0.81) 56.03 236.92 (p=0.81) -1074.46 251.61 (p<0.0001)*** Elephantidae 698.57 141.00 (p<0.0001)*** 821.80 190.35 (p<0.0001)*** -308.69 196.99 (p=0.16) 765.76 244.15 (p<0.002)* Equidae -0.46 0.23 (p=0.06) -0.13 0.28 (p=0.62) -0.89 0.30 (p=0.007)** Elephantidae 0.80 0.16 (p<0.0001)*** 1.13 0.22 (p<0.0001)*** 0.37 0.23 (p=0.11) 1.27 0.29 (p<0.0001)*** Equidae -14.40 4.25 (p=0.002)** -12.19 5.09 (p=0.03)* -22.49 5.42 (p=0.0003)*** Elephantidae 3.28 3.01 (p=0.30) 5.48 4.04 (p=0.25) -4.80 4.21 (p=0.30) 17.68 5.23 (p=0.002)** All herbivores grazer-browser grazer-mixed feeder mixed feeder- browser Bovids only grazer-browser grazer-mixed feeder mixed feeder- browser Table S4.4 Multiple-comparison testing of gut microbiota alpha-diversity among host dietary guilds. Shown are the estimate, std. error, and Benjamini-Hochberg adjusted p-values of multiple comparison testing to identify which host dietary guilds vary from one another in terms of their gut microbiota diversity. Comparisons are read from left to right. Chao1 Richness Shannon Diversity Phylogenetic Diversity 871.24 100.89 (p<0.001)*** 155.64 92.51 (p=0.09) 715.61 104.65 (p<0.001)** 492.9 115.3 (p<0.0001)*** 179.5 134.3 (p=0.18) 313.4 121.4 (p=0.014)* 0.95 0.12 (p<0.001)*** 0.014 0.10 (p=0.89) 0.93 0.12 (p<0.0001)*** 0.78 0.14 (p<0.0001)*** 0.10 0.17 (p=0.53) 0.67 0.15 (p<0.001)** 15.08 2.16 (p<0.0001)*** 2.06 1.98 (p=0.29) 13.01 2.24 (p<0.001)*** 7.5 2.39 (p=0.005)** 2.53 2.80 (p=0.36) 4.98 2.54 (p=0.07)* 183 Table S4.5 Multiple-comparison testing of gut microbiota alpha-diversity among host species (all study species). Shown are the estimate, std. error, and Benjamini-Hochberg adjusted p-values of multiple comparison testing to identify which host species varied from one another in terms of their gut microbiota diversity. All host species (N=11) were included in the analysis. A)Chao Richness, B)Shannon diversity index, C)Faith’s phylogenetic diversity. Comparisons are read from left to right. Accompanying linear models are in the manuscript text. A) Chao Richness Topi -523.19 125.08 (p=0.001)** 118.21 135.96 (p=0.99) Cattle -641.39 138.97 (p<0.001)*** Buffalo Eland -106.94 159.39 (p=0.99) 534.45 167.21 (p=0.05)* 416.25 155.74 (p=0.20) Impala -567.26 124.16 (p<0.001)*** 74.13 135.29 (p=0.99) -44.08 120.91 (p=0.99) -460.32 155.50 (p=0.09) Gazelle 170.71 133.02 (p=0.96) 812.10 143.80 (p<0.001)*** 693.90 129.71 (p<0.001)*** 277.65 162.24 (p=0.002)** 737.97 128.61 (p<0.001)*** Dikdik 482.59 114.48 (p=0.001)** 1123.99 124.80 (p<0.001)*** 1005.78 110.68 (p<0.001)*** 520.08 157.33 (p=0.03)* 1049.85 109.38 (p<0.001)*** 311.88 120.04 (p=0.23) Giraffe 413.14 127.31 (p=0.043)** 1054.53 138.15 (p<0.001)*** 936.32 123.48 (p<0.001)*** -352.84 207.38 (p=0.82) 980.40 123.69 (p<0.001)*** 242.43 132.44 (p=0.75) -69.45 112.51 (p=0.99) Zebra -459.78 187.16 (p=0.31) 181.61 194.95 (p=0.99) 63.40 185.01 (p=0.99) 887.32 182.03 (p<0.001)*** 107.48 184.99 (p=0.99) -630.49 190.76 (p=0.03)* -942.38 178.18 (p=0.001)*** -872.92 186.58 (p=0.001)*** Warthog 780.38 159.20 (p<0.001)*** 1421.78 166.86 p(<0.001)*** 1303.57 156.73 (p<0.001)*** 628.61 167.74 (p=0.007)** 1347.64 156.34 (p<0.001)*** 609.67 163.24 (p=0.008)** 297.79 148.38 (p=0.63) 367.24 158.51 (p=0.40) 1240.17 207.51 (p<0.001)*** Elephant 521.67 139.40 (p<0.001)** 1163.07 149.54 (p<0.001)*** 1044.86 136.42 (p<0.001)*** 737.97 128.61 (p<0.001)*** 1088.94 136.27 (p<0.001)*** 350.96 143.89 (p=0.32) 39.08 126.91 (p=0.99) 108.54 137.99 (p=0.99) 981.46 195.38 (p<0.001)*** -258.71 168.52 (p=0.90) Cattle Topi Eland Impala Gazelle Dikdik Giraffe Zebra Warthog B) Shannon Diversity Cattle -0.21 0.18 (p=0.98) Topi -0.28 0.16 (p=0.82) -0.07 0.18 (p=0.99) Buffalo Cattle Topi Eland Impala Gazelle Dikdik Giraffe Zebra Warthog Eland -0.12 0.21 (p=0.99) 0.08 0.22 (p=0.99) 0.16 0.21 (p=0.99) Impala -0.35 0.16 (p=0.53) -0.14 0.17 (p=0.85) -0.06 0.16 (p=0.99) -0.22 0.20 (p=0.99) Gazelle 0.10 0.18 (p=0.53) 0.31 0.19 (p=0.85) 0.39 0.17 (p=0.47) 0.23 0.22 (p=0.99) 0.46 0.17 (p=0.21) Dikdik 0.78 0.15 (p<0.01)** 0.99 0.16 (p<0.01)** 1.07 0.15 (p<0.01)** 0.91 0.19 (p<0.01)** 1.14 0.14 (p<0.01)** 0.68 0.16 (p<0.01)* Giraffe 0.47 0.16 (p=0.15) 0.68 0.18 (p<0.01)** 0.75 0.16 (p<0.01)*** 0.59 0.21 (p=0.14) 0.82 0.16 (p<0.01)** 0.36 0.17 (p=0.60) -0.31 0.15 (p=0.56) Zebra -0.63 0.25 (p=0.28) -0.42 0.26 (p=0.86) -0.34 0.25 (p=0.94) -0.50 0.28 (p=0.77) -0.28 0.24 (p=0.98) -0.74 0.25 (p=0.12) -1.42 0.24 (p=0.01)** -1.10 0.25 (p=0.01)** Warthog 0.28 0.21 (p=0.95) 0.49 0.22 (p=0.47) 0.57 0.21 (p=0.18) 0.41 0.24 (p=0.84) 0.64 0.20 (p=0.07) -0.17 0.22 (p=0.99) -0.50 0.20 (p=0.28) -0.18 0.21 (p=0.99) 0.92 0.28 (p=0.04)* Elephant 0.66 0.18 (p=0.015)* 0.87 0.20 (p<0.01)** 0.95 0.18 (p<0.01)** 0.79 0.22 (p=0.02)* 1.02 0.18 (p<0.01)** 0.56 0.19 (p=0.13) -0.12 0.17 (p=0.99) 0.19 0.18 (p=0.99) 1.30 0.26 (p<0.01)** 0.38 0.22 (p=0.84) 184 Table S4.5 (cont’d) C) Faith’s Phylogenetic Diversity Cattle -24.75 5.36 (p<0.01)** Topi -18.57 4.84 (p=0<0.01)** 6.17 5.25 (p=0.98) Eland -12.19 6.13 (p=0.64) 12.56 6.47 (p=0.67) 6.36 6.04 (p=0.99) Buffalo Cattle Topi Eland Impala Gazelle Dikdik Giraffe Zebra Impala -20.09 4.79 (p<0.01)** 4.66 5.21 (p=0.99) -1.53 4.68 (p=0.99) -7.89 6.01 (p=0.96) Gazelle 4.06 5.15 (p=0.99) 28.81 5.56 (p<0.01)** 22.62 5.03 (p=<0.01)** 16.25 6.29 (p=0.24) 24.15 4.98 (p<0.01)** Dikdik 13.24 4.42 (p=0.09) 37.99 4.83 (p<0.01)** 31.79 4.29 (p<0.01)** 25.43 5.70 (p<0.01)** 33.33 4.23 (p<0.01)** 9.17 4.85 (p<0.65)* Giraffe 10.92 4.91 (p=0.47) 35.77 5.32 (p<0.01)** 29.48 4.78 (p<0.01)*** 23.11 6.09 (p<0.01)** 31.04 4.76 (p<0.01)** 6.85 5.12 (p=0.95) -2.31 4.35 (p=0.99) Zebra -40.34 7.25 (p<0.01)** -15.59 7.5 (p=0.58) 21.79 7.18 (p=0.08) -28.15 8.05 (p=0.019)* -20.25 7.16 (p=0.14) -44.41 7.43 (p<0.01)** -53.58 6.91 (p<0.01)** -51.27 7.23 (p=0.01)** Warthog 2.46 6.16 (p=0.99) 27.21 6.46 (p<0.01)** 21.08 6.07 (p=0.02)* 14.65 7.07 (p=0.58) 22.55 6.04 (p<0.01)** -1.60 6.32 (p=0.99) -10.77 5.74 (p=0.72) -8.46 6.13 (p=0.95) 42.81 7.58 (p<0.01)* Elephant 1.02 5.4 (p=0.99) 25.77 5.78 (p<0.01)** 19.58 5.29 (p<0.01)** 13.21 6.50 (p=0.61) 21.15 5.27 (p<0.01)** -3.04 5.58 (p=0.99) -12.21 4.92 (p=0.30) -9.89 5.34 (p=0.73) 41.37 7.58 (p<0.01)** -1.43 6.53 (p=0.99) Warthog Table S4.6 Multiple-comparison testing of gut microbiota alpha-diversity among host species in bovids. Shown are the estimate, std. error, and Benjamini-Hochberg adjusted p-values of multiple comparison testing to identify which host bovid species varied from one another in terms of their gut microbiota diversity. A)Chao Richness, B)Shannon diversity index, C)Faith’s phylogenetic diversity. Comparisons are read from left to right. A) Chao 1 Richness Cattle -617.32 139.52 (p<0.001)*** Buffalo Topi -511.75 126.20 (p=0.0001)*** 105.56 136.86 (p=0.51) Eland -104.09 159.87 p=0.56) 513.22 168.51 (p=0.004)** 407.66 157.54 (p=0.013)* Impala -551.52 124.73 (p<0.001)*** 65.79 135.63 (p=0.65) -39.77 121.84 (p=0.74) -447.43 156.73 (p=0.006)** Gazelle 179.49 134.33 0.22 796.80 144.76 (p<0.0001)*** 691.24 131.44 (p<0.0001)*** 283.58 164.16 (p=0.11) 731.01 129.86 (p<0.0001)*** Dikdik 492.86 115.29 (p<0.001)*** 1110.17 125.78 p(<0.0001)*** 1004.61 111.98 (p<0.0001)*** 596.95 148.72 (p=0.0001)*** 1044.38 110.10 (p<0.0001)*** 313.37 121.39 0.013* Cattle Topi Eland Impala Gazelle 185 Table S4.6 (cont’d) B) Shannon Diversity Cattle -0.21 0.17 (p=0.41) Buffalo Cattle Topi Eland Impala Gazelle Topi -0.29 0.16 (p=0.15) 0.08 0.17 (p=0.69) Eland -0.12 0.20 (p=0.63) -0.08 0.21 (p=0.69) -0.16 0.20 (p=0.55) Impala -0.35 0.16 (p=0.06) -0.14 0.17 (p=0.55) -0.06 0.15 (p=0.69) -0.22 0.20 (p=0.41) Gazelle 0.10 0.17 (p=0.63) 0.32 0.18 (p=0.16) 0.40 0.17 (p=0.05)* 0.23 0.21 (p=0.41) 0.46 0.16 (p=0.019)* Dikdik 0.78 0.14 (p<0.0001)*** 0.99 0.16 (p<0.0001)*** 1.07 0.14 (p<0.0001)*** 0.91 0.19 (p<0.0001)*** 1.13 0.14 (p<0.0001)*** 0.67 0.15 (p<0.0001)*** Dikdik 7.52 2.39 (p=0.003) ** 20.34 2.61 (p<0.0001 )*** 17.92 2.34 (p<0.0001 )*** 12.95 3.11 (p<0.0001 )*** 18.87 2.28 (p<0.0001 )*** 4.98 2.54 (p=0.07) C) Faith’s Phylogenetic Diversity Topi Cattle Eland Impala Gazelle -12.82 2.87 (p<0.0001)*** -10.40 2.63 (p<0.0001)*** -5.43 3.33 (p=0.13) -11.35 2.57 (p<0.001)*** 2.53 2.80 (p=0.42) Buffalo Cattle Topi Eland Impala Gazelle 2.42 2.83 (p=0.43) 7.39 3.5 (p=0.05) 1.47 2.78 (p=0.62) 4.97 3.30 (p=0.16) -0.95 2.53 (p=0.70) 15.36 3.01 (p<0.0001)*** 12.93 2.76 (p0.0001)*** -5.92 3.25 (p=0.09) 7.96 3.44 (p=0.03)* 13.89 2.71 (p<0.0001)*** 186 Host Factors Table S4.7 Influences of host species and diet on the gut microbiota remain despite differences in host geography. Shown are the R2 values (% variance explained) and p-values for PERMANOVA tests that tested the influence of sample month, host geographic region (Masai Mara vs. Laikipia), host dietary guild, and host species (in this order) on the gut microbiota of 8 species of herbivores that overlapped our study and Kartzinel et al 2019. The PERMANOVA tests were based on 4 types of distance matrices. Bray-Curtis and Weighted Unifrac distance matrices take into consideration the proportions of bacterial taxa, and both Unifrac distances account for phylogenetic relatedness among bacterial types . Table S4.8. The relative contributions of host phylogenetic relatedness and diet in predicting gut microbiota similarity in Masai Mara and Laikipia herbivores. Shown are the rho, test statistic, and p- values associated with partial correlation coefficient tests that evaluated the correlation between 2 variables (e.g. gut microbiome similarity and phylogenetic relatedness, while controlling for a third (e.g. dietary similarity). The tests were conducted on 4 types of gut microbiome distance metrics, and significant p-values are bolded. The data from our study (Masai Mara) and Kartzinel et al. 2019 (Laikipia) were combined for this analysis. Bray-Curtis (% variance explained) 34.52, p=0.001 12.53, p=0.001 6.69 p=0.001 1.6, p=0.001 UW Unifrac (% variance explained) 43.84, p=0.001 11.31, p=0.001 6.20, p=0.001 1.52, p=0.001 W Unifrac (% variance explained) 40.02, p=0.001 7.80, p=0.001 11.65, p=0.001 2.36, p=0.001 Host species Host dietary guild Sample month Jaccard (% variance explained) 34.43, p=0.001 12.03, p=0.001 5.69, p=0.001 1.30, p=0.001 Host region Model Across all host study species (8 sp) Metric Bray-Curtis Jaccard Unifrac (weighted) Unifrac (unweighted) Dietary Similarity (% C4) R Phylogenetic Relatedness R Z stat p.val 0.6 3.78 0.64 4.25 0.61 3.9 <0.0001 0.0002 <0.001 <0.0001 -0.09 -0.04 Z stat p.val 0.001 0.99 0.63 -0.47 -0.24 0.8 0.65 4.33 <0.001 -0.1 -0.52 0.6 Across bovids (4 sp) Bray-Curtis Jaccard Unifrac (weighted) Unifrac (unweighted) 0.36 0.68 2.56 0 - -1.07 0.52 0.36 0.68 0.54 1 0.36 0.54 0.57 0.82 -0.1 0.57 1.2 0 -0.17 1.2 0.31 0.08 0.87 0.31 187 Table S4.9 Percent of ASVs differentially abundant between Masai Mara and Laikipia herbivore populations. Shown are the proportion of ASVs for each host species that are enriched in Masai Mara or Laikipia populations as determined by LEfSe (default parameters; LDA >2, α=0.05). ASVs with <0.01% relative abundances across samples were filtered from dataset prior to LEfSe. A LEfSe test was ran for each host species (8 tests in total). To calculate the % of enriched ASVs, we divided the # of statistically- significant ASVs (p-values <0.05 after Bonferroni-adjustment) by the total # of ASVs in the data frame. The raw LEfSe output can be downloaded from the GitHub repository for this project. ASVs enriched in Masai Mara ASVs enriched in Laikipia combined % enriched (%) (%) ASVs Host species Buffalo Cattle Eland Impala Giraffe Warthog Zebra Elephant # ASVs in data frame 1242 1491 1436 1518 1142 758 1537 1548 13.36 15.62 11.76 18.18 19.08 20.05 4.29 8.07 17.71 29.44 13.37 24.76 30.29 11.47 12.36 30.1 31.07 45.06 25.13 42.94 49.37 31.52 16.65 38.17 Table S4.10 Sample distribution for the combined Masai Mara and Laikipia dataset. Months considered part of the rainy season in each geographic region are highlighted as green [1–3], and the number of samples for each month is also shown. Rains in the Masai Mara follow a bimodal pattern, while those in Laikipia follow a trimodal pattern. Masai Mara Laikipia Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 19 42 21 19 48 90 14 41 7 188 APPENDIX B: SUPPLEMENTAL FIGURES Figure S4.1 Predominant bacterial phyla of African herbivore gut microbiotas. Stacked bar plots showing the relative percentage of 16S rRNA gene sequences assigned to each bacterial phylum across samples. Samples are grouped by host species, and each color represents a bacterial phylum. Buffalo Cattle Topi Eland Impala Gazelle Dikdik Giraffe Zebra Warthog Elephant Bacterial phylum Bacteroidetes Firmicutes Other Proteobacteria Spirochaetes Tenericutes Verrucomicrobia ) % ( e c n a d n u b A e v l i t a e R 100 75 50 25 0 189 Figure S4.2 Predominant bacterial genera of African herbivore gut microbiotas. Stacked bar plots showing the relative frequency of 16S rRNA gene sequences assigned to each bacterial genus across samples. Samples are grouped by host species, and each color represents a bacterial genus. Buffalo Cattle Topi Eland Impala Gazelle Dikdik Giraffe Zebra Warthog Elephant ) % ( e c n a d n u b A e v i t a l e R 100 75 50 25 0 Bacterial genus Akkermansia Alistipes Bacteroidales Bacteroides Christensenellaceae_R7 Clostridiales Kiritimatiellae_WCHB1−41 Mollicutes_RF39 Other Prevotellaceae_UCG−004 Rikenellaceae_RC9 Ruminococcaceae_NK4A214 Ruminococcaceae_UCG−005 Ruminococcaceae_UCG−010 Ruminococcaceae_UCG−013 Ruminococcaceae_UCG−014 Ruminococcus_1 Treponema_2 Figure S4.3 Relative abundance of 10 ASVs widely shared across host species. Heatmap showing the relative abundances (proportions) of 10 ASVs that were present in over 90% of the samples included in this study. Samples are grouped and color-coded by host species. Darker colors in the heatmap indicate higher relative abundances. unclass_Bacteroidales_RF16 unclass_Lachnospiraceae unclass_Planococcaceae unclass_Prevotellaceae unclass_Ruminococcaceae Buffalo Cattle Topi Eland Impala Gazelle Dikdik Giraffe ZebraWarthogElephant 190 Figure S4.4 Sample distribution by Month for Masai Mara herbivores. The total number of samples collected for each host species at each sampling month is shown. s e l p m a S f o r e b m u N 80 60 40 20 0 March April May June Host species Buffalo Cattle Topi Eland Impala Gazelle Dikdik Giraffe Zebra Warthog Elephant Figure S4.5 Rarefaction curves of gut microbiota ASV richness. Plotted are the number of ASVs (ASV Richness) that are recovered with an increasing number of sequences, after subsampling to 17,000 sequences/sample. Each curve represents a unique sample and is color-coded by host species. The y and x-axis are scaled equally to show what the curves would look if with each read came a new ASV. The curves all plateau. 191 REFERENCES 192 REFERENCES 1. Jovel J, Dieleman LA, Kao D, Mason AL, Wine E. The Human Gut Microbiome in Health and Disease. Metagenomics. 2018;:197–213. doi:10.1016/B978-0-08-102268-9.00010-0. 2. Ungerfeld E, Leigh M, Forster R, Barboza P. Influence of Season and Diet on Fiber Digestion and Bacterial Community Structure in the Rumen of Muskoxen (Ovibos moschatus). Microorganisms. 2018;6:89. doi:10.3390/microorganisms6030089. 3. Greene LK, McKenney EA, O’Connell TM, Drea CM. The critical role of dietary foliage in maintaining the gut microbiome and metabolome of folivorous sifakas. Sci Rep. 2018;8:1–13. 4. den Besten G, van Eunen K, Groen A, Venema K, Reijngoud D, Bakker B. The role of short- chain fatty acids in the between diet, gut microbiota, and host energy metabolism. J Lipid Res. 2013;54:2325–40. doi:10.1194/jlr.R036012. 5. Park W. Gut microbiomes and their metabolites shape human and animal health. J Microbiol. 2018;56:151–3. doi:10.1007/s12275-018-0577-8. 6. Münger E, Montiel-Castro AJ, Langhans W, Pacheco-López G. Reciprocal Interactions Between Gut Microbiota and Host Social Behavior. Front Integr Neurosci. 2018;12:1–14. doi:10.3389/fnint.2018.00021. 7. Pennisi E. Gut bacteria linked to mental well-being and depression. Science (80- ). 2019;363:569–569. doi:10.1126/science.363.6427.569. 8. Li J, Zhan S, Liu X, Lin Q, Jiang J, Li X. Divergence of Fecal Microbiota and Their Associations With Host Phylogeny in Cervinae. Front Microbiol. 2018;9:1–11. doi:10.3389/fmicb.2018.01823. 9. Amato KR, Sanders J, Song SJ, Nute M, Metcalf JL, Thompson LR, et al. Evolutionary trends in host physiology outweigh dietary niche in structuring primate gut microbiomes. ISME J. 2019;13:576–587. doi:10.1038/s41396-018-0175-0. 10. Nishida AH, Ochman H. Rates of gut microbiome divergence in mammals. Mol Ecol. 2018;27:1884–97. 11. Youngblut ND, Reischer GH, Walters W, Schuster N, Walzer C, Stalder G, et al. Host diet and evolutionary history explain different aspects of gut microbiome diversity among vertebrate clades. Nat Commun. 2019;10:1–15. doi:10.1038/s41467-019-10191-3. 12. Moeller AH, Suzuki TA, Lin D, Lacey EA, Wasser SK, Nachman MW. Dispersal limitation promotes the diversification of the mammalian gut microbiota. PNAS. 2017;114:13768–73. doi:10.1073/pnas.1700122114. 13. Kohl KD. Ecological and evolutionary mechanisms underlying patterns of phylosymbiosis in host-associated microbial communities. Philos Trans R Soc B Biol Sci. 2020;375. 14. Lim SJ, Bordenstein SR. An introduction to phylosymbiosis. Proc R Soc B Biol Sci. 2020;287. 193 15. Brooks AW, Kohl KD, Brucker RM, Van Opstal EJ, Bordenstein SR. Phylosymbiosis: Relationships and Functional Effects of Microbial Communities across Host Evolutionary History. PLoS Biol. 2016;14:1–19. doi:10.1371/journal.pbio.2000225. 16. Watson SE, Hauffe HC, Bull M, McKinney MA, Atwood TA, Perkins SE. Global change- driven use of onshore habitat impacts polar bear faecal microbiota. ISME J. 2019. doi:10.1038/s41396-019-0480-2. 17. Angelakis E, Yasir M, Bachar D, Azhar EI, Lagier JC, Bibi F, et al. Gut microbiome and dietary patterns in different Saudi populations and monkeys. Sci Rep. 2016;6:1–9. doi:10.1038/srep32191. 18. Goertz S, Menezes AB De, Birtles RJ, Id JF, Lowe E, Maccoll ADC, et al. Geographical location influences the composition of the gut microbiota in wild house mice ( Mus musculus domesticus ) at a fine spatial scale. PLoS One. 2019;14:1–16. 19. Orkin JD, Campos FA, Guadamuz A, Melin AD, Myers MS, Hernandez SEC. Seasonality of the gut microbiota of free-ranging white-faced capuchins in a tropical dry forest. ISME J. 2019;13:183–196. doi:10.1038/s41396-018-0256-0. 20. Knowles SCL, Eccles RM, Baltrūnaitė L. Species identity dominates over environment in shaping the microbiota of small mammals. Ecol Lett. 2019;22:826–37. 21. Kohl KD, Varner J, Wilkening JL, Dearing MD. Gut microbial communities of American pikas (Ochotona princeps): Evidence for phylosymbiosis and adaptations to novel diets. J Anim Ecol. 2018;87:323–30. 22. Kartzinel TR, Hsing JC, Musili PM, Brown BRP, Pringle RM. Covariation of diet and gut microbiome in African megafauna. PNAS. 2019;116:1–6. 23. Ottichilo WK, De Leeuw J, Skidmore AK, Prins HHT, Said MY. Population trends of large non-migratory wild herbivores and livestock in the Masai Mara ecosystem, Kenya, between 1977 and 1997. Afr J Ecol. 2000;38:202–16. 24. Goheen JR, Augustine DJ, Veblen KE, Kimuyu DM, Palmer TM, Porensky LM, et al. Conservation lessons from large-mammal manipulations in East African savannas: The KLEE, UHURU, and GLADE experiments. Ann N Y Acad Sci. 2018;1429:31–49. 25. Veldhuis MP, Ritchie ME, Ogutu JO, Morrison TA, Beale CM, Estes AB, et al. Cross- boundary human impacts compromise the Serengeti-Mara ecosystem. Science (80- ). 2019;363:1424–8. doi:10.1126/science.aav0564. 26. Augustine DJ. Response of native ungulates to drought in semi-arid Kenyan rangeland. Afr J Ecol. 2010;48:1009–20. 27. Broten MD, Said M. Population Trends of Ungulates in and around Kenya’s Masai Mara Reserve. In: Serengeti II: Dynamics, Management, and Conservation of an Ecosystem. 1995. p. 169–93. 28. Green DS, Zipkin EF, Incorvaia DC, Holekamp KE. Long-term ecological changes influence herbivore diversity and abundance inside a protected area in the Mara-Serengeti ecosystem. 194 Glob Ecol Conserv. 2019;20:e00697. doi:10.1016/j.gecco.2019.e00697. 29. Kingdon J. East African Mammals: Volume I. Chicago: The University of Chicago Press; 1984. 30. Kingdon J. East African Mammals: Volume IIIC. Chicago: The University of Chicago Press; 1982. 31. Kingdon J. East African Mammals: Volume IIID. Chicago: The University of Chicago Press; 1982. 32. Kingdon J. East African Mammals: Volume IIIB. Chicago: The University of Chicago Press; 1979. 33. Cerling TE, Harris JM, Passey BH. Diets of East African Bovidae based on stable isotope analysis. J Mammal. 2003;84:456–70. 34. Kartzinel TR, Chen PA, Coverdale TC, Erickson DL, Kress WJ, Kuzmina ML, et al. DNA metabarcoding illuminates dietary niche partitioning by African large herbivores. PNAS. 2015;112:8019–24. 35. Codron D, Codron J, Lee-Thorp JA, Sponheimer M, De Ruiter D, Sealy J, et al. Diets of savanna ungulates from stable carbon isotope composition of faeces. J Zool. 2007;273:21–9. 36. Codron J, Lee-Thorp JA, Sponheimer M, Codron D, Grant R, De Ruiter D. Elephant (Loxodonta Africana) diets in Kruger National Park, South Africa: Spatial and landscape differences. J Mammal. 2006;87:27–34. 37. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, et al. Ultra-high- throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 2012;6:1621–4. doi:10.1038/ismej.2012.8. 38. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. Development of a Dual- Index Sequencing Strategy and Curation Pipeline for Analyzing Amplicon Sequence Data on the MiSeq Illumina Sequencing Platform. Appl Environ Microbiol. 2013;79:5112–20. doi:10.1128/AEM.01043-13. 39. R Core Team. R: a language and environment for statistical computing. In: R Foundation for Statistical Computing. Vienna, Austria; 2019. https://www.gbif.org/tool/81287/r-a-language-and- environment-for-statistical-computing. Accessed 28 Aug 2019. 40. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: High- resolution sample inference from Illumina amplicon data. Nat Methods. 2016;13:581–3. 41. Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 2013;41:D590-6. doi:10.1093/nar/gks1219. 42. Wickham H. ggplot2: Elegant Graphics for Dat Analysis. New York, NY: Springer-Verlag New York; 2009. doi:10.1007/978-0-387-98141-3. 195 43. Kolde R. Pheatmap: Pretty Heatmaps. R package version 1.0.12. 2019;:1–8. https://cran.r- project.org/package=pheatmap. 44. Coordinators NR. Database resources of the National Center for Biotechnology Information. Nucleic Acids Res. 2018;46. doi:10.1093/nar/gkx1095. 45. De Cáceres M, Legendre P. Associations between species and groups of sites: Indices and statistical inference. Ecology. 2009;90:3566–74. 46. Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol. 2009;75:7537–41. doi:10.1128/AEM.01541-09. 47. Jin L, Wu D, Li C, Zhang A, Xiong Y, Wei R, et al. Bamboo nutrients and microbiome affect gut microbiome of giant panda. Symbio. 2020;80:293–304. 48. Bo T-B, Zhang X-Y, Wen J, Deng K, Qin X-W, Wang D-H. The microbiota-gut-brain interaction in regulating host metabolic adaptation to cold in male Brandt’s voles (Lasiopodomys brandtii). ISME J. 2019. doi:10.1038/s41396-019-0492-y. 49. McMurdie PJ, Holmes S. Phyloseq: An R Package for Reproducible Interactive Analysis and Graphics of Microbiome Census Data. PLoS One. 2013;8:1–11. 50. Schliep KP. phangorn: Phylogenetic analysis in R. Bioinformatics. 2011;27:592–3. 51. Kembel SW, Cowan PD, Helmus MR, Cornwell WK, Morlon H, Ackerly DD, et al. Picante: R tools for integrating phylogenies and ecology. Bioinformatics. 2010;26:1463–4. 52. Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw. 2015;67:1–48. doi:10.18637/jss.v067.i01. 53. Fox J, Weisberg S, Fox J. An R companion to applied regression. Thousand Oaks, CA: Sage; 2011. 54. Oksanen J, Blanchet F, Friendly M, Kindt R, Legendre P, McGlinn D, et al. vegan: Community Ecology Package. R Packag version 24-6. 2018. https://cran.r- project.org/web/packages/vegan/index.html. Accessed 28 Aug 2019. 55. Upham N, Esselstyn J, Jetz W. Inferring the mammal tree: species-level sets of phylogenies for questions in ecology, evolution, and conservation. PLoS Biol. 2019;17:e3000494. 56. Paradis E, Claude J, Strimmer K. APE: Analyses of phylogenetics and evolution in R language. Bioinformatics. 2004;20:289–90. 57. Kim S. ppcor: An R Package for a Fast Calculation to Semi-partial Correlation Coefficients. Commun Stat Appl Methods. 2015;22:665–74. 58. Ochieng EO. Characterizing the spatial distributions of elephants in Mpala, Kenya. 2015. 59. Kartzinel TR, Pringle RM. Multiple dimensions of dietary diversity in large mammalian 196 herbivores. J Anim Ecol. 2020;0:1–15. 60. Segata N, Izard J, Waldron L, Gevers D, Miropolsky L, Garrett WS, et al. Metagenomic biomarker discovery and explanation. Genome Biol. 2011;12:R60. doi:10.1186/gb-2011-12-6- r60. 61. Afgan E, Baker D, Batut B, van den Beek M, Bouvier D, Čech M, et al. The Galaxy platform for accessible, reproducible and collaborative biomedical analyses: 2018 update. Nucleic Acids Res. 2018;46:W537–44. doi:10.1093/nar/gky379. 62. Scheelings TF, Moore RJ, Van TTH, Klaassen M, Reina RD. Microbial symbiosis and coevolution of an entire clade of ancient vertebrates: the gut microbiota of sea turtles and its relationship to their phylogenetic history. Anim Microbiome. 2020;2:2–12. 63. Tinker KA, Ottesen EA. Phylosymbiosis across deeply diverging lineages in omnivorous cockroaches. Appl Environ Microbiol. 2020;86:e02513-19. doi:10.1128/AEM.02513-19. 64. Schmidt E, Mykytczuk N, Schulte-Hostedde AI. Effects of the captive and wild environment on diversity of the gut microbiome of deer mice (Peromyscus maniculatus). ISME J. 2019;13:1293–305. doi:10.1038/s41396-019-0345-8. 65. Perofsky AC, Lewis RJ, Meyers LA. Terrestriality and bacterial transfer: a comparative study of gut microbiomes in sympatric Malagasy mammals. ISME J. 2019;13:50–63. doi:10.1038/s41396-018-0251-5. 66. Greene LK, Clayton JB, Rothman RS, Semel BP, Semel MA, Gillespie TR, et al. Local habitat , not phylogenetic relatedness , predicts gut microbiota better within folivorous than frugivorous lemur lineages. Biol Lett. 2019;15:5–11. 67. Gogarten JF, Davies TJ, Benjamino J, Gogarten JP, Graf J, Mielke A, et al. Factors influencing bacterial microbiome composition in a wild non-human primate community in Taï National Park, Côte d’Ivoire. ISME J. 2018;12:2559–74. doi:10.1038/s41396-018-0166-1. 68. Amato KR, Mallott EK, Mcdonald D, Dominy NJ, Goldberg T, Lambert JE, et al. Convergence of human and Old World monkey gut microbiomes demonstrates the importance of human ecology over phylogeny. Genome Biol. 2019;20:201. 69. Groussin M, Mazel F, Sanders JG, Smillie CS, Lavergne S, Thuiller W, et al. Unraveling the processes shaping mammalian gut microbiomes over evolutionary time. Nat Commun. 2017;8:14319. doi:10.1038/ncomms14319. 70. Groussin M, Mazel F, Alm EJ. Co-evolution and Co-speciation of Host-Gut Bacteria Systems. Cell Host Microbe. 2020;28:12–22. doi:10.1016/j.chom.2020.06.013. 71. Gerardo NM, Hoang KL, Stoy KS. Evolution of animal Immunity in the light of beneficial symbioses. Philos Trans R Soc B2. 2020;375:20190601. 72. Popkes M, Valenzano DR. Microbiota-host interactions shape ageing dynamics. Philos Trans R Soc B2. 2020;375:20190596. 73. Perofsky AC, Lewis RJ, Abondano LA, Difiore A, Meyers LA. Hierarchical social networks 197 shape gut microbial composition in wild verreaux’s sifaka. Proc R Soc B Biol Sci. 2017;284. 74. Raulo A, Ruokolainen L, Lane A, Amato K, Knight R, Leigh S, et al. Social behaviour and gut microbiota in red-bellied lemurs ( Eulemur rubriventer ): In search of the role of immunity in the evolution of sociality. J Anim Ecol. 2017;87:1–12. doi:10.1111/1365-2656.12781. 75. Opie C, Atkinson QD, Shultz S. The evolutionary history of primate mating systems. Commun Integr Biol. 2012;5:458–61. 76. Shultz S, Opie C, Atkinson QD. Stepwise evolution of stable sociality in primates. Nature. 2011;479:219–22. doi:10.1038/nature10601. 77. Grieneisen LE, Charpentier MJE, Alberts SC, Blekhman R, Bradburd G, Tung J, et al. Genes, geology and germs: Gut microbiota across a primate hybrid zone are explained by site soil properties, not host species. Proc R Soc B Biol Sci. 2019;286. 78. Muegge BD, Kuczynski J, Kinghts D, Clemente JC, Fontana L, Henrissat B, et al. Diet drives Convergence in Gut Microbiome Functions Across Mammalian Phylogeny and Within Humans. Science (80- ). 2017;332:970–4. 79. Clemens ET, Maloiy GMO. Digestive physiology of East African wild ruminants. Comp Biochem Physiol -- Part A Physiol. 1983;76:319–33. 80. Illius AW, Gordon IJ. The functional significance of the browser-grazer dichotomy in African ruminants. Oecologia. 1994;98:167–75. 81. Pérez-Barbería JF, Gordon IJ, Illius AW. Phylogenetic analysis of stomach adaptation in digestive strategies in African ruminants. Oecologia. 2001;129:498–508. 82. Ley RE, Hamady M, Lozupone C, Turnbaugh PJ, Ramey RR, Bircher JS, et al. Evolution of Mammals and Their Gut Microbes. Science (80- ). 2008;320:1647. 83. McKenzie VJ, Song SJ, Delsuc F, Prest TL, Oliverio AM, Korpita TM, et al. The effects of captivity on the mammalian gut microbiome. Integr Comp Biol. 2017;57:690–704. 84. Phillips CD, Phelan G, Dowd SE, McDonough MM, Ferguson AW, Delton Hanson J, et al. Microbiome analysis among bats describes influences of host phylogeny, life history, physiology and geography. Mol Ecol. 2012;21:2617–27. 85. Godon JJ, Arulazhagan P, Steyer JP, Hamelin J. Vertebrate bacterial gut diversity: Size also matters. BMC Ecol. 2016;16:1–9. 86. Liu P, Cheng A, Huang S, Lu H, Oshida T. Body-size Scaling is Related to Gut Microbial Diversity, Metabolism and Dietary Niche of Arboreal Folivorous Flying Squirrels. Sci Rep. 2020;:1–12. doi:10.1038/s41598-020-64801-y. 87. Budd K, Gunn JC, Eggert LS, Finch T, Klymus K, Sitati N. Effects of diet, habitat, and phylogeny on the fecal microbiome of wild African savanna (Loxodonta africana) and forest elephants (L . cyclotis). Ecol Evol. 2020;00:1–14. 88. Metcalf JL, Song SJ, Morton JT, Weiss S, Seguin-Orlando A, Joly F, et al. Evaluating the 198 impact of domestication and captivity on the horse gut microbiome. Sci Rep. 2017;7. 89. Miller EA, Livermore J, Alberts SC, Tung J, Archie EA. Ovarian cycling and reproductive state shape the vaginal microbiota in wild baboons. Microbiome. 2017;5:1–14. 90. Moeller AHA, Foerster S, Wilson ML, Pusey AE, Hahn BH, Ochman H. Social behavior shapes the chimpanzee pan-microbiome. Sci Adv. 2016;2:e1500997–e1500997. doi:10.1126/sciadv.1500997. 91. Johnson KVA. Gut microbiome composition and diversity are related to human personality traits. Hum Microbiome J. 2020;15 December 2019:100069. doi:10.1016/j.humic.2019.100069. 92. Shipley L. Grazers and browsers: how digestive morphology affects diet selection. Grazing Behav Livest Wildl. 1999;:20–7. http://www.cnr.uidaho.edu/range456/readings/shipley.pdf. 93. Venter JA, Vermeulen MM, Brooke CF. Feeding Ecology of Large Browsing and Grazing Herbivores. In: Gordon IJ, Prins HHT, editors. The Ecology of Browsing and Grazing II. Cham: Springer International Publishing; 2019. p. 127–53. doi:10.1007/978-3-030-25865-8_5. 94. Hofmann RR. Evolutionary steps of ecophysiological adaptation and diversification of ruminants: a comparative view of their digestive system. Oecologia. 1989;78:443–57. doi:10.1007/BF00378733. 95. Spencer LM. Morphological Correlates of Dietary Resource Partitioning in the African Bovidae. J Mammal. 1995;76:448–71. doi:10.2307/1382355. 96. Henderson G, Cox F, Ganesh S, Jonker A, Young W, Janssen PH, et al. Rumen microbial community composition varies with diet and host, but a core microbiome is found across a wide geographical range. Sci Rep. 2015;5:14567. 97. Edwards JE, Shetty SA, Berg P Van Den, Burden F, Doorn DA Van, Pellikaan WF, et al. Multi-kingdom characterization of the core equine fecal microbiota based on multiple equine (sub) species. Anim Microbiome. 2020;2:1–16. 98. Ilmberger N, Güllert S, Dannenberg J, Rabausch U, Torres J, Wemheuer B, et al. A comparative metagenome survey of the fecal microbiota of a breast-and a plant-fed asian elephant reveals an unexpectedly high diversity of glycoside hydrolase family enzymes. PLoS One. 2014;9:1–12. 99. Roggenbuck M, Sauer C, Poulsen M, Bertelsen MF, Sørensen SJ. The giraffe (Giraffa camelopardalis) rumen microbiome. FEMS Microbiol Ecol. 2014;90:237–46. 100. Tan SC, Chong CW, Yap IKS, Thong KL, Teh CSJ. Comparative assessment of faecal microbial composition and metabonome of swine, farmers and human control. Sci Rep. 2020;10:1–13. 101. Zeng B, Zhang S, Xu H, Kong F, Yu X, Wang P, et al. Gut microbiota of Tibetans and Tibetan pigs varies between high and low altitude environments. Microbiol Res. 2020;235 November 2019:126447. doi:10.1016/j.micres.2020.126447. 102. Biddle A, Stewart L, Blanchard J, Leschine S. Untangling the genetic basis of fibrolytic specialization by Lachnospiraceae and Ruminococcaceae in diverse gut communities. Diversity. 199 2013;5. 103. Flint HJ, Scott KP, Duncan SH, Louis P, Forano E. Microbial degradation of complex carbohydrates in the gut. Gut Microbes. 2012; August:289–306. 104. Kohl KD, Denise Dearing M. The woodrat gut microbiota as an experimental system for understanding microbial metabolism of dietary toxins. Front Microbiol. 2016;7:1–9. 105. Shorrocks B. The Biology of African Savannas. New York: Oxford University Press; 2007. 106. Wright ADG, Auckland CH, Lynn DH. Molecular diversity of methanogens in feedlot cattle from Ontario and Prince Edward Island, Canada. Appl Environ Microbiol. 2007;73:4206–10. doi:10.1128/AEM.00103-07. 107. Daniel H, Gholami AM, Berry D, Desmarchelier C, Hahne H, Loh G, et al. High-fat diet alters gut microbiota physiology in mice. ISME J. 2014;8:295–308. doi:10.1038/ismej.2013.155. 108. Liu J, Pu YY, Xie Q, Wang JK, Liu JX. Pectin Induces an In Vitro Rumen Microbial Population Shift Attributed to the Pectinolytic Treponema Group. Curr Microbiol. 2014;70:67–74. doi:10.1007/s00284-014-0672-y. 109. Xie X, Yang C, Guan LL, Wang J, Xue M, Liu JX. Persistence of Cellulolytic Bacteria Fibrobacter and Treponema After Short-Term Corn Stover-Based Dietary Intervention Reveals the Potential to Improve Rumen Fibrolytic Function. Front Microbiol. 2018;9 JUN:1363. doi:10.3389/fmicb.2018.01363. 110. Milani C, Alessandri G, Mancabelli L, Mangifesta M, Lugli GA, Viappiani A, et al. Multi- omics Approaches To Decipher the Impact of Diet and Host Physiology on the Mammalian Gut Microbiome. Appl Environ Microbiol. 2020;86:1–21. 111. Kennedy K, Hall MW, Lynch MDJ, Moreno-Hagelsieb G, Neufeld JD. Evaluating bias of Illumina-based bacterial 16S rRNA gene profiles. Appl Environ Microbiol. 2014;80:5717–22. 112. De La Cuesta-Zuluaga J, Escobar JS. Considerations For Optimizing Microbiome Analysis Using a Marker Gene. Front Nutr. 2016;3:26. doi:10.3389/fnut.2016.00026. 113. Mallott EK, Malhi RS, Amato KR. Assessing the comparability of different DNA extraction and amplification methods in gut microbial community profiling. Access Microbiol. 2019. 200