The Role of Evolution in Maintaining Coexistence of Competitors Abigail I

Florida State University Libraries

Electronic Theses, Treatises and Dissertations The Graduate School

2017 The Role of Evolution in Maintaining Coexistence of Competitors Abigail I. (Abigail Ilona) Pastore

Follow this and additional works at the DigiNole: FSU's Digital Repository. For more information, please contact [email protected] FLORIDA STATE UNIVERSITY

COLLEGE OF ARTS AND SCIENCES

THE ROLE OF EVOLUTION IN MAINTAINING COEXISTENCE OF COMPETITORS

ABIGAIL I. PASTORE

A Dissertation submitted to the Department of Biological Science in partial fulﬁllment of the requirements for the degree of Doctor of Philosophy

2017

Thomas Miller Professor Directing Dissertation

Richard Bertram University Representative

Brian Inouye Committee Member

Scott Steppan Committee Member

Alice Winn Committee Member

The Graduate School has veriﬁed and approved the above-named committee members, and certiﬁes that the dissertation has been approved in accordance with university requirements.

ii For my Mom and for Kristofer Ad astra per aspera

iii ACKNOWLEDGMENTS

Tom Miller can not be thanked enough, he is a beacon of patience, generosity and fun. His intellectual guidance is pervasive through this document and this work should be seen as an extension of the Miller legacy of evolution among competitors. My committee members were aces; Alice Winn’s keen ability to cut straight to the heart of any bullshit and generally bring joviality into my day, Scott Steppan’s expert guidance through a milieu of phylogenetic inference and general supporter of my stage presence. Brian Inouye’s expertise in maximum likelihood estimation and bike mechanic skills were clutch. Thanks to Richard Bertram’s ability to bend my conceptual understanding into dimensions I wasn’t recognizing. Olivia Mason gave me the keys to the black box of bacterial ecology, and patiently helped me turn the locks. Charlotte Lee helped me struggle through coexistence theory and generally expanded my understanding of competition theory. All of the FSU E and E Faculty members have been delightful and intellectually invigorating including Don Levitan, Nora Underwood, David Houle, Kim Hughes, and Janey Wolfe. The E and E graduate students and post docs were the most supportive and fun people I’ve ever been around, particularly thanks to Wilbur Ryan and Andrew Merwin for always being down for a debate or a song. Elise Gornish and Josh Grinath’s guidance in my early years certainly should not go unrecognized. Casey terHorst and Catalina Cuellar-Gempler were also fantastic mentors. When I started grad school I had no idea I’d leave with such a big and wonderful family. But this would not have been possible without my biological family. My mom is the most supportive, loving and generous person I know. My dad instilled a sense of adventure and love of the natural world within me. Veronica is my rock and her seemingly endless support and encouragement were exactly what I needed to keep going more times than I can count. Omi and Opa, Kevin and Solara are my biggest cheer leaders. My aunts, uncles and cousins are just the best, and knowing I had such a loving and accepting family behind me helped me keep striving. I am awestruck that I am so blessed to have two big and wonderful families to support me on this journey. Thanks to Fallon Ringer for helping me develop the coping skills to ﬁnish strong. Undergraduate researchers that were tremendously helpful, fun and insightful include; Henry Gwynn, Pamela Betancourt, Deniece Wade, Tom Thornburg, Kennedy Wohlgemuth and Matthew

iv Green. Peter Adler, Nelson Hairston Jr., Doug Schemske, and Priyanga Amarasekare provided stimulating conversations that improved this work. Funding from the Florida State University Biology Department, Florida State University and the National Science Foundation’s Doctoral Dissertation Improvement Grant. Many thanks to Ruth Didier for troubleshooting ﬂow cytometry methods and the FSU College of Medicine for use of the ﬂow cytometer. Thanks to Sarah Owens and Argonne National Lab for library prep and illumina runs of the 16S bacterial samples.

v TABLE OF CONTENTS

List of Tables ...... viii List of Figures ...... ix Abstract ...... xiv

1 Introduction 1 1.1 Summary of Goals ...... 1 1.2 Background ...... 1 1.3 Outline ...... 4

2 Theoretical Evidence for the Role of Evolution in Competitor Coexistence 7 2.1 Introduction ...... 7 2.2 Model and Methods ...... 8 2.3 Results ...... 11 2.4 Discussion ...... 17 2.5 Conclusions ...... 19

3 Phylogenetic Signal does not Predict Competitive Interactions in Protist Mi- crocosms 20 3.1 Introduction ...... 20 3.2 Methods ...... 22 3.2.1 Protist Cultures ...... 22 3.2.2 Quantifying Competition ...... 23 3.2.3 Quantifying Protist Traits ...... 24 3.2.4 Quantifying Phylogenetic Distance ...... 24 3.3 Results ...... 26 3.4 Discussion ...... 28 3.4.1 Conclusions ...... 36

4 Evidence for the Role of Evolution in Competitor Coexistence in Protist Micro- cosms 37 4.1 Introduction ...... 37 4.2 Methods ...... 39 4.2.1 Bacterial Broth ...... 39 4.2.2 Protist Cultures ...... 39 4.2.3 Preparation of Stock Lines for Experiment ...... 40 4.2.4 Selection Experiment ...... 41 4.2.5 Response Surface Experiments ...... 41 4.2.6 Invasion Experiments ...... 43 4.2.7 Estimating Niche Overlap and Fitness Diﬀerences ...... 43 4.3 Results ...... 43 4.3.1 T etrahymena vs Colpoda ...... 43

vi 4.3.2 T etrahymena vs Maryna ...... 46 4.3.3 Maryna vs P aramecium ...... 51 4.3.4 Invasibility ...... 56 4.4 Discussion ...... 56 4.4.1 Conclusions ...... 59

5 Evolution of Bacterial Consumption in Competing Protists 60 5.1 Introduction ...... 60 5.2 Methods ...... 62 5.2.1 Bacterial Broth ...... 62 5.2.2 Protist Cultures ...... 64 5.2.3 Preparation of Stock Lines for Experiment ...... 64 5.2.4 Selection Experiment ...... 65 5.2.5 Measuring Effect of Protists on the Bacterial Community...... 65 5.2.6 Flow Cytometry ...... 66 5.2.7 Metagenomic Sequencing ...... 66 5.3 Results ...... 68 5.3.1 Effects on Bacterial Abundances ...... 68 5.3.2 Effects on Bacterial Community Composition ...... 68 5.4 Discussion ...... 73

6 Concluding Remarks 78

Bibliography ...... 81 Biographical Sketch ...... 90

vii LIST OF TABLES

3.1 Day 5: Table of interaction coefficients, ICij. Were column species are species, j, which effect the row species, i. Signicant differences between density of species in monocultures vs. competition are indicated. + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001...... 26

3.2 Day 46: Table of interaction coefficients, ICij. Were column species are species, j, which effect the row species, i. Signicant differences between density of species in monocultures vs. competition are indicated. + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001...... 28

4.1 t-statistics and P-values from multiple linear regression of the eﬀect of each species density on the per-capita growth rate of the focal species...... 44

4.2 t-statistics and P-values from multiple linear regression of the eﬀect of each species density on the per-capita growth rate of the focal species...... 46

4.3 t-statistics and P-values from multiple linear regression of the eﬀect of each species density on the per-capita growth rate of the focal species...... 50

5.1 F-statistics and P-values for permanovas of the eﬀects of protists on bacterial composition over time. Time is 0 hours or 9 hours, treatments are bacterial broth, and species lines. After selection, each species has a line selected in monoculture and in pairwise competition...... 70

viii LIST OF FIGURES

1.1 The balance between niche overlap and fitness difference among competitors determines outcomes of competition. Small fitness differences equalize competitors, whereas large niche difference stabilize population dynamics. When niche overlap is large, species must be similar in their fitness in order to coexist. (reproduced from Adler et al. 2007)...... 3

2.1 Three examples of how niche overlap and fitness differences change as a result of selection on the trait values of two competitors. Each point on the graph is a starting condition and vectors indicate the direction and relative magnitude of changes in niche and fitness differences (scaled down for illustration). The gray areas are regions in which one species will be competitively excluded, whereas the white region indicates coexistence between species. Red arrows represent areas where species originally coexisted, but evolution changed the ecological outcome to competitive exclusion. Blue regions represent where species would undergo competitive exclusion if only ecological dynamics were occurring, but species evolved in ways that allowed coexistence before exclusion occurred. The sigma value indicates the breadth of the environmental landscape; species are more likely to evolve to coexist in environments with a broader environmental niche breadth...... 12

2.2 The initial niche overlap of the two species predicts the ﬁnal niche overlap of two competing species (for the parameters shown here: r2 = 0.67, p < 0.001). The grey area represents divergence in the species niches, and the white area represents convergence in the species niches. Where the linear regression intersects the shaded region, no change in niche overlap is predicted as a consequence of evolutionary dynamics, referred to subsequently as the evolutionarily stable value of niche overlap (H2 = 0.1, σ = 0.2)...... 13

2.3 The breadth of the resource environment determines how tightly species will pack after evolving in trait values in response to competition. The vertical dashed line is the trait variance of both species. When intraspeciﬁc variance of species is equal to the breadth of the environment, species maximize niche overlap. A. The evolutionarily stable value of niche overlap (see Fig. 2.1) tends to decrease in broader environments as species can decrease competition and still have access to resources. B. The change in niche overlap between species indicates whether species are converging or diverging in resource use. The grey area represents divergence in the species niches, and the white area represents convergence in the species niches. Each point represent the average tendency of species in an environment to move toward or away from each other in resource use, error bars are standard deviations of 9 runs of the model with diﬀerent initial starting values of species i...... 14

2.4 Changes in fitness differences as a function of the breadth of the environment. Fitness differences were more likely to both increase and decrease in narrower environments.

ix Each point represent the average tendency of species in an environment to change, given diﬀerent initial starting values of species i...... 15

2.5 The relative magnitude of changes in niche overlap and fitness differences as predicted by the breadth of the environment. A. The relative importance of changes in niche overlap and changes in fitness differences due to trait evolution in competitors. B. Absolute value of the average change in niche overlap due to evolution. C. Absolute value of the average change in fitness differences due to evolution. Each point is the average change in niche overlap and fitness difference of 900 runs with varying initial niche overlap and fitness differences (H2 = 0.1)...... 16

3.1 Interaction Coeﬃcients ICij tended to be negative at day 5 and day 46, indicating competition among many species pairs was occurring. Competition was stronger at day 46 than day 5...... 27

3.2 Ultrametric maximum likelihood tree reconstruction of our sequences added to the Gao and Katz 18S alignment, pruned to included only species present in this study. Species abbreviations are given for comparison to Table 3.3 and 3.3...... 29

3.3 With all taxa included, there was no significant relationship between phylogenetic distance and competitive interactions at either (A) 5 days or (B) 46 days in competition. ICij = 0 indicates no effect of interspecific competitors, negative values indicate negative effects on species abundances in the presence of an interspecific competitors, and positive values indicate facilitation. The dashed line represents a marginally positive significant relationship between phylogenetic distance and the interaction coefficient found on day 5...... 30

3.4 With only ciliates (Alveolata), (A) there was no relationship between phylogenetic distance and competitive interactions at 5 days, (B) but at 46 days there was a marginally significant correlation between phylogenetic distance and the strength of competition between heterospecifics, suggesting that species that diverged longer ago have stronger competitive interactions. This could occur if closely related taxa experienced more niche divergence in sympatry. Interpretation of interaction coefficients is explained in Fig. 3.3...... 31

3.5 The size of a protist did not predict its (A) average competitive eﬀect (Day 5: P = 0.268,R2 = 0.134, Day 46: P = 0.156,R2 = 0.263), or (B) average competitive response (Day 5: P = 0.386,R2 = 0.084, Day 46: P = 0.251,R2 = 0.143), at either time point...... 32

3.6 The tolerance of each protist species to stagnation (density in monoculture at day 46/density in monoculture at day 5) did not predict its (A) average competitive eﬀect (Day 5: P = 0.545,R2 = 0.042, Day 46: P = 0.585,R2 = 0.034), or (B) average competitive response (Day 5: P = 0.118,R2 = 0.249, Day 46: P = 0.313,R2 = 0.113), at any either time point...... 33

x 4.1 Example response surface experiment density levels for two species. Dots represent target density levels of each species and asterisks represent the carrying capacity of each species. This experimental design will allow us to parameterize linear models of competition between two species...... 42

4.2 Parameter estimates and 90% conﬁdence intervals generated by bootstrap analysis for a Lotka-Volterra competition model. Colpoda is species 1 and T etrahymena is species 2...... 44

4.3 Estimates of niche overlap and fitness difference before selection and in different selection environments. Error-bars are 90% confidence intervals. B. Negative values indicate a competitive advantage for Colpoda, and positive values indicate a competitive advantage for T etrahymena...... 45

4.4 Parameter estimates and 90% conﬁdence intervals generated by bootstrap analysis for a Lotka-Volterra competition model. Maryna is species 1 and T etrahymena is species 2...... 47

4.5 Parameter estimates and 90% conﬁdence intervals generated by bootstrap analysis for a Lotka-Volterra competition model. Maryna is species 1 and T etrahymena is species 2. B. Negative values indicate a competitive advantage for Maryna, and positive values indicate a competitive advantage for T etrahymena...... 48

4.6 Population dynamics of T etrahymena and Maryna over the selection period. A. T etrahymena density through time. B. Maryna density through time...... 49

4.7 Parameter estimates and 90% conﬁdence intervals generated by bootstrap analysis for a Lotka-Volterra competition model. Maryna is species 1 and P aramecium is species 2...... 50

4.8 Parameter estimates and 90% conﬁdence intervals generated by bootstrap analysis for a Lotka-Volterra competition model. B. Negative values indicate a competitive advantage for Maryna, and positive values indicate a competitive advantage for P aramecium. 52

4.9 Population dynamics of P aramecium and Maryna over the selection period. A. P aramecium density through time. B. Maryna density through time...... 53

4.10 Change in niche overlap and fitness difference in three different pairs of species. Un- shaded areas are where species coexist, and grey areas are where one species will outcompete the other. A. Colpoda andT etrahymena evolving in pairwise mixtures (2 species) or monocultures (1 species). B. Maryna and T etrahymena have changing competitive outcomes as a result of evolution. C. Maryna and P aramecium coexist initally but evolve to exclude each other in both pairwise mixtures and monocultures. 54

4.11 Invasibility analysis of P aramecium and Maryna before and after selection. Protists were selected in either one species mixtures (mono) or two species mixtures (comp). If each species can increase from low density (positive growth rate) in the presence of

xi its competitor, the species theoretically should coexist. If a species’ growth rate is not positive in the presence of the competitor then it is the competitive subordinate. The box-plots indicate medians and 50% of the data, whiskers indicate 95% conﬁdence intervals...... 55

5.1 Number of particles counted in 11 µL by a flow cytometer as a proxy for bacterial cell density in the 9 standardized bacterial broths (RSE), prepared in the same way at different time points. The box-plots indicate medians and 50% of the data, whiskers indicate 95% confidence intervals...... 63

5.2 NMDS of bacterial community composition gives a two-dimensional visual representation of the compositional similarity of the 9 standardized bacterial broths (RSE), prepared in the same way at diﬀerent time points. Here, the minimum stress after 20 runs was 0.1754653 for k =2...... 63

5.3 Change in the density of bacteria approximately 9 and/or 96 hours after protists were added for diﬀerent experimental species pairs in two selection environments. On the x-axis, ’B’ is where no protists were added and serves as a baseline for determining the eﬀects of protists on the bacterial community, otherwise it indicates the selection environment that a species evolved in; either alone (1 sp.) or in pairwise competition (2 sp.). (A) Colpoda with Maryna at 9 hrs. (B) T etrahymena with Colpoda at 9 hrs. (C) Maryna with P aramecium at 9 hours. (D) T etrahymena with Maryna at 96 hours. (E) Maryna with P aramecium at 96 hours. Box-plots are as described in Fig. 5.1...... 69

5.4 NMDS ordinations of the bacterial communities associated with different selection treatments. Ellipses are the 70% confidence intervals for each treatment. Grey ellipses are the bacterial communities at time zero for the different treatments. Light blue and light red ellipses are species that evolved in monocultures. Dark blue and dark red are species that evolved in pairwise competition. (A) Colpoda with Maryna at 9 hrs, stress= 0.1417709. (B) T etrahymena with Colpoda at 9 hrs, stress = 0.1316615. (C) Maryna with T etrahymena at 96 hours, stress = 0.09588695. (D) Maryna with P aramecium at 96 hours stress = 0.08369383...... 71

5.5 Niche evolution is estimated as the difference in the bacterial communities associated with lines selected in monocultures and pairwise competition, scaled by initial differences in the bacterial communities. Error bars are +/− 1 sd. The grey bar indicates the difference between bacterial communities associated with different selected lines neither increased nor decreased with time. two treatments was the same at time zero and 9 and/or 96hours. (A) Colpoda with Maryna at 9 hrs. (B) T etrahymena with Colpoda at 9 hrs. (C) Maryna with T etrahymena at 96 hours. (D) Maryna with P aramecium at 96 hours...... 72

5.6 Niche diﬀerence is estimated as the scaled dissimilarity between species associated bacterial assemb;ages before selection, after selection in monoculture and after selection in pairwise competition. The grey bar indicates no change in similarity of associated

xii bacterial communities between two species after protists consumption has occurred for 9hours and/or 96hours, i.e. equivalent niches. (A) Colpoda with Maryna at 9 hrs. (B) T etrahymena with Colpoda at 9 hrs. (C) Maryna with T etrahymena at 96 hours. (D) Maryna with P aramecium at 96 hours...... 74

xiii ABSTRACT

Species interactions can regulate a population’s density and therefore can act as a selective force on that population. Such evolutionary responses have the potential to feedback and change ecological interactions between species. For species that compete for resources, the interaction between ecological and evolutionary dynamics will regulate the stability of the species interactions, determining whether competing species can coexist. The outcome of competition between species is determined by two factors: (1) niche overlap, or the similarity in how species use resources and are affected by their environment, and (2) fitness differences, or differences in how efficiently each species uses resources in their environment. Decreasing niche overlap will decrease competitive interactions, thereby stabilizing coexistence. Decreasing fitness differences makes species more equal in their competitive abilities, facilitating coexistence. In the absence of evolutionary constraints, both niche overlap and fitness differences among species are subject to change as a consequence of evolution among competitors, and thus ecological dynamics between two species will also change. In this dissertation, I develop a broader understanding of (1) how niche overlap and fitness differences between species change after evolution in response to competition, (2) how changes in niche overlap and fitness differences are mediated through changes in resource use of protists, and (3) what role evolutionary history plays in shaping ecological and evolutionary dynamics. I address these goals with a suite of approaches including theoretical models, an experimental lab system, and comparative methods. I constructed a quantitative genetic model of trait evolution, where the trait of a species determined its resource use, and found that species are prone to change in their niche overlap as well as their fitness differences as a result of trait evolution. However, the magnitude of changes in niche overlap and fitness differences were determined by the resource availability within the environments. When resources were broadly available, species changed more in their niche overlap, whereas when resources were narrowly available, species changed more in their fitness difference. To test these predictions, I developed a system in the laboratory where protists competed for a bacterial resource. Species were allowed to evolve in either monoculture or a two-species mixture; the effects of evolution on competition, niche overlap and fitness differences were quantified using parameterized models. In general I found that species tended to converge in their niche as a result of evolution, however, changes in fitness differences between species were

xiv larger and more influential on coexistence than changes in niche differences. Both increases in niche overlap, and increases in fitness differences decreased coexistence among species pairs. By describing the bacterial communities associated with these protists before and after selection I determined that protists tended to converge or not change in which bacteria they were consuming as a result of selection. Additionally, for eleven protist species, I determined whether traits or relatedness predicted competitive ability by placing species on a molecular phylogeny and conducting pairwise competition experiments for all pairs. I found no correlations, suggesting neither traits, nor evolutionary history was informative for explaining current ecological and evolutionary interactions in this deeply divergent clade. There are two major conclusions from this dissertation: (1) when species evolve in response to competition, changes in fitness differences may often be more important than changes in niche overlap, (2) evolution can, and may be likely to, decrease the ability of species to coexist through increases in niche overlap and increases in fitness differences. This work suggests that one must simultaneously consider the role of evolutionary and ecological processes to understand community processes. Specifically, when researchers are attempting to explain mechanisms of coexistence between species, they must consider how evolutionary dynamics may change the ecological interactions within communities of competitors.

xv CHAPTER 1

INTRODUCTION

1.1 Summary of Goals

Ecologists and evolutionary biologists share the goal of explaining the diversity found on earth. While evolutionary biologists investigate the origins of diversity, ecologists ask how it is maintained. Currently ecologists are beginning to consider the effects evolution has on the dynamics of natural systems through study of feedback loops between ecology and evolution (Becks et al. 2012). Theory has recently emerged that frames coexistence between species as a balance between niche overlap between competing species and differences in the ecological fitnesses of these species (Chesson 2000). But ecologists have barely begun to explore evolutionary dynamics in this context (Lankau 2011). Past theoretical work has assumed competing species will evolve to differentiate in their niches, but changes in fitness differences are likely to occur with evolution and have ecological consequences as well. Notably, understanding evolutionary trajectories will reveal when interactions between species will lead to coexistence even when the ecological dynamics predict competitive exclusion, and vice versa. The goal of this research is to understand the mechanisms behind the evolutionary and ecological effects of competitive interactions.

1.2 Background

The interplay between ecological and evolutionary forces has always been of interest to ecologists (Hutchinson 1965), but has recently been the subject of much discussion (e.g. Kokko and L`opez-Sepulcre 2007, Becks et al. 2012). ’Eco-evo feedbacks’ are particularly interesting when considering species that compete for resources because competitive interactions that suppress a species population size will act as a selective force, which can change the ecological interactions of competing species. As such, coevolution between species that originally cannot coexist may result in coexistence between species (and vice-versa). Theoretical work on the coevolution among resource competitors has been a common theme of ecology (e.g. MacArthur and Levins 1967, Rummel and Roughgarden 1985, Abrams 1987, Fox and Vasseur 2008 etc). In general, this theory predicts that

1 species will evolve to use different resources and avoid competition. Only in environments with ex- treme constraints on resource use will species evolve to use the same resources, thereby undergoing convergence in the niche (Fox and Vasseur 2008, terHorst et al 2010). Yet, few empirical studies have tested how species will evolve in competition (but see terHorst et al. 2010, and Miller et al. 2014). Theory has also shown that niche differentiation is not the only factor that is important in determining coexistence between species. Chesson (2000) has shown that coexistence between species depends on two important factors; niche overlap between species as well as the ecological fitness differences between species. For example, if two species have complete niche overlap, both species can persist as long they are equally fit in consuming those resources, but species with relatively small niche overlap may not coexist if one is far superior at consuming shared resources (Fig. 1.1). Furthermore, it has been shown in annual plant communities that fitness differences as well as niche overlap are important for determining coexistence between species in nature (Levine and Hille Ris Lambers 2009). In order to have a holistic view of the factors that lead to coexistence or extinction between species it is essential to consider how species evolution will change competitive interactions. For example, two species may not appear to coexist at a given time, but as selection on species changes their traits, they may evolve to stable coexistence (e.g. Gomulkeiwicz and Holt 1995). A key to understanding how species’ ecological dynamics change as a result of evolution is parsing out how competitors evolve in not only niche overlap but also fitness differences (Lankau 2011). Current theory does not explicitly address how fitness differences between species are expected to evolve in conjunction with niche overlap, nor are there any experimental demonstrations that distinguish between changes in niche overlap and fitness differences in evolving species. An exciting prospect in community ecology is that evolutionary history may be predictive of contemporary ecological and evolutionary dynamics among competitors (e.g. Webb 2000). As a result of a shared evolutionary history between close relatives, species’ traits will be correlated with phylogenetic distance; therefore phylogenetic distance should be a good indicator of ecological similarity among species (Harvey and Pagel 1991). As a corollary, closely related and therefore ecologically similar species have strong competitive interactions. Interestingly, closely related species that have similar niches will be more likely to exclude each other, whereas closely related species that are similar in ecological fitness will be more likely to coexist (Mayfield and Levine 2010). Thus

2 Figure 1.1: The balance between niche overlap and fitness difference among competitors determines outcomes of competition. Small fitness differences equalize competitors, whereas large niche difference stabilize population dynamics. When niche overlap is large, species must be similar in their fitness in order to coexist. (reproduced from Adler et al. 2007).

3 there must be a complex relationship between evolutionary history, current ecology and future evolution.

1.3 Outline

In this work, I attempt to thoroughly investigate how evolution due to competition between species will change competitive outcomes. I use protist microcosms as an experimental system, combined with parameterized population models, phylogenetic techniques, and the development of new theory to understand the relationship between species evolutionary history, current ecology and future evolution. I approach these goals from the perspective of coexistence theory and I attempt to quantify two between-species measures of competition, niche overlap between competing species and differences in the ecological fitness of these species, which ultimately predict the outcome of competition between species pairs (Chesson 2000). The idea of the ’niche’ is a central component of this research because species often experience competition through their shared resource base. Protists in laboratory microcosms consume bacteria as their primary resource in an otherwise homogeneous environment, and thus competition for bacteria will drive species interactions and subsequent evolution. In chapter 2, I develop theoretical models of evolution among competitors that incorporate niche overlap and fitness differences. I combine a classical ecological model (Lotka-Volterra) with new techniques in modeling quantitative genetic evolution (Screiber et al 2011). The model developed is quite general, and calls to mind classic theory of the evolution of niche partitioning and limiting similarity (MacArthur and Levins 1967, Dieckman and Doebli 1999). I then employ Chesson’s method (2013) for decomposing the Lotka-Volterra model into niche overlap and fitness difference to determine how these two components of coexistence change due to the selective forces of competition. The model predicts that in general, niche overlap will decrease and fitness differences will increase, but these two consequences have opposite effects on coexistence! Thus, the exciting aspect of this model is that it predicts that, depending on environmental conditions, sometimes changes in niche overlap are more important for ecological outcomes, but sometimes changes in fitness differences are more important. In chapter 3, I investigate the role of evolutionary history in determining contemporary competitive interactions. I hypothesized that species niche use would be evolutionarily conserved and that

4 phylogeny could be used in place of ecologically informative traits to predict competitive outcomes, and subsequent evolutionary outcomes. However, when I allowed species to compete for resources, I found no relationships between competitive interactions, phylogenetic distance or even species traits. This is perhaps a consequence of the deeply divergent group of organisms I used, and serves to remind us that one must be very careful in considering the phylogenetic scale when attempting to make inferences from evolutionary history. In chapter 4, I used experimental evolution in the lab to determine how selection in interspecific competition changed competitive interactions and subsequently coexistence between species pairs. The state of the art method for determining to what degree species are coexisting due to niche and fitness differences is combining response surface experiments with competition model parameterization (as in Law and Watkinson 1987, Inouye 2001). I performed such model parameterization before and after selection experiments to determine how niche and fitness differences between species change due to evolution. Surprisingly, I saw little change or increases in niche overlap between species, concurrent with larger increases in fitness differences. Both components decrease the probability of species coexistence. In chapter 5, I attempted to understand changes in niche overlap and fitness differences from the previous chapter more mechanistically by taking a closer look at how protist affected their bacterial resource base. As a proxy for a protists niche, I used high-throughput sequencing to quantify the composition of bacterial communities to determine how resource use changes as a result of evolution in competition. Additionally, I looked at the protists’ effect on the abundance of bacteria as a proxy for their fitness. In line with the results from chapter 4, niche overlap (similarity in protists effect on bacterial composition) tended to not change or to decrease. However, the effect of protists on the abundance of bacteria evolved in some cases but not others. Taken together, this work begins to tell a story of evolution among competitors that is a bit different from the well-recognized outcome of character displacement. Evolution seems to be more fickle than making room for everyone. There are four possible scenarios (1) niche overlap decreases and fitness differences decrease - increasing coexistence, (2) niche overlap increases and fitness differences increase -decreasing coexistence, and (3 and 4) two ambiguous cases where niche and fitness have opposite effects, and the relative magnitude of change in the two will determine effects on coexistence. It seems that the historical expectation that character displacement will be the

5 most likely evolutionary outcome may have biased our understanding of competitive outcomes, and equally likely may be situations in which species evolutionary responses to competition result in extinction.

6 CHAPTER 2

THEORETICAL EVIDENCE FOR THE ROLE OF EVOLUTION IN COMPETITOR COEXISTENCE

2.1 Introduction

A wealth of theoretical work has investigated the consequences of competition for the evolution of species traits and, proximally, stable coexistence. Much of this literature predicts character displacement as a major outcome of species competition, which will result in niche partitioning and subsequently the ability of both species to increase from low density in the presence of the other competitor (i.e. stable coexistence) (e.g. MacArthur and Levins 1967, Rummel and Roughgarden 1985, Abrams 1983). However, the full suite of possible outcomes of selection on competitors are diverse and strongly dependent on the assumptions of the theoretical framework. A surprising outcome of several models is coexistence or long term cooccurrence arising as a consequence of convergence in resource use when niche space is limited, because in these situations species can only converge or face extinction (Abrams 1986, terHorst et al. 2010). Such convergence may result in ecologically equivalent or ’neutral’ species (sensu Hubbell 1979). Theory shows that coexistence depends not just on niche overlap between species, but also on similarity in the average fitness of the competing populations (Chesson 2000). In this context, niche differences is the degree to which species limit their own populations more than the populations of their competitors. Therefore niche overlap can be thought of as how similarily species are affected by the qualities of their environment either through resource use, predation or climatic constraints. In contrast, the average fitness differences between two species populations is the difference in how efficiently species convert resources to growth in their average environment and can be thought of as the relative competitive ability of two species, even when they do not share resources (Adler et al. 2007). Both niche and fitness differences have consequences for coexistence between species. For example, species that are similar in resource use may not be ecologically equivalent if they have large differences in their average fitness, thus resulting in unstable dynamics that quickly lead to exclusion. Convergence in trait values can only result in coexistence if species also have similar

7 fitness. Conversely, species that have large differences in their niches, may experience competitive exclusion if they also have large disparity in their fitness. It is unknown how evolution among competitors will result in changes to ecological coexistence when species can evolve both in their niche and their fitness simultaneously, especially if niche and fitness must necessarily covary as a consequence of environmental constraints. This is a crucial link in understanding the dynamics and structure of real communities (Lankau 2011). Further- more, for over 15 years, there has been debate over whether population densities are regulated by niche partitioning or stochastic neutral forces (Hubbell 2001, Loreau 2004). Although consensus is growing that both forces are likely acting to structure communities, the relative importance and mechanisms for generating such communities is not well understood. Broader understanding of how evolution leads to changes in species’ niches and fitnesses would provide context for when and how niche and neutral forces are operating to structure communities. In this study, I use a simple model based on classic models of trait evolution combined with contemporary quantitative genetic models to investigate the interactions between ecological and evolutionary forces. I determine how niche overlap and fitness differences change as a result of selection under competition and develop insight into the relative magnitude of expected changes in niche overlap and fitness differences depending on the environmental context of species interactions. Ultimately, I address when and how coexistence conditions change as a result of evolution due to competition.

2.2 Model and Methods

I describe the population dynamics of two competing species and the evolutionary changes in a trait value xi of species i associated with resource acquisition. This trait value can be conceptualized as determining resource acquisition for an implicit continuous and limited resource set. For example, for a protist, the size of its oral groove is a trait that may determine the size of bacteria that it can consume. Species i has intraspeciﬁc variance τ, in the trait xi described by a normal distribution with mean trait value,x ¯i(t) at time t such that the species trait distribution, βi is described as

2 −(x−x¯i(t)) βi = kie 2τ2 , (2.1)

8 where ki is the maximum value of the distribution. The environmental landscape determines ki, such that

2 −(θ−x¯i(t)) ki = kmaxe 2σ2 , (2.2) where there is a trait value θ that maximizes the carrying capacity of both species. This optimal trait value allows a species to consume the most abundant resources in the environment. Both species have the same optimum trait value, which results in stabilizing selection on both species.

Here, kmax is the maximum value of the distribution such that ki = kmax whenx ¯i = θ. Species trait distributions decrease in height as they move away from θ at a rate inverse to the environmental niche breadth σ. This is tantamount to a smaller carrying capacity for a species as a result of scarcer resources for individuals with suboptimal trait values. Each species i (i = 1, 2) experiences changes in population size due to ecological forces and changes in mean trait value due to selection. Changes in population size is described by the Lotka−Volterra model of population growth in competition:

dNi Ki − Ni − βijNj = riNi (2.3) dt Ki where ri, is the intrinsic growth rate of species i, Ki is the population carrying capacity deﬁned as the area under the trait distribution curve βi (eq. 3.1). Finally, βij is the proportion of βi that overlaps with the trait distribution of species j such that forx ¯i < x¯j;

R f R ∞ 0 βjdx + f βidx βij = R ∞ , (2.4) 0 βidx where f is the value of x where βi = βj. βij represents the per−capita effect of species j on species i. The mean fitness,w ¯i, of an individual of species i is defined as the per−capita growth rate 1 dN i . Assuming that trait variance is maintained as a result of mutation−selection balance, Ni dt selection acts on the mean trait of each species,x ¯i as

dx¯i 2 δw¯i = Hiτ × , (2.5) dt δx¯i

9 δw¯i where Hi is the heritability of trait x, and is the change in mean fitness per change in mean δx¯i trait of a population (Lande 1976, Schreiber et al. 2011). The trait value of each species affects fitness through equation 4.1 in two ways. First, species will have increased carrying capacity and therefore increased fitness as the trait value gets closer to the environmental optimum θ. Second, the proximity of trait values between the two species determines the per capita effect of interspecific competition, thus species with more similar trait values will experience stronger selection due to interspecific competition. Thus selection will drive the change in the trait value of each species depending on the balance between these two pressures, such that in monocultures, either species will evolve to the environmental optimum, θ. κ As species evolve, changes in niche overlap ρ and fitness differences j between the species pair κi were calculated by letting

r α α ρ = ij ji , (2.6) αjj αii and

κ r α α j = ij ii . (2.7) κi αjj αji

1 βij as given by Chesson (2013), where αii = and αij = . Niche overlap, ρ can be manipulated to Ki Ki show that it is the geometric mean per capita effects of intraspecific competition. As a consequence of the assumptions of the model, ρ directly relates to the amount of overlap in the resource distri- κ bution curves of the two species. The average fitness difference j is intrinsic to the species pair κi and is not density dependent, this is essentially a measure of the relative resource use efficiency of the two species in the absence of intra and interspecific competition. One should cautiously note κ the distinction between the average fitness difference between two species, j and the mean fitness κi w¯i of a species, which depends on the density of both intra and interspecific competitors. For a given set of parameters numerical solutions to equations 3 and 4 were found using Math- ematica’s NDsolve function which gave time dependent solutions for the population size and trait values of two species. Solutions were generated until one species went extinct or 100 time steps elapsed, which produced stable dynamics or populations that had nearly converged on a stable dN dx equilibrium ( i and i were negligible). For each set of environmental parameters (H2 and σ), dt dt

10 the model was run with 900 distinct combinations of niche overlap and fitness differences. The initial value of niche overlap ρ was manipulated by varying the distance between x1 and x2 to cre- κ ate relatively uniform steps between 30 values of ρ, and fitness difference j was manipulated by κi varying the relative magnitude of kmax of each species to create relatively uniform steps between κ κ 30 values of j . At the end of each model run ρ and j were recalculated to determined how κi κi evolution changed both niche overlap and fitness differences. Additionally, the width of the environmental niche breadth σ was varied relative to the width of the species frequency distributions τ σ ( = 8, 5, 3, 2.2, 2, 1.25, 1, 0.75, 0.5, 0.25), which changed the amount of resources available for the τ species to use and share.

2.3 Results

The model framework followed species interacting through the implicit shared resource (Lotka- Volterra), which caused selection on trait values. Evolution of trait values of both species resulted in changes in niche overlap between species as well as changes in fitness differences (Fig. 2.1). The environmental niche breadth σ had large effects on the magnitude and direction of these changes, with subsequent effects on the ecological outcomes of competition, specifically species coexistence (Fig. 2.1). More detailed descriptions of these patterns follow. Change in niche overlap between species is strongly dependent on the initial niche overlap of the two species (Fig. 2.2), as well as the environmental niche breadth. In general species are more likely to increase their niche overlap when initially dissimilar, and diverge when similar, consequently there emerges a value of niche overlap that is evolutionarily stable. The evolutionarily stable value of niche overlap depends on the breadth of the environmental landscape: species tend to stably occupy more similar niches in narrower environments, whereas in broader environments species take advantage of the opportunity to use less similar niches (Fig 2.3A). As a corollary, convergence is less likely as the breadth of the environment increases (Fig 2.3B). Fitness differences tended to change as a result of evolution in competition in environments with narrow breadth (Fig. 2.1A), though this included both increases and decreases in the absolute magnitude of fitness differences (Fig. 2.4). This is because in narrow niche environments, small changes in trait values resulted in large changes in carrying capacity and therefore species average fitness. Whereas, in environments with wider niche breadths, changes in trait values did not result

11 A B C 1.5 1.5 1.5

σ = 0.1 σ = 0.2 σ = 0.3 1.0 1.0 1.0 0.5 0.5 0.5 0.0 0.0 0.0 log(fitness difference) log(fitness -0.5 -0.5 -0.5 -1.0 -1.0 -1.0 -1.5 -1.5 -1.5

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Niche overlap Niche overlap Niche overlap

Figure 2.1: Three examples of how niche overlap and fitness differences change as a result of selection on the trait values of two competitors. Each point on the graph is a starting condition and vectors indicate the direction and relative magnitude of changes in niche and fitness differences (scaled down for illustration). The gray areas are regions in which one species will be competitively excluded, whereas the white region indicates coexistence between species. Red arrows represent areas where species originally coexisted, but evolution changed the ecological outcome to competitive exclusion. Blue regions represent where species would undergo competitive exclusion if only ecological dynamics were occurring, but species evolved in ways that allowed coexistence before exclusion occurred. The sigma value indicates the breadth of the environmental landscape; species are more likely to evolve to coexist in environments with a broader environmental niche breadth.

12 1.0

0.8

0.6

0.4

0.2 Final niche overlap niche Final

0.0

0.0 0.2 0.4 0.6 0.8 1.0 Initial niche overlap

Figure 2.2: The initial niche overlap of the two species predicts the ﬁnal niche overlap of two competing species (for the parameters shown here: r2 = 0.67, p < 0.001). The grey area represents divergence in the species niches, and the white area represents convergence in the species niches. Where the linear regression intersects the shaded region, no change in niche overlap is predicted as a consequence of evolutionary dynamics, referred to subsequently as the evolutionarily stable value of niche overlap (H2 = 0.1, σ = 0.2) .

13 )

ρ 0.4 0.6 0.2

0.4 0.0

0.2 -0.2

-0.4 0.0 ( overlap niche in Change

Evolutionarily stable value of niche overlap of niche value stable Evolutionarily 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8

Environmental niche breadth (σ) Environmental niche breadth (σ)

Figure 2.3: The breadth of the resource environment determines how tightly species will pack after evolving in trait values in response to competition. The vertical dashed line is the trait variance of both species. When intraspeciﬁc variance of species is equal to the breadth of the environment, species maximize niche overlap. A. The evolutionarily stable value of niche overlap (see Fig. 2.1) tends to decrease in broader environments as species can decrease competition and still have access to resources. B. The change in niche overlap between species indicates whether species are converging or diverging in resource use. The grey area represents divergence in the species niches, and the white area represents convergence in the species niches. Each point represent the average tendency of species in an environment to move toward or away from each other in resource use, error bars are standard deviations of 9 runs of the model with diﬀerent initial starting values of species i.

14 1.0

0.5

0.0

-0.5 Average change in fitness difference in change Average -1.0

0.2 0.4 0.6 0.8

Environmental niche breadth (σ)

Figure 2.4: Changes in fitness differences as a function of the breadth of the environment. Fitness differences were more likely to both increase and decrease in narrower environments. Each point represent the average tendency of species in an environment to change, given different initial starting values of species i. in large changes in carrying capacity or fitness, thus species could more easily partition niche space without major decreases in fitness. Species were more likely to evolve to coexist in environments with broad niches and less likely to evolve to coexist in narrow niche environments. In broad environments coexistence arose through decreases in niche overlap between species. In narrow environments, exclusion arose as an outcome of increases in fitness differences. The relative magnitude of change of niche overlap versus fitness differences also depends on the niche breadth of the environment. Changes in fitness differences are relatively larger when the breadth of the environment is small, whereas in broader environments changes in niche overlap predominate (Fig. 2.5A). This pattern is however driven by decreasing changes in fitness as the breadth of the environment increases (Fig. 2.5C) rather than changes in niche, which has a much weaker correlation with niche breadth (Fig. 2.5B). Consequently evolution of trait values results in changes to coexistence of species in most environments. Species are more likely to evolve to coexist in broad environments.

15 3

0.25 1.0 2

0.8

1 0.20

0.6

0.15 0.4 Change in niche overlap niche in Change Change in fitness difference in Change

-1

0.2

0.10

-2 0.0

0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8 More change in fitness differences fitness differences in overlap niche change in More change More Niche breadth Niche breadth Niche breadth

Figure 2.5: The relative magnitude of changes in niche overlap and fitness differences as predicted by the breadth of the environment. A. The relative importance of changes in niche overlap and changes in fitness differences due to trait evolution in competitors. B. Absolute value of the average change in niche overlap due to evolution. C. Absolute value of the average change in fitness differences due to evolution. Each point is the average change in niche overlap and fitness difference of 900 runs with varying initial niche overlap and fitness differences (H2 = 0.1).

16 2.4 Discussion

Most models of evolution of competition predict that species will often diverge in trait values, which should increase coexistence between species. Here I analyzed a model that demonstrates selection on trait values results in changes in both niche overlap and fitness differences among species, which in turn effects coexistence. Two novel findings emerge from the model. First, fitness differences indeed change as a result of evolution, and often times these changes are bigger than changes in niche overlap. Secondly, species will converge to a tighter packing in their niches in landscapes with narrow resource availability. In order to have a complete picture of how ecological and evolutionary processes feed back on one another to regulate diversity in a community, it is important to consider not just changes in species niches, but also changes in fitness differences (Lankau 2011). The model demonstrates that evolution can cause large changes in fitness differences, especially in narrow environments. This is a consequence of the steeper slopes in the resource availability of the landscape. Thus changes in trait values that determine resource use have strong effects on the fitness of competitors because lower resources will result in lower species carrying capacity; a fundamental component of species fitness. Such large changes in fitness differences with changes in trait values makes maintenance of coexistence difficult in narrow environmental landscapes. In narrow environments, it is harder for species to coexist in general, but evolution towards the environmental optimum makes the space unable to be shared by both species. Conversely, in broad landscapes, resource availability slopes are flatter and so fitness changes less as a consequence of changes in the trait value associated with fitness. The model predicts that convergence in niche is more likely to occur between species in a narrow environmental landscape (Fig 3A). This is because the selective pressure of the narrow environment is stronger than the effects of interspecific competition. In any landscape, species will converge until selection due to competition is stronger than selection toward the environmental optimum. How- ever it is much more likely for species to diverge in trait values and niche overlap as long as the environment is sufficiently broad (Fig 3B); convergence may be a relatively special case. This is analogous to the results of the classic speciation models that found divergence and consequently speciation occurred only when the environmental gradient was broader than the variation within a species (Dieckmann and Doebeli 1999). Additionally the concept that a limit to the similarity

17 between competing species is based on the ratio of the breadth of the environment and intraspecific variation is not new (Macarthur and Levins 1967), but considering changes in niche overlap and fitness differences separately helps to disentangle some of the subtlety of how differences in competitive ability interact with niche overlap in evolutionary models (Abrams 1983). In general, we do not know the distribution of potential resources relative to intraspecific variation for populations in nature; thus it is difficult to predict the outcomes of eco−evo feedbacks on competition and coexistence. One might assume that since competition occurs frequently in nature and resources appear to be limiting in most populations, environmental landscapes may be rather narrow relative to the number of species present and thus convergence may be expected to occur. However most studies of evolution of traits in response to competition show niche divergence through experiments (Bono et al. 2013, Stuart et al. 2014, Zuppinger−Dingley et al. 2014, Ellis et al. 2015) or phylogenetic inference (Peterson et al. 2013, Agnarsson et al. 2015, Fiˇseret al. 2015). Thus if the world has unfilled niches communities are not at their species carrying capacity (Harmon and Harrison 2015). There are far fewer studies that allow inference into changes in fitness differences compared to studies that look at the evolution of niche overlap. In a study of Escherichia coli researchers found that evolution caused decreases in niche overlap and idiosyncratic changes in fitness differences (Zhao et al. 2015). However, interestingly in another study of an invasive plant introduction, plants seemed to increase coexistence with another species by increasing tolerance to allelopathy, which suggests changes in the fitness of the plants rather than the niche (Lankau 2012). Additionally, in a study of protists over succession, weak competitors tended to become stronger competitors over time in terms of suppressing the dominant species abundances (Miller et al 2014), this seems to indicate decreases in fitness differences between species. According to this model this would only occur in environments with very narrow resource breadth. Although the change in the niches of these species was not measured, the model would predict that these protists would also converge in their niche. The simplicity of this model results in some over simplifications that may not generalize to all systems. First, resources are modeled on a single niche axis. The approach presented here is relevant for species interactions that are driven by a single limiting resource, or covarying resources, but higher dimensional environments could complicate these results (Falster et al. 2017). Next, the

18 assumption of a continuous trait axis corresponding to a continuous resource may not be appropriate for systems with discrete limiting resources, like different types of nitrogen for plants unless the trait was associated with proportional use of two different discrete resources. Finally, I assume the resource base does not evolve, this may be true in a chemostat like system or a system with an abiotic resource. However, if predators are competing for evolving prey, dynamics would become more complex (Abrams 2000, Hairston et al. 2005). Predictions of this model may be difficult to test if species have already reach evolutionarily stable equilibria, therefore invasive species could be exploited to test these ideas. One would expect initially strong selection pressure both on the invader as a result of the environment and competition with natives, and selection on the natives as a result of competition with the invader. One might expect to see more evolution in the niche of the invader than the natives which may be evolutionary constrained. Additionally one would expect large changes in fitness difference between the native and invasive, as the invasive species fine tunes its efficiency in the new environment, whereas natives would have already exhausted this opportunity.

2.5 Conclusions

Here we find that evolution among competitors changes both components of coexistence of competitors; niche overlap and fitness differences, but the relative magnitude of changes depends on the availability of different resource types in the environment. In narrow resource environments, even species that initially coexist will generally evolve in ways that result in exclusion due to one species increasing its competitive ability while simultaneously increasing niche overlap. However there is a narrow range of conditions where species can evolve to coexist if they converge in both niche and fitness. In broader landscapes, it is relatively easy for species to evolve to coexist through diverging in niche overlap. Ultimately this model reveals that in order to predict how species will evolve in response to competition it is essential to consider not just changes in the niche of species but also changes in the average fitness differences between species, as they could be more important for predicting the outcome of evolution on coexistence under constrained environments.

19 CHAPTER 3

PHYLOGENETIC SIGNAL DOES NOT PREDICT COMPETITIVE INTERACTIONS IN PROTIST MICROCOSMS

3.1 Introduction

Resource competition is thought to affect community composition and regulate diversity in communities (Macarthur and Levins 1967, Tilman 1980, Chase and Leibold 2003). But quantifying competitive interactions in natural systems is often quite challenging. One possible predictor of the strength of competitive interactions in nature is the phylogenetic distance between a pair of species (Webb 2000, Webb et al. 2002, Cavender-Bares et al. 2009). Closely related organisms are expected to be more ecologically similar because of a shared evolutionary history (Harvey and Pagel 1991, Prinzing et al. 2001, Burns and Strauss 2011), and more ecologically similar species should have stronger competitive interactions due to niche similarity. Community phylogenetics is an alternative to trait-based approaches that may be particularly useful when multiple traits determine competitive interactions and traits are phylogenetically conserved, or when the traits that mediate competition are unknown. Furthermore, if phylogeny can predict the strength of competitive interactions it could also predict future evolutionary trajectories of different lineages. If phylogenetic signals in communities were informative, it would allow for inferring the role of competition in communities without the need for directly measuring species interactions (Cavender- Bares et al. 2009). However, the relationship between phylogenetic distance and strength of competition would break down if more distantly related species experience convergent evolution or closely related species experience character displacement in sympatry (Mouquet et al. 2012). Even when species traits have strong phylogenetic signal, the relationship between phylogeny and coexistence between species is less straightforward than originally thought. Modern coexistence theory proposes that two factors are important for coexistence between species, niche overlap and average competitive ability (fitness) differences (Chesson 2000). If niches are phylogenetically conserved there would be less coexistence between closely related species, but if competitive ability

20 differences are conserved, closely related species would be more equivalent, making competitive exclusion less likely (Mayfield and Levine 2010). In fact, in annual plants in California grasslands, fitness differences (determined with field based parameter estimates of competition) are conserved across the phylogeny whereas niches are not (Godoy et al. 2014). Perhaps the complicated relationship between phylogeny, trait conservatism and competitive interactions explains why there is mixed evidence supporting the so called ’competitive-relatedness hypothesis’ which states that more closely related species will be less likely to coexist due to competition (Cahill et al. 2008). For example, in many plant systems phylogeny does not predict competitive interactions (Cahill et al. 2008, Godoy et al. 2014, Zhang et al. 2016) even though ecological similarity can be related to phylogeny (Burns and Strauss 2011, Anacker and Strauss 2016). However, in protists (Jiang et al. 2010, Violle et al. 2010) and fungal (Maherali and Klironomos 2007) systems, a significant relationship between phylogeny and the strength of competitive interactions has been identified. These mixed experimental findings suggest that the competitive-relatedness hypothesis may depend on the types of species, the nature of the limiting resources, the time scales of experiments or some other aspect of the environment. An alternative hypothesis is that species traits should be a better predictor of competitive interactions as the link between traits and ecological processes is more direct (Goldberg 1996, Westoby et al. 2002, Chaveet al. 2009). In fact, in the case of forests in the French Alps, wood density has undergone convergent evolution and is a better predictor of competitive outcomes than phylogeny (Kunstler et al 2012). Understanding the evolutionary history of the species in a community can provide an informative context for other ecological and evolutionary questions. For example, trait evolution within a clade can provide valuable insight into the trade-offs and evolutionary constraints on ecologically meaningful traits (Sargent et al. 2007, Stephens and Wiens 2008). Also, the presence of phylogenetic patterns in communities across environmental gradients or through time (e.g. Pastore and Scherer 2016), may point to ecological mechanisms that could otherwise go undetected. Finally, relationships between phylogeny and ecological processes could predict future evolutionary processes (Weber et al. 2017). Exploring phylogenetic patterns in communities can aid in a more holistic understanding of community interactions, structure and function.

21 Here we investigated the relationship between phylogenetic distance and the strength of competitive interactions among pairs of protists in a direct test of the competitive-relatedness hypothesis. In laboratory microcosms, we quantified the competitive interactions of protists that co-occur in natural ponds. Additionally, we estimated the phylogenetic distance between each species pair by sequencing a portion of the 18s rRNA gene of each taxa and placing them on a well-resolved phylogeny of the Intramacronucleata protist clade (Gao and Katz 2014). Contrary to the findings of Violle et al. (2010) we found no significant relationship between phylogenetic distance and a species effect on or response to competition. Interestingly, protist traits like size, average competitive ability, and tolerance to stagnation also failed to predict competitive interactions.

3.2 Methods 3.2.1 Protist Cultures

Samples were taken from shallow waters of eight ponds in the Apalachicola National Forest, south of Tallahassee, FL. Individual cells of 12 species were isolated from these samples by serial dilution until species were in monoculture. Several cells of each species from a given pond were combined to create a line for that sample; between three and eight lines were created for each species by sampling from different ponds. Each line was maintained in a separate 50 mL macrocentrifuge tube, using 10 mL of a standard bacterial broth. The bacterial broth was made by autoclaving Tetramin fish food, and DI water and then adding 1 mL of a standardized bacterial inoculate that had been cryopreserved previously. The broth then incubated for 24 hours at room temperature on a stir plate. Protist cultures were refreshed every one to two months by inoculating 10 mL of newly prepared standard bacterial broth with 0.1-1 ml of the previous culture. Protists were maintained in the incubators for 24-24 months before the experiment began. Protists obtained with this method were single celled, free-living organisms that swam by beat- ing cilia or flagella and consumed bacteria in their habitat. Species had life spans on the order of six hours and reproduced predominantly through asexual cell division. One species, T etrahymena sp., was seen conjugating (sexual reproduction) when resources became scarce in their environments. Individuals ranged in size from 10-3200 µm2 when viewed in two dimensions. We identified species to genus by their placement on the broader phylogeny. Three flagellates were included in the experiment; Chilomonas sp. (ff), Entosiphon sp. (tv), P oterioochromonas sp. (ch) and nine

22 ciliates; Chilodonella sp. (wn), Colpoda sp. (bc), Cyclidium sp. (sw), Cyrtolophosid sp. (sc), Crytolophosis sp. (of), Maryna sp. (ca), P aramecium sp. (pm), P seudocyrtolophosis sp. (bj), T etrahymena sp. (ac).

3.2.2 Quantifying Competition

For each of 12 species, three lines from different ponds were combined. These mixed lines were used for inoculating the competition experiment. The competition experiment consisted of monoculture treatments for each of the 12 species and pairwise competition treatments for each of 78 species pairs, with three replicates for each treatment. A broth was prepared by combining 700 mL of DI water with 700 mg of Tetramin. The broth was then sterilized by autoclaving and inoculated with cryopreserved stock bacteria as describe above. The broth was incubated in the growth chamber for 24 hours and then diluted with three parts sterile water to one part bacterial broth. Each experimental unit was initiated by adding 10 mL of the diluted broth to a 50 mL conical macrocentrifuge tube and then inoculating the broth with 100 µL of either one species for monocultures or each of two species for competition treatments. Tubes were shaken and incubated in the growth chamber at 24◦C for 5 days until protists reached peak high densities. On the 5th day after inoculation (here after ’day 5’), the density of protists in each tube was quantified by counting species abundances in 100 µL of fluid on a Palmer cell under 100x magnification. Tubes were returned to the growth chamber and allowed to incubate for 41 more days, vortexed, returned to growth chamber and counted again 24 hrs later (here after ’day 46’). At day 5, when species were reaching peak densities and day 46 when species resources were depleted, a competitive index,

ICij, quantifying the eﬀect of species i on species j, was measured as the diﬀerence between each species population size in competition and monocultures using the formula (Miller et al. 2014):

log(xij) − log(xii) ICij = , (3.1) log(xii)

Here, xij is the average density of species i when species j is present and xii is the average density of species i in monoculture. We performed Z tests in R to determine if species tend to have negative interactions, and we compared the strength of competition at day 5 and day 46 with a t-test. To estimate protist competitive ability, we computed the average competitive eﬀect and response of each species, such that the competitive eﬀect of a species, j was the average ICij across

23 all species i, and the competitive response of a species, i was the average ICij across all species j. Although competitive eﬀects and responses have been found to be correlated (e.g. Miller et al. 2014), competitive eﬀects should predict dominant species in a community context, whereas competitive response relates to ability of species to persist amongst other dominant competitors.

3.2.3 Quantifying Protist Traits

On day 5 of the experiment, 0.5 mL of each monoculture was preserved separately in 27 µL of formaldehyde and 15 µL of iodine and stored at 4◦C. At least 30 cells of each species were photographed under 100x magniﬁcation; photos were loaded into ImageJ and the area of each cell was measured and scaled with a hemocytometer. P oterioochromonas preserved in formaldehyde were still alive and moving at the time of being photographed, so only iodine preserved protists were used for size measurements. Species tolerance to stagnant conditions (no resource input or waste output) was quantiﬁed as the density of species in monoculture at day 5 over the density of the species in monoculture at day 46. Tolerating stagnant conditions could predict a species ability to avoid competition in a stagnant environment.

3.2.4 Quantifying Phylogenetic Distance

We constructed a phylogenetic tree by sequencing the protists in this experiment and placing these sequences within a well-resolved phylogeny of the intramacronucleata clade (Gao and Katz 2014). We first extracted DNA from several lines of each species using the DNeasy Blood and Tissue kit (Qiagen) according to the manufacturer’s protocol. We amplified the 18s SSU rDNA gene region of each species using one forward primer (5’-CAA CCT GGT TGA TCC TGC CAG T-3’; Medlin et al. 1988, Chantangsi et al. 2007) and two reverse primers that we developed (5’- ATA TAC GCT ATT GGA GCT GGA-3’; starting at the 587th base pair of the forward primer, and 5’-TAA CTA TGT CTG GAC CTG GTA-3’; starting at the 1217th base pair of the forward primer) on each sample. PCR products were cleaned using EXO-SAP-IT (Affymetrix, Cleveland, Ohio, USA) according to the manufacturers protocol. DNA was sequenced with the Sanger method at the FSU core facilities using the three primers described above. Sequence reads were aligned and concatenated to generate 1217 base pair fragments within the 18s gene. Gao and Katz (2014) constructed a 9 gene, 537 taxon phylogeny of species within the intramacronucleata clade. We obtained their alignment of the 18S gene region which had been aligned

24 in GUIDANCE (Penn et al. 2010) with ambiguous columns in the alignment removed. We aligned their sequences with ours using MAFFT (Katoh and Standley 2013) on the Cipres Science Gateway using the default settings, and then removed the same ambiguous regions from our sequences. We then used add alignment in MAFFT to align our sequences with ambiguities removed to Gao and Katz’s original alignment. A maximum likelihood tree was constructed in PAUP* v4.0a152 (Swofford 2003). A heuristic search was performed using parsimony analyses, retaining 100 trees. An arbitrarily selected parsimony tree was used to estimate rate parameters for a likelihood analysis. A GTR+G+I model was the best fit model for the data and was used to perform a heuristic search using likelihood. The parsimony tree in memory was used as the starting tree, asis stepwise addition and TBR branch swapping were performed on the best tree with a reconnection limit of 8. Branch lengths were optimized with Newton’s method. The search was stopped after 24 days at which point the search had not generated a tree with a log likelihood score less than 2 for two days. Taxa with exceptionally long branches and basal clades that included no taxa used in the experiment were then pruned from the phylogeny, leaving 227 taxa. Model parameters were re-estimated and the heuristic search was continued for 24 more hours. This tree was time calibrated using penalized likelihood in r8s (Sanderson 2003). Point estimates for the divergence of P aramecium and T etrahymena, T etrahymena and Chilodonella, T etrahymena and P hytophthora, and T halassiosira and P haeodactylum were found in Parfrey et al. (2011). Phylogenetic distance for each species pair was determined as the sum of the branch lengths between a pair of species on the ultrametric tree (Fig. 3.2). Branch lengths were then correlated with the ICij of each species pair with linear regression. We tested for phylogenetic signal in the average competitive effect and response of species as well as protist size and intolerance to stagnant conditions using Blomberg’s K (Blomberg et al. 2003) and Pagel’s lambda (Pagel 1999) with the phytools package in R (Revell 2012), K < 1 indicates over-dispersion of traits within a clade and K > 1 indicates trait clustering within a clade. We performed all analyses with all species, or with just the Alveolata (no flagellates) to test for these patterns among a less diverged group.

25 Table 3.1: Day 5: Table of interaction coefficients, ICij. Were column species are species, j, which effect the row species, i. Signicant differences between density of species in monocultures vs. competition are indicated. + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001.

3.3 Results

The measures of competition, ICij, were different from zero at day 5 (Z = −4.344, P < 0.001) and at day 46 (Z = −18.70, P < 0.001), indicating competition was occurring in this system(Table 3.3 and 3.3). Competitive effects were stronger at day 46 than day 5 (t = 3.842, P < 0.001), but the competitive effects at day 5 did not predict competitive effects after 46 days of competition (P = 0.69,R2 = 0.001; Fig. 3.1. One of the flagellate species included in the experiment, Entosiphon sp., is part of the Excavata clade (paraphyletic with ciliates), and was excluded from the alignment and downstream analyses because this taxon was too divergent to meaningfully align in this set of species. Of the 11 remaining taxa, nine were ciliates within the Alveolata superphylum, and two were flagellates within the Heterokont superphylum (Fig. 3.2). Patristic distance between each species pair on the ultrameteric tree (hereafter phylogenetic distance) had a weak positive correlation with interaction coefficients at 5 days (P = 0.065,R2 = 0.031), but not at 46 days (Fig. 3.3). When only the Alveolata were included in analyses, there was no relationship between phylogenetic distance and interaction coefficient at 5 days. However, there was a weak negative relationship between phylogenetic distance and interaction coefficient after 46 days (P = 0.086,R2 = 0.042) indicating that on average more distantly related species had stronger negative effects on each other (Fig. 3.4).

26 1.5 1.0 ) ij IC ( 0.5 0.0 -0.5 Interaction coefficient coefficient Interaction -1.0 -1.5

Day 5 Day 46

Figure 3.1: Interaction Coeﬃcients ICij tended to be negative at day 5 and day 46, indicating competition among many species pairs was occurring. Competition was stronger at day 46 than day 5.

27 Table 3.2: Day 46: Table of interaction coefficients, ICij. Were column species are species, j, which effect the row species, i. Signicant differences between density of species in monocultures vs. competition are indicated. + p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001.

Phylogenetic signal results were consistent between Blomberg’s K and Pagel’s λ. No phylogenetic signal was detected in the protist traits measured including protist size (P = 0.383,K = 0.615), tolerance to stagnation (P = 0.455,K = 0.813). The average competitive eﬀect of a species was not related to the phylogeny (Day 5: P = 0.255,K = 0.638, Day 46: P = 0.144,K = 0.758), nor was the average competitive response of a species (Day 5: P = 0.668,K = 0.440, Day 46: P = 0.229,K = 8.15) as determined by linear regression in the absence of phylogenetic signal.

3.4 Discussion

If phylogenetic conservatism is a dominant force in structuring the niches of competing species, then there should be a strong relationship between phylogenetic distance and interaction strength between competitors. Presumably such an eﬀect would be mediated through multifaceted trait conservatism. To test this hypothesis, we measured competitive interactions between 11 species of protists, and estimated their phylogenetic relationships. At best, we observed only marginally signiﬁcant relationships between phylogenetic distance and competition, and even these patterns were either in the opposite direction than was predicted (Fig. 3.4) or were not robust over time or phylogenetic scale (Fig. 3.3). Furthermore, traits did no better in predicting competitive outcomes and were also not phylogenetically conserved (Fig. 3.5 and 3.6). From this we can conclude

28 Oligohymenophorea Phyllopharyngea Colpodea Apicomplexa Dinoflagellates Stramenopiles 1500 Chilomonas sp. (ff) Chilomonas sp. (ch) Poterioochromonas Sr St Aano Sr St Pinf Sr St Tpse Sr St Ptri Sr Ap Cpar Sr Pmar Di Sr Atam Di Sr Kbre Di Sr Ap Tpar Sr Ap Eten Sr Ap Tgon Sr Wmet Co Sr Sasp Co Sr Rspa Co Sr Pvor Co Sr Pbro Co Sr Ssto Co Sr Odra Co Sr Ph Tste sp. (wn) Chilodonella Sr Ph Cunc Sr Ph Tfau Sr Ph Hfan Sr Ph Hsin Sr Ph Hder Sr Ph Ispd Sr Ph Ispc Sr Ph Dpro Sr Ph Dder Sr Ph Ctri Sr Ph Cexc Sr Ph Apua Sr Ph Espi Sr Ph Egem Sr Ph Plim Sr Ph Herh Sr Ph Pspi Sr Ph Dcol Sr Ph Atub Sr Ph Acom Sr Zcf.Na Sr Zaga Na Sr Ospm Na Sr Oapo Na Sr Pdua Na Sr Lsp. Na Sr Lcos Na Sr Pr Pstr Sr Pr Ssoc Sr Pr Sbud Sr Pr Plsa Sr Ol Lbul Sr Ol Sver Sr Ol Fspj sp. (pm) Paramecium Sr Ol Ptet Sr Ol Pcau Sr Ol Fdid Sr Ol Adoh Sr Ol Ocat Sr Ol Imul Sr Ol Cstr Sr Ol Ccol Sr Ol Gbro Sr Ol Gsci Sr Ol Bbra Sr Ol Dcam Sr Ol GchQ sp. (ac) Tetrahymena Sr Ol Tthe Sr Ol Tpyr Sr Ol Lspa Sr Ol Aasp Sr Ol Omic Sr Ol Ospc Sr Ol Ospd Sr Ol Oall Sr Ol Cumb Sr Ol Vcry Sr Ol Zsin Sr Ol Zalt Sr Ol Tmat Sr Ol Pspf Sr Ol Epla Sr Ol Echr Sr Ol Omin Sr Ol Vmic Sr Ol Ohen Sr Ol Aenr Sr Ol Zarb Sr Ol Vcam Sr Ol Eabr Sr Ol Ppun Sr Ol Ppaa Sr Ol Over Sr Ol Cpol Sr Ol Aros Sr Ol Aarn Sr Ol Tsin Sr Ol Tnob Sr Ol Thet Sr Ol Tmya Sr Ol Tepi Sr Ol Uure Sr Ol Hlwo Sr Ol Gspb Sr Ol Gpit Sr Ol Cova Sr Ol Pper Sr Ol Plon Sr Ol Phar Sr Ol Pmag Sr Ol Msin Sr Ol M502 Sr Ol McaQ Sr Ol McaG Sr Ol Hset Sr Ol Sdog Sr Ol Pnot Sr Ol Cvea Sr Ol Pdig Sr Ol Papo Sr Ol Parm Sr Ol U173 Sr Ol Umar Sr Ol P494 Sr Ol Pvir Sr Ol Ebor Sr Ol Uhet Sr Ol U491 Sr Ol Upar Sr Ol Ufil Sr Ol Ploa Sr Ol Ppac Sr Ol Tvor Sr Ol Etea Sr Ol Epil Sr Ol Pdic Sr Ol Gtri Sr Ol Mavi Sr Ol Ahae Sr Ol Pden Sr Ol Spla Sr Ol Shol Sr Ol Utur Sr Ol Uspc Sr Ol Dpan Sr Ol Pwas Sr Ol P218 Sr Ol Cver Sr Ol Pset Sr Ol Pcor Sr Ol Pgro Sr Ol P363 Sr Ol Saes Sr Ol S336 Sr Ol Hnat Sr Ol Hmin Sr Ol Wtyp Sr Ol Esia Sr Ol Hsal Sr Ol Cplo sp. (sw) Cyclidium Sr Ol Cgla Sr Ol Aspb Sr Ol Bsub Sr Ol Ccf. Sr Ol Acra Sr Ol Dspa Sr Ol Dsp. Sr Ol Npro Sr Ol Pelo Sr Ol Pcon Sr Ol Mspc Sr Ol Ecom Sr Ol Amar Sr Ol Mspb Sr Ol Mint Sr Ol Abiv Sr Pr Tfus Sr Pr Chia Sr Pr Lbiw Sr Pr Cnol Sr Pr Nsp. Sr Pr Aspf Sr Pr Amag Sr Pr Pter Sr Pr Povu Sr Pr Pala Sr Ol Bmas Sr Cr Cirr Sr Ccau Na Sr Ogeo Na Sr Nspb Na Sr Fblo Na Sr Marc Co (sc) Cyrtolophosid sp. (cm) Colpidium sp. (bj) Pseudocyrtolophosis Sr Palp Co Sr Ol Aspk Sr Ol Amac sp. (of) Cyrtolophosis Sr Cmuc Co Sr Npaa Co Sr Bpul Co Sr Bgem Co Sr Ipal Co sp. (ca) Maryna Sr Hdis Co Sr Pspa Co Sr Mumb Co Sr Mova Co Sr Eaug Co Sr Mter Co Sr Pnan Co Sr Cmag Co sp. (bc) Colpoda Sr Cinf Co Sr Bvor Co Sr Ccuc Co Sr Bdis Co Sr Bsph Co Sr Btru Co Sr Baty Co 1250 1000 750 Time from root (million years) Time from root (million 500 250 0

Figure 3.2: Ultrametric maximum likelihood tree reconstruction of our sequences added to the Gao and Katz 18S alignment, pruned to included only species present in this study. Species abbreviations are given for comparison to Table 3.3 and 3.3. 29 Day 5 Day 46 A B 1.5 1.5

1.0 1.0 ) ij IC

( 0.5 0.5

0.0 0.0 cor$week6

-0.5 -0.5 Interaction coefficient coefficient Interaction

-1.0 -1.0

-1.5 -1.5

500 1000 1500 2000 2500 3000 500 1000 1500 2000 2500 3000

Phylogenetic distance (my) Phylogenetic distance (my)

Figure 3.3: With all taxa included, there was no significant relationship between phylogenetic distance and competitive interactions at either (A) 5 days or (B) 46 days in competition. ICij = 0 indicates no effect of interspecific competitors, negative values indicate negative effects on species abundances in the presence of an interspecific competitors, and positive values indicate facilitation. The dashed line represents a marginally positive significant relationship between phylogenetic distance and the interaction coefficient found on day 5.

30 Day 5 Day 46 A B 1.5 1.5

1.0 1.0 ) ij IC

( 0.5 0.5

0.0 0.0 cori$week6 -0.5 -0.5 Interaction coefficient coefficient Interaction

-1.0 -1.0

-1.5 -1.5

500 1000 1500 2000 500 1000 1500 2000

Phylogenetic distance (my) Phylogenetic distance (my)

Figure 3.4: With only ciliates (Alveolata), (A) there was no relationship between phylogenetic distance and competitive interactions at 5 days, (B) but at 46 days there was a marginally significant correlation between phylogenetic distance and the strength of competition between heterospecifics, suggesting that species that diverged longer ago have stronger competitive interactions. This could occur if closely related taxa experienced more niche divergence in sympatry. Interpretation of interaction coefficients is explained in Fig. 3.3.

31 A B 0.0 0.0

-0.2 -0.5

-0.4

-1.0 Average competitive effect competitive Average Average competitive response competitive Average -0.6 Day 5 Day 46 -1.5

0.0000 0.0005 0.0010 0.0015 0.0020 0.0000 0.0005 0.0010 0.0015 0.0020

Protist size (mm2) Protist size (mm2)

Figure 3.5: The size of a protist did not predict its (A) average competitive eﬀect (Day 5: P = 0.268,R2 = 0.134, Day 46: P = 0.156,R2 = 0.263), or (B) average competitive response (Day 5: P = 0.386,R2 = 0.084, Day 46: P = 0.251,R2 = 0.143), at either time point.

32 A B 0.0 0.0

-0.2 -0.5

-0.4

-1.0 Average competitive effect competitive Average Average competitive response competitive Average -0.6 Day 5 Day 46 -1.5

-0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2

Tolerance to stagnation Tolerance to stagnation

Figure 3.6: The tolerance of each protist species to stagnation (density in monoculture at day 46/density in monoculture at day 5) did not predict its (A) average competitive eﬀect (Day 5: P = 0.545,R2 = 0.042, Day 46: P = 0.585,R2 = 0.034), or (B) average competitive response (Day 5: P = 0.118,R2 = 0.249, Day 46: P = 0.313,R2 = 0.113), at any either time point.

33 that phylogeny is not an immediately helpful tool for understanding the ecological dynamics in this laboratory system and should not be used to infer concurrent ecological and evolutionary processes. Phylogenetic analyses have already been shown to be useful in other systems where they can predict ecological patterns due to trait conservatism (Cavender-Bares et al. 2009), or if coupled with solid phylogenetic studies of trait evolution (Kraft et al. 2007). It appears that the questions should now shift to understanding at what scale (Swenson et al. 2013) and under what conditions phylogenetic conservatism is meaningful for competitive interactions. When working with clades as deeply divergent and diverse as protists, the scale at which ecology is conserved is unclear. If ecological traits are labile relative to the depth of the phylogeny, only very shallow scales should show relationships between ecology and phylogeny. On the other hand, closely related species occurring sympatrically will have been asserting evolutionary pressures on each other, and so phylogenetic signal may not be conserved even among shallow clades. Most models predict that such pressure should result in niche divergence between species (e.g. Abrams 1983), as well as changes in fitness differences (Chapter 2). One surprising aspect of our results is that they are quite different from a similar study that found a strong relationship between coexistence and phylogenetic distance with protists in the laboratory (Violle et al. 2010). The contrast between Violle et al. (2010) and the work presented here is noteworthy given the similarity of the lab system and taxa used. The pattern they found was remarkably strong, whereas patterns in our work are remarkably weak or absent. There are three key differences between the studies that may account for the differential results. First, they used only protists within the ciliate clade (intramacronucleata), whereas our study contains flagellates as well. When we analyzed our data to only include ciliates, a pattern in the opposite direction as Violle et als’ study emerged (Fig. 3.4). Second, Violle et al. (2010) used only three species of bacteria for the protists to consume, whereas we used several hundred bacterial OTU’s. Simplified bacterial communities would likely decrease the ability for niche differentiation thus forcing species to compete more than would be expected in other conditions: they observed a range of responses in populations from nearly total reduction in population size due to competition, to almost none. Third, their protists were obtained from biological supply companies, whereas our taxa where cultured from sympatrically occurring wild protists. In situ evolution of our co- occuring protists in the recent evolutionary history of the Apalachicola area could have resulted

34 in niche differentiation between closely related taxa thus decreasing niche overlap and competitive interactions, subsequently reducing the relationship between ecological function and phylogenetic distance. The differential roles of niche overlap and fitness difference in contributing to phylogenetic patterns is interesting to note in the context of this study. Indeed, if niche overlap is phylogenetically conserved then one could expect strong competitive interactions and less coexistence between closely related taxa. Whereas if fitness differences are phylogenetically conserved, more closely related taxa will be more equivalent and less likely to go extinct, despite their niche overlap (May- field and Levine 2010). Here, we do not find evidence that competitive ability is phylogenetically conserved. Given that all species were grown in the same competitive environment, this is a good approximation for fitness differences between species, suggesting that in this study, fitness is not conserved across the phylogeny (but see Godoy et al 2015). Additionally, since we observed no relationship between competitive interactions and phylogeny, niche overlap must not be phylogenetically conserved among species. Interestingly, when just considering the intramacronucleata clade after 46 days of co-occurring, more distantly related species compete more strongly, and closely related species tend to facilitate each other. Such relationships between phylogeny and facilitation are predicted to occur in stressful environments (Verdu et al. 2009). Indeed after 46 days, resources are likely quite low and a large amount of metabolic waste products from both bacteria and protists would have been generated, likely shifting the composition of bacteria present in a microcosm to those OTUs that can metabolize waste. If protists can utilize the bacteria associated with another protists metabolic waste, this could be a mechanism of facilitation. In general, the applicability of laboratory microcosms for inference about phylogenetic patterns in natural systems is unclear. First, the niche of species is more complex than the draw down of a simplified resource base (Tilman 1990), differential responses to spatial and temporal variability of the quality of the environment as well as apparent competition through a shared predator afford other opportunities for niche differentiation (Amarasekare 2003). Access to mutualists can increase access to resources and weaken or intensify competitive interactions (Cushman and Addicott 1989, Bastolla et al. 2009). Notwithstanding, competition may be experienced differently or not at all in the context of a full community with many indirect effects occurring (Miller 1994).

35 3.4.1 Conclusions

With the advent of cheap and easy sequencing techniques, the promise of phylogeny to predict ecology is ﬁnally being widely tested. However, we are discovering that the relationship between phylogenetic distance and ecological similarity is not as straightforward as expected. As more evidence accumulates, it appears that complicating factors such as community context, phylogenetic scale, niche convergence and phylogenetic repulsion can obscure phylogenetic eﬀects. Perhaps phylogeny can be predictive for species that have recently been introduced, as phylogenetic distance does appear to be predictive of invasion success (Strauss et al. 2006, Hill and Kotanen 2009, Jiang et al. 2010, Peay et al. 2012), but is less useful for co-occuring species that have been coevolving for a long time. A clear understanding of the conditions under which phylogenies can be useful is still being resolved.

36 CHAPTER 4

EVIDENCE FOR THE ROLE OF EVOLUTION IN COMPETITOR COEXISTENCE IN PROTIST MICROCOSMS

4.1 Introduction

The enigma of diversity is two-fold; how does diversity emerge, and how is it maintained? While evolution is the study of the processes that generate new species, ecology investigates the maintenance of species, often in the face of inherently unstable interactions that act to degrade species diversity. The intersection of these very different processes can be found at the scale of in situ selection occurring on ecological timescales and in the realm of feedbacks between the ecological and evolutionary processes (Weber et al. 2017, terHorst et al in Press). As species adapt to their environments they interact with other species and both exert and experience selective pressure that can change ecological interactions and subsequent selection pressures (Kokko and Lòpez-Sepulcre 2007, Becks et al. 2012). As such, the effects of ecology and evolution are intertwined. Previous work in predator-prey systems has shown that both ecological and evolutionary dynamics are needed to explain changes in population dynamics that ecology alone cannot (e.g. Yoshida et al. 2003, Hairston et al. 2005). Similarly, models have suggested that evolution of competitors can lead to coexistence or extinction, depending on the nature of the resources and initial conditions (Rummel and Roughgarden 1985, Gomulkeiwicz and Holt, 1995). However, how selection facilitates the maintenance of diversity among competing species is largely unknown. A key concept in ecology is that evolution leads to decreasing niche overlap via character displacement, facilitating coexistence between species (e.g. Macarthur and Levins 1967, Abrams 1983). Subsequently, modern coexistence theory has revealed that there are two essential components for determining coexistence between species (Chesson 2000, Adler et al. 2007). First, niche overlap, or the similarity in the usage of resources or effects of predators in space and time, stabilizes coexistence between species. Second, fitness differences, which can be thought of as the difference in average reproductive output between two species, make species less equal in their competitive

37 ability. These two factors are best thought of as continuous variables that both contribute to coexistence. Thus, two species that use the same resources in the same way cannot coexist in the presence of any fitness differences. But competitors that are equal in fitness can coexist even when they require nearly the same resources. Unlike niche partitioning, very little research has considered how fitness differences should change in response to evolution of competitors (but see: Zhao et al 2016, Chapter 2). If species evolve to be more equal in fitness, this process should facilitate coexistence, but if one species evolves to increase its fitness faster than the other species, the ability for these species to coexist will decrease (Aarssen 1983). Thus whether species evolve to coexist depends on the relative magnitude and direction of changes in both niche overlap and fitness differences. In fact, the theoretical models presented in Chapter 2 of this dissertation predicts changes in fitness differences due to evolution will be more important for competitive outcomes than changes in niche overlap in environments with few opportunities for niche divergence (Chapter 2). Such evolutionary outcomes would decrease coexistence between species. However, it can be difficult to infer the breadth of potential resources for species to use alongside the evolutionary constraints of an organism to evolve to use those resources in a given environment in order to predict whether niche or fitness differences will change more due to evolution. Niche divergence has long been inferred from species trait and distribution patterns (e.g. Macarthur 1958, Pacala and Roughgarden 1985). Even in recent years, evidence that competition can result in niche divergences continues to accumulate. For example, there are experimental demonstrations of niche differentiation in competing plants (Zuppinger-Dingley et al. 2014), anolis lizards adapting to different habitat types after a competitor was introduced (Stuart et al. 2014), and niche expansion in viruses and biofilms (Bono et al. 2012, Ellis et al. 2015). Additionally, there is evidence of character displacement though phylogenetic inference (e.g. Peterson et al. 2013, Ag- narsson et al. 2015, F˘iser et al. 2015). Although some models predict species can converge in their resource use (Abrams 1987, Fox and Vasseur 2008, terHorst et al. 2010), there are few explanations for how equivalent species emerge and persist in nature as they evidently do (Leibold and McPeek 2006). In this study, three pairs of protists in laboratory microcosms were used to quantify changes in both niche overlap and fitness differences resulting from evolution under pairwise competition.

38 We performed selection experiments where species evolved in either monocultures or pairs. And we estimated niche overlap and fitness differences between species pairs (Godoy el al. 2014) before and after selection using Lotka-Volterra populations models parameterized with response surface experiments (Law and Watkinson 1987, Inouye 2001). We asked how selection due to competition changes coexistence between species pairs and if this occurs through changes in niche overlap or fitness differences. In this system, selection due to competition tends to result in a lower probability of coexistence between species through increases in fitness differences, with little change or increases in niche overlap.

4.2 Methods 4.2.1 Bacterial Broth

4.2.2 Protist Cultures

39 Protists obtained with this method were single celled, free-living organisms that swam by beat- ing cilia and consumed bacteria in their habitat. Four species were used in this experiment were identiﬁed to genus level by sequencing their 18s rRNA gene region and placing them on a well resolved phylogeny (Chapter 3). All four species were among the Intramacronucleata clade of cil- iated protists, two species were among the Colpodea class; Colpoda sp. (Colpoda) and Maryna sp. (Maryna), and two were in the Oligohymenophorea class; T etrahymena sp. (T etrahymena), and P aramecium sp. (P aramecium). Species had life spans on the order of six hours and reproduced predominantly through asexual cell division, however T etrahymena was seen conjugating (sexual reproduction) when resources became scarce in their environments. P aramecium were the biggest (∼2000 µm2), then Colpoda (∼900 µm2), and T etrahymena (∼400 µm2), and smallest was Maryna (∼300 µm2).

4.2.3 Preparation of Stock Lines for Experiment

To prepare each species line for experimental use, 0.1-1 mL of each distinct protist line was added to a separate 50 mL centrifuge containing 10 mL of a standardized bacterial broth. These cultures were incubated for 4-6 days on a shaker tray in the growth chamber. When cultures reached high densities, all lines of the same species were combined into one tube. We then attempted to remove most of the bacteria and nutrients from these cultures while retaining the protists. Combined cultures were first passed through a 50 µm mesh filter to remove larger particulate matter. A gravity and suction filtration system with a 0.5 µm filter was assembled with the goal of retaining the 10-50 µm protists above the filter and essentially washing the bacteria through by the addition of sterile water. Beads were added to the super-filtrate to prevent protists from having their tiny selves smashed against the mesh like waffles, which we assume they don’t like, but we have little data to support anthropomorphizing these beasts. After adding the protist culture to the filter, an equal volume of sterile DI water was slowly added to prevent all the water from draining out. When 50 mL was left above the filter the super-filtrate was obtained and used for subsequent experiments. Although this method tends to decrease protist abundances by up to half, it appears to remove most bacteria, so negligible amounts of nutrients and bacteria are added to the experiment.

40 4.2.4 Selection Experiment

Three selection experiments were run at different times with three species pair combinations; Colpoda with T etrahymena, Maryna with T etrahymena and Maryna with P aramecium. For each species in each pair, five replicates of two different selection environments were implemented: pairwise competition or alone in monocultures. To initiate these environments, each experimental unit received 5 mL of standardized bacterial broth diluted to 0.5 mg/ml of Tetramin fish food. To initiate the pairwise selection environment, 4 mL of each species filtered culture and 8 mL of sterile water was added to the falcon tube for a total of 21 mL of fluid. For the monoculture treatments, we used 5 mL of diluted broth as before, 4 mL of the monoculture species and 12 mL of sterile water. Tubes were sealed and kept on a shaker tray in the incubator for six weeks. Tubes were refreshed every four days by shaking vigorously, removing 10 mL of fluid and replacing that fluid with fresh standardized broth diluted to 0.25 mg Tetramin/mL to maintain similar nutrient densities and decrease the accumulation of waste products. For two of the three species pairs, densities were measured every week for six weeks. At the end of the selection period, for each replicate, 10- 50 cells were isolated and used to initiate monoculture lines. For each of the five replicates of each experiment, four lines were generated; both species selected in either a monoculture or in pairwise competition. These lines were then grown to high density and prepared for subsequent experimentation with the above filtration methods, keeping each replicate separate.

4.2.5 Response Surface Experiments

For each pair of species, we performed response surface experiments before and after selection, these experiments allowed us to parameterize Lotka-Volterra competition models (Law and Watkin- son 1987, Inouye 2001). Variables from these models can be used to quantify species niche overlap and species fitness differences (Chesson 2013, Kraft et al. 2015). Before selection, combined and filtered protist cultures were used to set up these experiments. Each experimental unit contained 5 mL of 0.5 mg/mL standardized broth, then each species was added at three different density levels to monoculture and competition tubes. A total of ten density treatments were used for each pair of species, three densities for each monoculture, and four density combinations of pairwise competition (Fig. 4.1). For high density treatments, 8 mL of protists were added to each tube, for medium density, 4 mL were added and for low density 2 mL were added. To standardize the

41 150

j 100

50 Density of species Density

0 50 100 150 200 Density of speciesi

Figure 4.1: Example response surface experiment density levels for two species. Dots represent target density levels of each species and asterisks represent the carrying capacity of each species. This experimental design will allow us to parameterize linear models of competition between two species. volume of the tube, enough sterile water was added to bring the tube to 21 mL. Upon completion of the experimental set-up, monoculture densities in 0.1 mL were counted under a microscope using a palmer cell. Tubes were stored on the shaker tray in the growth chamber and counted at two days and four days. For the selection lines, response surface experiments were performed as described above, with species selected in two species mixtures grown together and species grown in monoculture grown together. There were ﬁve replicates of each selection line and each monoculture line; to perform the response surface experiments, only one replicate was used at a time to keep replicates independent of each other.

42 4.2.6 Invasion Experiments

For one species pair (Maryna and Paramecium), we performed additional invasion experiments before and after selection as an alternate assessment of coexistence. Species can theoretically coexist if both species can increase from low density in the presence of its competitor. These experimental treatments were set up similarly to the response surface experiment, where 8 mL of one species was added to 5 mL of broth, and then the tube was ﬁlled to 21 mL with sterile water. Next, 0.1 mL of the invading species was added to the tube. Density was measured at two days and four days.

4.2.7 Estimating Niche Overlap and Fitness Diﬀerences

To determine niche overlap and ﬁtness diﬀerences between a pair of either preselection lines, lines selected in monoculture, or lines selected in competition, we used data from the response surface experiments and maximum likelihood estimation to parameterize a Lotka-Voltera competition model;

dN i = rN (1 − a N − a N ). (4.1) dt i ii i ij j dN We used initial counts and counts after 4 days to estimate the growth rate, i of each line under dt each density treatment. We then used the growth rate and the initial densities from the response surface experiment as the input data. We used the nlminb function in R (R Core Team, 2017) to optimize parameter values, r1, a12, a11, r2, a22, a21 while bounding them by zero, initial values were set to 0.0001. We used a parametric bootstrapping to generate 90% confidence intervals for the parameter estimates. Niche overlap and fitness differences were calculated from these parameters using the equations in Chesson (2013).

4.3 Results 4.3.1 T etrahymena vs Colpoda

The density treatments we implemented before and after the selection regimes allowed us to quantiy the per-capita growth rate of a focal species in response to changes in intra- and interspeciﬁc mixtures. Multiple regression showed that both protist species’ densities tended to aﬀect the per- capita growth rate of the focal species both before and after selection (Table 4.3.1). However, after

43 Table 4.1: t-statistics and P-values from multiple linear regression of the eﬀect of each species density on the per-capita growth rate of the focal species.

Preselection Monoculture Competition Focal Species Tetrahymena Colpoda Tetrahymena Colpoda Tetrahymena Colpoda Tetrahymena -3.489 0.0015 -3.856 <0.001 -4.774 <0.001 -2.717 0.009 -2.968 0.0086 -4.578 <0.001 Colpoda -2.077 0.0462 -6.669 <0.001 -2.713 0.0101 -2.561 0.0146 -0.024 0.9810 -2.487 0.0196

Preselection Monoculture Competition Focal Species Tetrahymena Maryna Tetrahymena Maryna Tetrahymena Maryna Tetrahymena -4.111 <0.001 -2.998 0.005 -4.263 <0.001 -1.498 0.143 -2.541 0.0158 -1.570 0.126 Maryna 0.309 Intraspecific0.760 -3.460 effects0.002 -1.895 0.066 -1.857 0.072Interspecific-0.302 0.767 effects-3.684 <0.001 A B

2.22 Preselection Monoculture Competition Focal Species Maryna Paramecium Maryna Paramecium Maryna Paramecium Maryna -3.656 <0.001 -1.042 0.304 -6.131 <0.001 -1.428 0.155 -4.127 <0.001 -2.345 0.0204 0.028 Paramecium2.2 3.918 <0.001 -1.489 .144 2.541 0.012 -3.642 <0.001 -0.735 0.463 -4.618 <0.001 Colpoda 11 12 a a 0.005 0.03 Effects on 0.002 0.01 0 Before 1 sp. 2 sp. Before 1 sp. 2 sp. Selection environment Selection environment

C D 0.005 0.020

0.004 0.015

0.003 22 21 Tetrahymena

a a 0.010 0.002

0.005 0.001 Effects on 0.000 0.000 Before 1 sp. 2 sp. Before 1 sp. 2 sp. Selection environment Selection environment

Figure 4.2: Parameter estimates and 90% conﬁdence intervals generated by bootstrap analysis for a Lotka-Volterra competition model. Colpoda is species 1 and T etrahymena is species 2.

44 A B 1.0 2

0.8 1

0.6 0

0.4 Niche overlap Niche

-1 Log(fitness difference) Log(fitness

0.2 Sp.1 wins Sp.2 wins

-2

0.0 Before 1 sp. 2 sp. Before 1 sp. 2 sp.

Selection environment Selection environment

Figure 4.3: Estimates of niche overlap and fitness difference before selection and in different selection environments. Error-bars are 90% confidence intervals. B. Negative values indicate a competitive advantage for Colpoda, and positive values indicate a competitive advantage for T etrahymena. selection in competition, T etrahymena no longer had a significant effect on the per capita growth rate of Colpoda. One might expect to see similar trends in the maximum likelihood parameter estimates for the Lotka-Volterra models. However, although estimates converged, the bootstrapped confidence intervals tended to be large, and differences between parameter estimates before and after competition could not be detected (Fig. 4.3.1). For example, the linear models suggest the effect of T etrahymena on Colpoda decreased due to the interspecific selection, concurrently we see a decrease in the estimate of a12 (effect of T etrahymena on Colpoda), but this value has very large error associated with it. There is marginal evidence that the intraspecific effect of Colpoda increased in pairwise selection compared to monoculture. Niche overlap and fitness differences are calculated from the alpha parameters, therefore when

45 Table 4.2: t-statistics and P-values from multiple linear regression of the eﬀect of each Preselection Monoculture Competition speciesFocal Species densityTetrahymena on the per-capitaColpoda growthTetrahymena rate of the focalColpoda species. Tetrahymena Colpoda Tetrahymena -3.489 0.0015 -3.856 <0.001 -4.774 <0.001 -2.717 0.009 -2.968 0.0086 -4.578 <0.001 Colpoda -2.077 0.0462 -6.669 <0.001 -2.713 0.0101 -2.561 0.0146 -0.024 0.9810 -2.487 0.0196

Preselection Monoculture Competition Focal Species Tetrahymena Maryna Tetrahymena Maryna Tetrahymena Maryna Tetrahymena -4.111 <0.001 -2.998 0.005 -4.263 <0.001 -1.498 0.143 -2.541 0.0158 -1.570 0.126 Maryna 0.309 0.760 -3.460 0.002 -1.895 0.066 -1.857 0.072 -0.302 0.767 -3.684 <0.001

Preselection Monoculture Competition Focal Species Maryna Paramecium Maryna Paramecium Maryna Paramecium Maryna -3.656 <0.001 -1.042 0.304 -6.131 <0.001 -1.428 0.155 -4.127 <0.001 -2.345 0.0204 calculatingParamecium niche overlap3.918 <0.001 and fitness-1.489 difference.144 2.541 for the0.012 species-3.642 pair,<0.001 error-0.735 within0.463 the alpha-4.618 parameter<0.001 estimates was propagated. Niche overlap had a non-significant increase after selection in monoculture and selection in competition. Changes in fitness differences suggest a switch in the competitive dominant from Colpoda as the initially dominant competitor to T etrahymena, but this effect was not significant (Fig. 4.3). Evolution appears to have shifted species from stably coexisting due to small fitness difference and relatively little niche overlap to a region where species are on the knife edge between coexistence and exclusionary dynamics (Fig. 4.10A).

4.3.2 T etrahymena vs Maryna

Again, multiple regression indicated the extent of density dependence in the pre-capita growth rate of species. Response surface experiments tended to show density dependent effects on the per- capita growth rate of intra and interspecific competitors both before and after selection. However, T etrahymena did not have a significant effect on Maryna’s per capita growth in any condition. Additionally, after selection Maryna no longer had a significant effect on T etrahymena (Table 4.3.1). Parameter estimates for the Lotka-Volterra models converged, but the bootstrapped confidence intervals tended to be large, and differences between parameter estimated before and after competition could not be detected. We found marginal support suggesting Maryna evolved to be less suppressed in monoculture as compared to competition, which is consistent with the linear model results. Additionally, consistent between the statistical models is the effect of Maryna on

T etrahymena (a21) decreased as a result of both species evolving in monocultures. However, in- consistent with the linear model is that T etrahymena evolved to suppress Maryna (a12) in both monoculture and competition (Fig. 4.4).

46 Intraspecific effects Interspecific effects A B 0.105 0.168 Maryna 11 12 0.003 a a 0.005 Effects on 0.002 0.001 0 Before 1 sp. 2 sp. Before 1 sp. 2 sp. Selection environment Selection environment

C D 0.9 0.485 22 21 Tetrahymena a a 0.006 0.005 0.002 Effects on 0 Before 1 sp. 2 sp. Before 1 sp. 2 sp. Selection environment Selection environment

Figure 4.4: Parameter estimates and 90% conﬁdence intervals generated by bootstrap analysis for a Lotka-Volterra competition model. Maryna is species 1 and T etrahymena is species 2.

47 A B 1.0 2

0.8 1

0.6 0

0.4

Niche overlap Niche -1

0.2 Log(fitness difference) Log(fitness

-2 Sp.1 wins Sp.1 wins Sp.2 wins 0.0 Before 1 sp. 2 sp. Before 1 sp. 2 sp.

Selection environment Selection environment

Figure 4.5: Parameter estimates and 90% conﬁdence intervals generated by bootstrap analysis for a Lotka-Volterra competition model. Maryna is species 1 and T etrahymena is species 2. B. Negative values indicate a competitive advantage for Maryna, and positive values indicate a competitive advantage for T etrahymena.

When calculating niche overlap and fitness difference for Maryna and T etrahymena, error within the parameter estimates was propagated. Niche overlap had a non-significant increase after selection in monoculture and selection in competition. Fitness differences suggested a switch in the competitive dominant as a result of selection in competition from Maryna to T etrahymena, but this effect was not significant for species selected in competition (Fig. 4.5). Evolution during competition appears to have shifted species from a region in which Maryna excludes T etrahymena to a region in which T etrahymena excludes Maryna, though these changes are non-significant (Fig.4.10B). When considering the temporal dynamics of the populations over the selection regime, the dynamics of T etrahymena do not appear to fit with the narrative that T etrahymena is driving Maryna to extinction. Monoculture and competition densities for T etrahymena are initially very similar, whereas over time, T etrahymena has lower densities in competition compared to monocultures (Fig. 4.6A). This could be because either T etrahymena is suppressed by Maryna in the competition treatment, T etrahymena quickly evolved to have higher densities in the monoculture

48 A B

2000 2000 Selection environment 1 species 2 species 1500 1500 density density 1000 1000 Maryna Tetrahymena 500 500 0 0

0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35

Days Days

Figure 4.6: Population dynamics of T etrahymena and Maryna over the selection period. A. T etrahymena density through time. B. Maryna density through time. treatment, or historical effects of competition resulted in lower T etrahymena densities even after they evolved to be less suppressed by Maryna. However, the population dynamics of Maryna seem more in line with the parameterized models that suggest Maryna may be driven extinct; initially, the presence of T etrahymena does not have a significant effect on Maryna abundances, however over time, Maryna densities in monoculture are higher than Maryna densities in competition (Fig. 4.6B). Again, this could be due to three reasons, evolution of the monoculture line to have higher densities, evolution of T etrahymena to more effectively suppress Maryna, or evolution of Maryna to be suppressed by T etrahymena.

49 Preselection Monoculture Competition Focal Species Tetrahymena Colpoda Tetrahymena Colpoda Tetrahymena Colpoda Tetrahymena -3.489 0.0015 -3.856 <0.001 -4.774 <0.001 -2.717 0.009 -2.968 0.0086 -4.578 <0.001 Colpoda -2.077 0.0462 -6.669 <0.001 -2.713 0.0101 -2.561 0.0146 -0.024 0.9810 -2.487 0.0196

Table 4.3: t-statisticsPreselection and P-values fromMonoculture multiple linear regressionCompetition of the eﬀect of each speciesFocal Species densityTetrahymena on the per-capitaMaryna growthTetrahymena rate of the focalMaryna species. Tetrahymena Maryna Tetrahymena -4.111 <0.001 -2.998 0.005 -4.263 <0.001 -1.498 0.143 -2.541 0.0158 -1.570 0.126 Maryna 0.309 0.760 -3.460 0.002 -1.895 0.066 -1.857 0.072 -0.302 0.767 -3.684 <0.001

Intraspecific effects Interspecific effects A B 0.08 0.005

0.004 0.06 Maryna

11 0.003 12

a a 0.04

0.002 0.02

Effects on 0.001

0.000 0.00 Before 1 sp. 2 sp. Before 1 sp. 2 sp. Selection environment Selection environment

0.07 C D 0.06 8e-04 0.05 6e-04 0.04 22 21 Paramecium a a 0.03 4e-04

0.02 2e-04 0.01 Effects on 0.00 0e+00 Before 1 sp. 2 sp. Before 1 sp. 2 sp. Selection environment Selection environment

Figure 4.7: Parameter estimates and 90% conﬁdence intervals generated by bootstrap analysis for a Lotka-Volterra competition model. Maryna is species 1 and P aramecium is species 2.

50 4.3.3 Maryna vs P aramecium

Response surface experiments tended to show density dependent effects on the per-capita growth rate of intra and interspecific competitors both before and after selection (Table 4.3.2). Surprisingly, the effect of Maryna on P aramecium before selection and after selection in monoculture was positive. Additionally, P aramecium did not have a significant effect on itself or Maryna before selection, and did not affect Maryna after both underwent selection in monoculture, but did effect Maryna after both underwent selection in competition. Finally, the positive effect of Maryna on P aramecium disappeared after both evolved in competition. Parameter estimates for the Lotka-Volterra models converged, but differences between parameter estimated before and after competition could not be detected except the intraspecific effect of

P aramecium (a22) decreased in both competition and monoculture (Fig. 4.7C). This is the opposite of what the linear models find. Parameter estimates for the effest of Maryna on P aramecium were constrained to be negative and likely are very near zero (Fig. 4.7D) due to the positive effect of Maryna on P aramecium observed in the linear models (Table 4.3.2). When calculating niche overlap and fitness differences for the species pair, error within the parameter estimates was propagated. Niche overlap had a non-significant increase after selection in monoculture and selection in competition (Fig. 4.8A). Fitness differences additionally increased non-significantly after selection in monoculture and competition, maintaining P aramecium as the competitive dominant (Fig. 4.8B). Evolution in competition and monoculture appear to have shifted species from a region of coexistence to a region were P aramecium will outcompete Maryna, though these changes are non-significant (Fig. 4.10C). When considering the temporal dynamics of the populations over the selection regime, the density of P aramecium in monoculture was not significantly different from the density of P aramecium in competition, although the density of P aramecium in competition was consistently higher, suggesting some positive effect of Maryna on P aramecium or perhaps some level of intraguild predation (Fig. 4.9A). The density of Maryna was consistently significantly lower in competition than in monoculture(Fig. 4.9B). This result is surprising because the response surface experiment did not detect a significant effect of P aramecium on Maryna before selection.

51 A B 1.0

1.0

0.8

0.5 0.6

0.4

Niche overlap Niche 0.0 Log(fitness difference) Log(fitness

0.2

-0.5 Sp.1 wins Sp.2 wins

0.0 Before 1 sp. 2 sp. Before 1 sp. 2 sp.

Selection environment Selection environment

Figure 4.8: Parameter estimates and 90% conﬁdence intervals generated by bootstrap analysis for a Lotka-Volterra competition model. B. Negative values indicate a competitive advantage for Maryna, and positive values indicate a competitive advantage for P aramecium.

52 200 A Selection environment B 1 species 2 species 1500 150 density 1000 density 100 Maryna Paramecium 500 50 0 0

10 15 20 25 30 35 40 45 10 15 20 25 30 35 40 45

Days Days

Figure 4.9: Population dynamics of P aramecium and Maryna over the selection period. A. P aramecium density through time. B. Maryna density through time.

53 1.0 1.0 1.0

A B C wins wins wins

0.5 0.5 0.5Paramecium Tetrahymena Tetrahymena

0.0 0.0 0.0 Fitness Fitness difference Fitness difference Fitness difference

-0.5 -0.5 -0.5 Selection environment Before wins wins wins wins wins 1 species 2 species Maryna Maryna Colpoda

-1.0 -1.0 -1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Niche overlap Niche overlap Niche overlap

Figure 4.10: Change in niche overlap and fitness difference in three different pairs of species. Unshaded areas are where species coexist, and grey areas are where one species will outcompete the other. A. Colpoda andT etrahymena evolving in pairwise mixtures (2 species) or monocultures (1 species). B. Maryna and T etrahymena have changing competitive outcomes as a result of evolution. C. Maryna and P aramecium coexist initally but evolve to exclude each other in both pairwise mixtures and monocultures.

54 Paramecium invasibility Maryna invasibility 3.5 0.8 3.0 0.6 2.5 2.0 0.4 1.5 Per capita growth Per capita 1.0 0.2 0.5 0.0 0.0

Before 2 sp. 1 sp. Before 2 sp. 1 sp.

Selection Environment Selection Environment

Figure 4.11: Invasibility analysis of P aramecium and Maryna before and after selection. Protists were selected in either one species mixtures (mono) or two species mixtures (comp). If each species can increase from low density (positive growth rate) in the presence of its competitor, the species theoretically should coexist. If a species’ growth rate is not positive in the presence of the competitor then it is the competitive subordinate. The box-plots indicate medians and 50% of the data, whiskers indicate 95% conﬁdence intervals.

55 4.3.4 Invasibility

As an alternative to estimating niche overlap and ﬁtness diﬀerences with response surface experiments, we performed invasibility analyses for P aramecium and Maryna both before and after selection. Before selection, P aramecium on average cannot invade the Maryna population, suggesting that Maryna is the competitive dominant and these species cannot coexist. However, after selection in both monoculture and competition P aramecium can increase from low density in the respective Maryna population (Fig. 4.11A). When Maryna is introduced at low density into the P aramecium monoculture before selection, it can on average increase in abundance. However, after selection in both monoculture and competition, Maryna does not have a growth rate greater than zero and therefore cannot invade the population suggesting that P aramecium is in fact the competitive dominant after these selection regimes (Fig. 4.11B).

4.4 Discussion

Ecologists have often assumed that the evolution of niche partitioning is the primary mechanism for facilitating coexistence among competitors (e.g. Hutchinson 1959). While on the other hand, they have been perplexed by the apparent high degree of overlap or even convergence in niches (Hubbell 2001, Leibold and McPeek 2006). Despite the preoccupation with changes in niche overlap, fitness differences can have profound effects on coexistence (Chesson 2000). Convergence in fitness can facilitate coexistence but divergence would degrade species ability to co-occur even when niche differences are present. To fully understand long term population dynamics of species, we must understand how ongoing evolution changes both niche overlap and fitness differences among competitors. Here, we demonstrate with competition experiments and models parametrized with response surface experiments that pairwise evolution in a laboratory does not facilitate coexistence among species. When results for three separate trials are taken together, evolution among competitors appears to be decreasing the probability that species will coexist rather than facilitating coexistence. Specifically, we found that species pairs tended to increase or not change their niche overlap as a result of selection in competition, contrary to what most models predict. Furthermore, fitness differences among species pairs tended to change in response to evolution; species pairs always

56 ended up with larger fitness differences after evolution. Consequently, both changes in niche overlap and fitness differences decreased the ability for species to coexist. Although broad patterns seemed generally consistent across species pairs (large changes in fitness differences, smaller changes and increases in niche overlap), within species pairs, different experimental components seemed to have different results. One would expect that the linear models of the response surface experiment and parameter estimates of species effects would be consistent, however they did not always match. Additionally, predictions about competitive outcomes were not always observed in the temporal dynamics or invasibility analyses (Fig. 4.11). One explanation for these inconsistencies seems to be the large error estimates in model parameters, thus their ability to predict competitive outcomes should be taken with caution. However the most incongruous results were found among a species pair that had a positive interaction of one species on the other, this is not competition in the canonical sense but could reflect intraguild predation. All though competition could be occurring between the pair for the bacterial resources, the effects of competition on their population growth are obscured by the positive interaction, and so niche differences and fitness differences are difficult to intrepret. The model presented in Chapter 2 as well as other models generally predict that evolution among competitors should result in niche divergence. Additionally, the model in Chapter 2 predicts that competing species should generally evolve to have greater, rather than lesser, fitness differences. In this study, although we observed increases in fitness differences, there were also increases (or no change) in niche overlap. In most models of evolution of competitors, no independent consideration is given to fitness differences, and convergence in the niche only occurs when there is considerable constraint of the niche and species are unable to diverge from each other. Consequently, in such environments, there is a race between species to the optimal niche, and whoever gets there first will be conferred with fitness advantages that will most likely result in the exclusion of the more slowly evolving species. For P aramecium with Maryna and to a lesser degree, T etrahymena with Maryna, evolution appeared to operate consistently when the selection environment included only conspecifics, or het- erspecifics and conspecifics. This seems to imply that the environmental condition the protists were in was a stronger selective force than competition with a heterospecific. This also seems consistent with the predictions in Chapter 2 of an environment with only a narrow range of resource availabil-

57 ity. But although protist are filter feeders and would appear to not discriminate on which bacteria they are eating, protist species are known to discriminate mainly based on bacterial size (Jürgens and Matz 2002), but also, bacterial motility, physicochemical surface characters and toxicity (Sherr and Sherr 2002). Therefore if protists could evolve offensive traits (Cortez and Weitz 2014), they could have capacity for niche differentiation through specialization. However, evolutionary constraint could be playing a role here, these protists perhaps lack the genetic variation or the genetic architecture to differentiate along such axes. But how common is it for niche evolution to be constrained, and therefore how often will species evolution result in convergence rather than divergence? Or phrased differently; are there unfilled niches in natural communities? If niches in an environment are filled, even with weak competitors, then niche convergence may be a relatively common even if there are many niches. These results lend some evidence to Hubbell’s (2001) neutral theory of biodiversity that suggests that diversity is maintained in communities simply through inherently unstable neutral dynamics were species are ecologically equivalent. However, since convergence in the niche should be coupled with divergence of fitness differences, ecologically equivalent species are unlikely to co-occur and the rate of competitive exclusion increase. There are limitations to this study’s ability to detect niche overlap since the response surface experiment and invasibility analyses implemented here are only able to detect niche overlap at the time of measurement (temporal dynamics that occur on scales larger than 4-days will not be captured). These experiments should detect population level effects caused by partitioning of resources, spatial structure and parasites. However, we were not able to homogenize the environment temporally, and thus species could have partitioned temporal niches either through relative non-linearities in growth rate or through the storage effect without detection. In fact, two species used here, Maryna and Colpoda, form resting cysts with low metabolic demands, thus they have a mechanism for buffering against adverse conditions. Additionally, the conditions in these experiments may be the conditions under which the temporal storage effect can evolve since nutrient inputs were scheduled and predictable (Snyder and Adler 2011). Not only are the timescales of ecological interactions important, but the time scales of evolutionary effects can have implications for communities as well. Although we saw evolution over six weeks under different selection regimes, it’s unclear how evolutionary processes proceeded. Did

58 evolutionary responses occur in the first few generations until genetic variation was quickly used up? Or was there a mutation selection balance that resulted in a slow march toward evolutionarily stable strategies? Such different evolutionary processes would likely have different consequences for ecological and evolutionary dynamics in complex communities, and the speed at which they reach stable dynamics.

4.4.1 Conclusions

The results of this study raise a few major concerns with assumptions about evolution among competitors. First, niche space may be limited in most communities and thus species may be more likely to converge in their niches rather than diverge; thus decreasing coexistence between species. Second, ﬁtness diﬀerences can change as a result of evolution among competitors and are likely to increase; thus decreasing coexistence between species. Consequently, sympatric evolution appears to be an antagonistic force for diversity in communities, thus aggravating the paradox of diversity.

59 CHAPTER 5

EVOLUTION OF BACTERIAL CONSUMPTION IN COMPETING PROTISTS

5.1 Introduction

The concept of ’the niche’ is at the cornerstone of competition theory (Chase and Leibold 2003). In this context, the definition of ’the niche’ requires a density dependent process by which the niche is ’used up’ and thus regulates the density of a focal species and its competitor (Amarasekare 2003). Consequently, competition is an indirect effect of one species on another through a shared resource, and inherently, how each species uses the shared resource determines the outcomes of competition. Despite this elegant conceptual basis, measuring the direct effects of a species on its resource can be very difficult. It is often much easier to infer the degree of niche overlap between competitors using parameterized models, rather than measuring the loops of direct effects of species interacting through their resources (e.g. Law and Watkinson 1987). And while measuring indirect effects has the advantage of precision in capturing a multidimensional niche phenomenon, it is not mechanistic and obscures the true nature of these species interactions (Letten et al. 2017). Ultimately, these indirect effects between competing species will act as selective pressures on what, when, and how species use resources, which can lead to the evolution of traits associated with competitive interactions. But without being able to observe resource utilization directly, it is nearly impossible to understand the mechanism by which selection is acting. Once the mechanisms of evolutionary processes among competitors is understand, ecologist can begin to disentangle the contribution of evolutionary processes relative to ecological processes in filling niches in communities, thus informing our understanding of community assembly and the functional structure of communities (Ackerly 2003, Kraft et al. 2007). Most theory of species evolving in competition suggest that whenever possible species will evolve to differentiate in their resource use by specializing (Macarthur and Levin 1967, Abrams 1983, terHorst et al 2010). This results in niche differences that causes a species to limit their own population growth more than they limit the growth of a competitor and thus will stably

60 coexist. But some dimensions of this body of theory suggests that limited resource spaces can lead to convergence in resource use between two species because there is no capacity for divergence in resources (Abrams 1986, Fox and Vasseur 2008, terHorst et al. 2010). In such environments, small changes in a species niche result in large changes in the viability of the population due to the limited ecological options (Chapter 2). Thus we predict that large changes in a species niche as a result of evolution in an environment with unfilled niches should have relatively small effects on the average fitness of the population. Whereas, in an environment with limited ability for niche differentiation, any change in the niche will come with a large cost or benefit to fitness. Direct observations are needed to understand how competitors evolve while simultaneously considering their effects on shared resources and indirect effects on their competitors. The intra- and interspecific interactions strengths between a species pair can be measured by manipulating the density of competitors, and such parameters provide an approximation of the degree to which species share resources as well as how efficiently species compete for limiting resources (Godoy and Levine 2014, Kraft et al. 2015). But the direct effects of species on the relevant axes of their niche is often difficult to observe, thus concealing the mechanisms of competition and subsequent selection. In Chapter 4 of this dissertation, we applied parameterized Lotka-Volterra competition models to follow the changes in species interactions as a result of evolution in pairs of protist species, which compete for limiting bacteria. But the mechanisms of competition were not identified. Recent methods have been developed to describe bacterial assemblages using high-throughput molecular methods (e.g. Paisie et al. 2014) and flow cytometry (Sherr and Sherr 2002, First et al. 2012). Because we can now quantify bacterial species presence and abundance in these communities, we can also determine the effects of protist predation on these diverse bacterial communities, which could inform how protists are interacting with each other indirectly through overlap in their bacterial ’resource niche’. The tractability of protist-bacteria systems allows us to investigate how evolution acts indirectly among competing predators through changing effects on the bacterial community. Protists have short generations times and past studies have demonstrated that competitive ability can evolve (terHorst 2011, Miller et al. 2014). However, these studies are entirely phenomenological and do not contain mechanisms on the underlying resource dynamics. Now, using modern microbial methods, we can provide a mechanistic explanation of how the protists interact through effects on shared bacterial resources.

61 In the work presented here, we allowed four different pairs of protists to evolve in monoculture and pairwise competition and measured the effects on the bacterial community both before evolution and under the different selection environments. We measured the effect of protists on both the abundance and composition of bacterial communities. We determined how evolution in response to intra and interspecific competition changed the ability of protists to consume resources as well as which resources a competitor will eat. Although one might expect that species evolve to use less similar bacterial communities, we found no evidence for niche differentiation in the composition of bacterial communities. Additionally, we found little evidence that protist species evolved in their ability to drawn down the bacterial community (a potential mechanism for increased competitive ability), except for one species pair.

5.2 Methods 5.2.1 Bacterial Broth

Bacterial inoculates for a standardized broth were prepared by adding 500 mg of Tetramin fish food to 500 mL of sterile deionized water. The solution incubated for 24 hours at room temperature on a stir plate. Bacteria that colonized this broth was from the air and from the fish food. Then the broth was pipetted in 1mL aliquots into microcentrifuge tubes using sterile techniques, and frozen at 80◦C until needed. All subsequent broth media was prepared by combining 700 mL of deionized water with 700 mg fish food. This solution was then autoclaved to sterilize and then cooled to room temperature in a water bath. Once the solution had cooled, a 1 mL aliquot of the previously described bacterial inoculate was thawed in a water bath and added to the sterile solution. This broth incubated at room temperature (∼23◦C) for 24 hours. Attempts to use bacteria from the natural habitats of these species resulted in failed protist growth. Standardized broth was prepared using these methods nine times over the course of the experiments to compare protist consumption, but there was variation in the communities produced. It appears that the cell density gradually got higher over the course of the experiment (Fig. 1); the cause for this is unknown, as the growth conditions (e.g., light, temperature) were constant. Species composition, however, was relatively similar across the different broths (Fig. 2), however again some broths were clearly different (e.g., the 5th broth).

62 Figure 5.1: Number of particles counted in 11 µL by a flow cytometer as a proxy for bacterial cell density in the 9 standardized bacterial broths (RSE), prepared in the same way at different time points. The box-plots indicate medians and 50% of the data, whiskers indicate 95% confidence intervals.

Figure 5.2: NMDS of bacterial community composition gives a two-dimensional visual representation of the compositional similarity of the 9 standardized bacterial broths (RSE), prepared in the same way at diﬀerent time points. Here, the minimum stress after 20 runs was 0.1754653 for k = 2.

63 5.2.2 Protist Cultures

Water samples were taken from the shallows of eight ponds in the Apalachicola National Forest, south of Tallahassee, FL. Samples were immediately transported to the laboratory where individual cells of 4 species were isolated from other pond flora and fauna. Several cells of each species were gathered from individual ponds to create monoculture lines, with replicate lines created by sampling different ponds (between three and eight ponds per species). Each line of each species was maintained in a separate 50 mL macrocentrifuge tube and raised on 10 mL of the standardized bacterial broth. Protist cultures were refreshed every one to two months by inoculating 10 mL of newly prepared standardized bacterial broth with 0.1-1ml of the previous culture. Protists obtained with this method were single celled, free-living organisms that swam by beat- ing cilia and consumed bacteria in their habitat. Four species were used in this experiment were identified to genus level by sequencing their 18s rRNA gene region and placing them on a well resolved phylogeny (Chapter 3). All four species were among the Intramacronucleata clade of cil- iated protists, two species were among the Colpodea class; Colpoda sp. (Colpoda) and Maryna sp. (Maryna), and two were in the Oligohymenophorea class; T etrahymena sp. (T etrahymena), and P aramecium sp. (P aramecium). Species had life spans on the order of six hours and reproduced predominantly through asexual cell division, however T etrahymena was seen conjugating (sexual reproduction) when resources became scarce in their environments. P aramecium were the biggest (∼2000 µm2), then Colpoda (∼900 µm2), and T etrahymena (∼400 µm2), and smallest was Maryna (∼300 µm2).

5.2.3 Preparation of Stock Lines for Experiment

64 through by the addition of sterile water. Beads were added to the super-filtrate to prevent protists from deforming against the mesh filter. After adding the protist culture to the filter, an equal volume of sterile DI water was slowly added to prevent all the water from draining out. When 50 mL was left above the filter the super-filtrate was obtained and used for subsequent experiments. Although this method tends to decrease protist abundances by up to half, it appears to remove most bacteria, so negligible amounts of nutrients and bacteria are added to the experiment.

5.2.4 Selection Experiment

Four selection experiments were run at different times with four species pair combinations; (1) Colpoda with Maryna, (2) Colpoda with T etrahymena, (3) Maryna with T etrahymena and (4) Maryna with P aramecium. For each species in each pair, five replicates of two different selection environments were implemented: pairwise competition or alone in monocultures. To initiate these selection environments, first a uniform volume and density of standardized broth was to all tubes such that the final concentration of Tetramin in each tube was 0.12 mg/mL. To initiate the pairwise selection environment, 4 mL of the filtered culture of each species and 8 mL of sterile water was added to the falcon tube for a total of 21 mL of fluid. To initiate the monoculture selection environment, 4 mL of the filtered culture of one species and 12 mL of sterile water was added to the falcon tube for a total of 21 mL of fluid. Tubes were capped and kept on a shaker tray in the incubator for six weeks. Tubes were refreshed every four days by shaking vigorously, removing 10 mL of fluid and replacing that fluid with fresh standard broth diluted to 0.25 mg Tetramin/mL to maintain similar nutrient densities and decrease the accumulation of waste products. At the end of the selection period, for each of five replicates, 10-50 cells were isolated and used to initiate monoculture lines. For each of the five replicates of each experiment four lines were generated; both species evolved in either a monoculture or in selection. These lines were then grown to high density and prepared for subsequent experimentation with the above filtration methods, keeping each replicate separate.

5.2.5 Measuring Eﬀect of Protists on the Bacterial Community.

Samples to determine the eﬀects of protists on bacteria were taken before and after selection for each of ﬁve separate replicates. For each species line, before and after selection, I combined 5 mL of 0.5 mg/mL standardized bacterial broth, 8 mL of the focal species line and 8 mL of sterilized water

65 (this resulted in the same food concentration and volume as in the selection experiment, but double the protist density). Separate 1 mL initial samples were taken from these tubes for flow cytometry and high-throughput sequencing and immediately preserved. The remaining 20 mL of fluid were stored on the shaker tray in the growth chamber. For Colpoda with Maryna and T etrahymena with Colpoda, tubes were again sampled for flow cytometry and sequencing about 9 hours later. For T etrahymena with Maryna, tubes were sampled 96 hours later, and for P aramecium with Maryna tubes were sampled at 9 and 96 hours.

5.2.6 Flow Cytometry

To prepare for ﬂow cytometry, 53 µL of Formaldehyde was added to 1 mL of each sample to create a 2% formaldehyde solution, which was then stored at −80◦C until processing (First et al. 2012). Tubes were transported to the FSU College of Medicine Flow Cytometry Lab on ice. Samples were thawed and ﬁltered through 50 µm nylon mesh, and mixed with 2 µL of Syto 13 nucleic acid stain (del Giorgio et al. 1996). Samples sat covered for 10-30 minutes to allow the stain to penetrate the cells, and then processed through a BD FACS Canto two laser instrument (BD Biosciences, San Jose, CA) for 30 seconds at a rate of 22 µL/min, and events were counted.

5.2.7 Metagenomic Sequencing

For high throughput sequencing, 1 mL of fluid was pelleted at 10,000 x g for 10 min. The top 0.8 mL were removed and samples were stored at −80◦C. To extract DNA, samples were thawed, and Mo Bio’s PowerViral DNA Isolation kits were used to process the samples according to the manufacturers protocol. Samples were then shipped to Argonne National Lab for library preparation and Illumina Miseq itag sequencing. 16S rRNA gene amplification was performed in triplicate using archaeal and bacterial primers 515F and 806R, which targets the 104 base pair V4 region of E.coli in accordance with the protocol described by Caporaso et al. (2011b, 2012) and used by the Earth Microbiome Project (http://www.earthmicrobiome.org/emp-standard-protocols/16s/). Reverse primers contained a sample-specific barcode incorporated into each amplicon, allowing multiplexing of samples for high-throughput sequencing with the Argonne National Lab MiSeq (Illumina, San Diego, CA). Between 112 and 27,801 (mean 14603) paired-end sequences of 250- base length were generated from each sample, three samples had less than 1,700 reads and were dropped from subsequent

66 analyses. Forward and reverse sequences that had paired-end reads were stitched together to generate longer contiguous sequences spanning nearly the entire V4 region. Raw sequences were de-multiplexed, quality filtered and clustered into operational taxonomic units (OTUs) using 16S rRNA gene sequence similarity QIIME. The resulting representative sequences set were aligned using PyNAST and given a taxonomic classification using RDP (Cole et al. 2009). Representative sequences were checked for chimeras and phiX contamination. The resulting OTU table was filtered to keep only OTUs that had at least two observations, and then normalized using cumulative sum scaling in R with the MetagenomeSeq package (Paulson et al. 2013). To analyze changes in microbial community composition due to selection, we compared the community matrices of each sample using non-metric multidimensional scaling ordination (NMDS) to quantify bacterial communities associated with each sample with 20 runs and 2 dimensions (see, e.g., Miller et al. 2010, Bokulich et al. 2012) using the Vegan package version 2.4-3 in R version 3.3.3 (R Core Team 2017). Permanova (adonis) was used to test for significant differences in the variance in microbial communitie composition between different treatment cultures. Soreson’s index of dissimilarity was calculated in R with the Betapart package (Baselga 2010, Baselga and Orme 2012) to quantify the niche overlap between each pair of communities using presence absence data. While all treatments were initiated with the standardized bacterial broth, some differences in initial bacterial communities occurred due to unknown differences in culture conditions (Fig. 5.2) or through bacterial contamination from the protozoa inoculates. To account for this initial dissimilarity, I divided initial dissimilarity between communities from final dissimilarity as my measure of niche overlap, between two communities. It is important to realize the limits of these methods. (1) They do not directly quantify which species of bacteria are being consumed by protists, but rather which species protists are not consuming. I infer what protists do consume, by comparing bacterial communities of protists to control bacterial communities that did not experience predation. (2) Inferences about the bacterial community assume minimal indirect competitive effects among the bacteria. (3) It is possible that bacteria will evolve to resist predation, however protist effects on bacterial communities were always measured from a bacterial broth initiated from cryopreserved bacterial communities. Therefore, even if bacteria in the protist selection environment evolved, protist evolution was always compared with a standardized broth.

67 5.3 Results 5.3.1 Eﬀects on Bacterial Abundances

In most experiments, the presence of a competitor did not significantly change the effect of protists on the abundance of bacteria. Colpoda and Maryna lines grown in pairwise competition with each other or in monoculture did not significantly differ in their effects on bacterial abundance after 9 hours (Fig. 5.3A). However, a second pair of species shows significant evolution: T etrahymena selected in competition with Colpoda had a significantly smaller effect on the bacterial abundances than T etrahymena selected in monoculture (P = 0.037) after 9 hours (Fig. 5.3B). Protists selected in pairs of T etrahymena and Maryna or monoculture all resulted in fewer bacteria after selection, relative to the control abundance. However, overall, there were no differences among the effects of lines on bacterial abundance after 96 hours of protist consumption (Fig. 5.3D). Protists selected in pairs of P aramecium and Maryna or monoculture had no effect on bacterial abundance after 9 hours (Fig. 5.3C), but effects emerged after 96 hours of bacterial consumption. P aramecium reduced bacterial abundance more than Maryna across selection lines (P < 0.01), but there were no differences between selection lines of the same species (Fig. 5.3E).

5.3.2 Eﬀects on Bacterial Community Composition

NMDS analysis of bacterial abundances revealed differences in bacterial community composition where flow cytometry cell counts alone did not. All permanovas were highly significant, showing strong effects of time and treatment, and the interaction between the two showed that different treatments changed in different ways (Table 5.1). However, time tended to affect bacterial composition was more strongly than the environment that a protist experiences selection. Time zero communities tended to group together in NMDS plots (Fig. 5.3; grey ellipses), indicating starting communities were similar. After protists had time to consume bacteria, bacteria-only communities tended to be the most different from the other treatments. NMDS showed differences between the communities associated with species that evolved in monoculture or pairwise competition, however care was needed to distinguish the causes of differences in bacterial communities. Bacterial composition often differs between selected protist lines, but these differences appear to be driven by initial differences in bacterial communities due to bacterial contamination from protist culture rather than evolution of protists in their effect on the

68 9 hours A 16000 10000 Change in bacteria in Change 4000

B 1 sp. 2 sp. 1 sp. 2 sp. Colpoda Maryna

96 hours B D 15000 500 0 5000 Change in bacteria in Change bacteria in Change -1000 -5000 B 1 sp. 2 sp. 1 sp. 2 sp. B 1 sp. 2 sp. 1 sp. 2 sp. Tetrahymena Colpoda Tetrahymena Maryna

C E 10000 2000 0 0 Change in bacteria in Change bacteria in Change -2000 -15000 B 1 sp. 2 sp. 1 sp. 2 sp. B 1 sp. 2 sp. 1 sp. 2 sp. Maryna Paramecium Maryna Paramecium

Figure 5.3: Change in the density of bacteria approximately 9 and/or 96 hours after protists were added for diﬀerent experimental species pairs in two selection environments. On the x-axis, ’B’ is where no protists were added and serves as a baseline for determining the eﬀects of protists on the bacterial community, otherwise it indicates the selection environment that a species evolved in; either alone (1 sp.) or in pairwise competition (2 sp.). (A) Colpoda with Maryna at 9 hrs. (B) T etrahymena with Colpoda at 9 hrs. (C) Maryna with P aramecium at 9 hours. (D) T etrahymena with Maryna at 96 hours. (E) Maryna with P aramecium at 96 hours. Box-plots are as described in Fig. 5.1.

69 Table 5.1: F-statistics and P-values for permanovas of the eﬀects of protists on bacterial composition over time. Time is 0 hours or 9 hours, treatments are bacterial broth, and species lines. After selection, each species has a line selected in monoculture and in pairwise competition.

RSE 1 Before selection 2 After selection 3 Before selection 4 After selection Species Colpoda and Maryna Colpoda and Maryna Colpoda and Tetrahymena Colpoda and Tetrahymena Time 15.2528 0.001 22.986 0.001 15.731 0.001 28.0995 0.001 Treatment 13.1292 0.001 14.7055 0.001 43.998 0.001 6.9874 0.001 Interaction 4.5024 0.005 4.2882 0.001 4.728 0.002 2.8706 0.002

RSE 5 Before selection 6 After selection 7 Before selection 8 After selection Species Maryna and Tetrahymena Maryna and Tetrahymena Maryna and Paramecium Maryna and Paramecium Time 50.571 0.001 52.398 0.001 33.007 0.001 29.2496 0.001 Treatment 62.3 0.001 7.346 0.001 31.053 0.001 6.2129 0.001 Interaction 20.272 0.003 4.002 0.001 13.884 0.001 2.3985 0.002

bacterial community (e.g. diet preference). For example, in RSE 4: Colpoda selected in monoculture had a different bacterial community than Colpoda selected in pairwise competition with T etrahymena after 9 hours. This could have been a result of selection, however, Colpoda selected in monoculture also had similar differences in the bacterial community than Colpoda selected in pairwise competition when it was initiated at time zero. Comparing dissimilarity (Sorenson’s index) between two communities before and after protist consumption was used to determine if evolution changed the protist’s effect on a bacterial community. To determine if selection environment had a significant effect on how the protists used resources, I calculated the dissimilarity between bacterial communities associated with different selection lines of the same species after protist consumption divided by the initial differences in bacterial communities as a measure of the change in the niche of a species between different selection lines. Values were all greater than one, indicating that dissimilarities tended to increase (Fig. 5.5). However, no measures were significantly greater than one, suggesting that differences between treatments were a result of differences in the initial bacterial community. Further, there were no consistent differences between 9 hour and 96 hour samples.

70 RSE 2 RSE 4 A B T0 T0

Colpoda 1 sp. 1.0 Tetrahymena 1 sp. 0.6 Maryna 1 sp. Colpoda 1 sp. T0.MCA Colpoda 2 sp. Tetrahymena 2 sp. T7.MAC Maryna 2 sp. T0.MCA T0.SBC Colpoda 2 sp. T7.MAC T0.MCA 0.5 T7.MAC

0.2 T0.MAC T0.MBC T0.SCA T7.CBC T7.MBC T0.MAC T7.CBC T0.SBC T0.SCA T0.MBC T7.MAC T0.SBC T0.MBC T7.MCA 0.0 T0.MBC T7.CBC

NMDS2 T7.MBC NMDS2 T7.SCA T7.SBC T7.MBC T7.SBC T0.MBC T7.MBC

-0.2 T7.SCA T7.MBC T7.MBC

T7.MBC -0.5 -0.6 -1.0

-0.5 0.0 0.5 1.0 -1.5 -1.0 -0.5 0.0 0.5 1.0

NMDS1 NMDS1 RSE 6 RSE 8

C 1.0 D

0.5 D4.MAC D4.CPM D4.MCA T0.MCA D4.CCA T0.MPM D4.MAC 0.5 D4.MCA T0.CCA D4.MCA D4.CCA D4.CCA T0.MCA D4.MCA T0.CPM

0.0 D4.CCA D4.MAC T0.CCA T0.MPM T0.MPM

T0.CAC 0.0 D4.MPM T0.CPM D4.CAC T0.MCA D4.CCA D4.MPM T0.CCA

NMDS2 D4.CCA NMDS2 D4.MCA

-0.5 T 0 D4.CPM Tetrahymena 1 sp. T0 -0.5 D4.CPM Maryna 1 sp. Paramecium 1 sp. D4.CAC Tetrahymena 2 sp. Maryna 1 sp. D4.CCA Paramecium 2 sp.

-1.0 Maryna 2 sp.

-1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 1.5

NMDS1 NMDS1

Figure 5.4: NMDS ordinations of the bacterial communities associated with different selection treatments. Ellipses are the 70% confidence intervals for each treatment. Grey ellipses are the bacterial communities at time zero for the different treatments. Light blue and light red ellipses are species that evolved in monocultures. Dark blue and dark red are species that evolved in pairwise competition. (A) Colpoda with Maryna at 9 hrs, stress= 0.1417709. (B) T etrahymena with Colpoda at 9 hrs, stress = 0.1316615. (C) Maryna with T etrahymena at 96 hours, stress = 0.09588695. (D) Maryna with P aramecium at 96 hours stress = 0.08369383.

71 RSE 2 RSE 4 A B 1.20 1.15 1.10 1.10 1.05 Niche evolution Niche evolution Niche 1.00 1.00 0.90 Colpoda Maryna Tetrahymena Colpoda

RSEx 6 RSEx 8 C D 1.25 1.3 1.2 1.15 1.1 1.05 1.0 Niche evolution Niche evolution Niche

9 hrs 0.9 4 days 0.95

Tetrahymena Maryna Maryna Paramecium

Figure 5.5: Niche evolution is estimated as the difference in the bacterial communities associated with lines selected in monocultures and pairwise competition, scaled by initial differences in the bacterial communities. Error bars are +/− 1 sd. The grey bar indicates the difference between bacterial communities associated with different selected lines neither increased nor decreased with time. two treatments was the same at time zero and 9 and/or 96hours. (A) Colpoda with Maryna at 9 hrs. (B) T etrahymena with Colpoda at 9 hrs. (C) Maryna with T etrahymena at 96 hours. (D) Maryna with P aramecium at 96 hours.

72 Additionally, the Sorenson’s index of dissimilarity was used as a measure of niche difference between two species. Differences in bacterial communities were divided by initial dissimilarity to account for initial dissimilarities in the bacterial communities. In three of the four cases, niche difference increased after selection. In one case, T etrahymena with Maryna, the two species converged in their niches in monoculture and in competition (Fig. 5.6). However, niche difference was never significantly different between protists selected in monoculture and protists selected in pairwise competition, suggesting that evolution of protist niches was not common in this experimental design.

5.4 Discussion

When understanding the ecological and evolutionary nature of resource competition, it is essential to consider each predator’s effect on their shared prey; and how selection changes effects on prey and therefore indirect effects on competing predators. In this experiment, protists had clear effects on the composition (Table 5.1) and often abundance (Fig. 5.3) of bacteria. Protists tended not to change in their effect on the abundance of a bacterial community due to selection (Fig. 5.3). But surprisingly, one species evolved to have a much smaller effect on the bacterial community in pairwise competition (Fig. 5.3B). In general, species also tended not to differentially evolve in their effect on the composition of bacterial communities as a result of selection environment (Fig. 5.5). Additionally, species tended not to evolve in their niche compared to before selection, except for one pair of species that seemed to converge in their niche in both monoculture and pairwise competition (Fig. 5.6). Taken together it seems that selection environment had little effect on the feeding habits of protists. Here the effect of protist presence on the bacterial community was intended to serve as a proxy for the niche of the protist. However, resource use is only one facet of the niche and our measurements do not include differences in temporal use greater than 9 or 96 hours. Although efforts were made to homogenize microcosms spatially by shaking, temporal dynamics could be at play due to the feeding regime and ability of Colpoda and Maryna to form resting cysts, thus opening the possibility a storage effects (Chesson 1994). On the other hand, mutualism and parasitism could be acting as changing components of a protists niche. If protists were evolving in their ability to form mutualistic relationships or their susceptibility to pathogenic bacteria, the

73 Colpoda with Maryna Colpoda with Tetrahymena

A B 1.10 1.15 1.00 1.05 Niche difference Niche difference Niche 0.90 Before 1 sp. 2 sp. 0.95 Before 1 sp. 2 sp. Selection environment Selection environment

Tetrahymena with Maryna Paramecium with Maryna

C 1.4 D 9 hrs 1.3 4 days 1.3 1.2 1.2 1.1 1.1 Niche difference Niche difference Niche 1.0 1.0 0.9 Before 1 sp. 2 sp. Before 1 sp. 2 sp. Selection environment Selection environment

Figure 5.6: Niche diﬀerence is estimated as the scaled dissimilarity between species associated bacterial assemb;ages before selection, after selection in monoculture and after selection in pairwise competition. The grey bar indicates no change in similarity of associated bacterial communities between two species after protists consumption has occurred for 9hours and/or 96hours, i.e. equivalent niches. (A) Colpoda with Maryna at 9 hrs. (B) T etrahymena with Colpoda at 9 hrs. (C) Maryna with T etrahymena at 96 hours. (D) Maryna with P aramecium at 96 hours.

74 relative proportion of these bacteria would change in the community and thus leave a signal in the bacterial assemblage. Thus our measure of niche difference is likely to detect such changes. Here we saw the niche not changing or converging, which is what we also saw in the fourth chapter of this dissertation, where estimates of the niche were derived from population models. When two species have the same niche, the one with the greater average fitness should outcompete the other. Thus, fitness differences between species can be thought of as differences in the competitive ability of two species (Adler et al. 2007). As competition is mediated through shared resources, the ability to take up resources and suppress the bacterial community should be a predictor of competitive ability and therefore species average fitness. It is surprising that species tended not to change in their ability to suppress the bacterial community given that in the parameter estimations changes in fitness differences tended to be much bigger than changes in niche overlap. However, bacterial consumption rate is only one parameter that determines competitive ability; assimilation efficiency and mortality were not directly measured, and thus could be contributing to changes in competitive ability with evolution. A next step, could be to estimate protist mortality and assimilation efficiency before and after evolution to determine if these parameters account for the change in fitness. Together, these proxies for niche and fitness suggest that evolution among competitors is not changing their ability to coexist with one another in three out of four species pairs. This work suggests that either competition is not playing a strong role in the top trophic level of this model ecosystem (but see Chapter 4) or there is a lack of genetic variation within the protist cultures, as evolution does not appear to be occurring differentially in monocultures or competition. Or perhaps species are perfectly substitutable or perfectly non-overlapping so most evolution occurs similarly between competition and monoculture, though both situations seem unlikely. Though changes in bacterial consumption due to evolution were rare, in one case, T etrahymena with Maryna changes in consumption rate were observed, that made species less equal in their ability to draw down resources. Changes in fitness will make species less equal and decrease the probability of coexistence. For this pair, evolution of the resource niche of these species appeared to convergence. Both of these processes working together, would force species to compete more, while one species decreases its ability to compete efficiently, subsequently reducing coexistence between species.

75 Experimental evidence for convergence in the niche of two species is elusive even though it is often suggested through theory (Abrams 1983, Hubbell 2005) and biogeographic patterns (Ackerly 2004, Elias et al. 2008). To the best of my knowledge, no other studies have observed convergence in the niche, therefore this study marks a novel contribution. This work suggests that although convergence is rare it is not impossible, however it is associated with increased fitness differences and therefore the increased probability of extinction between species, suggesting that it will be difficult to detect such phenomenon as they are fundamentality transient dynamics in ecological systems. Subsequently, this outcome, in addition to the theoretical model in Chapter reiterates the long held aciom that indeed species must have niche differences to coexist (Gause 1934). Both before and after selection, in all experiments, the presence of protists had a significant effect on the composition of the bacterial community 9 and/or 96 hours later (Fig. 4). However, this was not always true for the abundance of bacteria. Effects on the abundance of bacteria was likely to be seen at 96 hours, but not necessarily at 9 hours. This suggests density dependent dynamics within the bacterial community for which bacteria can compensate for consumption by protists through changes in community composition on the timescales of bacterial life, but in the longer term, protists increase in density results in suppression of the bacterial community. More work remains to be done in associating protist resource use with competitive interactions observed among protists in Chapter 4. Larger measures of niche overlap should be associated with increases competitive interactions for both species. Larger differences in consumption rate should be associated with competitive ability. Techniques in describing bacterial assemblages as a measure of the niche of protists are new, and it’s unclear to what degree the indirect effects in the bacterial community would obscure the signal of predation by protists, and subsequently the link between competitive interactions. The bacterial communities effect on the protists is probably much more complicated than we can currently grapple with experimentally. There could be species of bacteria that facilitate protists and thus expand the protists niche, or there could be species that act as parasites that could drive density or frequency dependence in protists and thus have the potential to drive evolved resistance to such pathogens. Additionally, indirect effects of this biotic resource could be innumerable as bacterial species are likely experiencing their own competitive dynamics (Abrams and Cortez 2015). Thus, it remains hard to generalize even in model systems that we can manipulate experimentally. More

76 work is needed on much shorter timescales to resolve the complex interactions between predators and competing prey. In general more work is needed that reveals the mechanisms by which species are experiencing competitive interactions in order to understand the complex nature in which species respond and adapt to their environments.

77 CHAPTER 6

CONCLUDING REMARKS

When I started this project, I was compelled by the paradox of diversity. As a theory oriented person, I clung tightly to the competitive exclusion principle as one of the extremely scarce ’laws’ of ecology; but at the same time, I marveled at the diversity, not just from the maintenance perspective, but also the origins. I meandered through community ecology concepts such as neutral theory of biodiversity as quirky exceptions to Gause’s fundamental principle, and inspired by theory on the evolution of convergence (terHorst et al 2010) as a possible dulling of the coexistence knife edge. I began to ask: can evolution reveal how diversity is maintained? Are the conditions for ecological convergence common or meaningful? Modern coexistence theory called to me like a beacon of hope for the reconciliation of these concepts. Upon conception, I thought the outcome of my dissertation would be to enumerate ways in which evolution could act to facilitate coexistence. However, the story that emerged when the dust of theoretical work and experiments had settled was: (1) evolution of fitness differences can be more important than evolution of niche overlap, and (2) evolution can often degrade the ability of species to coexist. The call to action here is that changes in fitness differences must be considered in conjunction with changes in niche overlap to understand the ecological and evolutionary dynamics of competitors. In the theory chapter, we found that environmental context is an important driver of evolution among competitors. When niches are constrained, species will tend to evolve more in their fitness differences which often leads to extinction, even when species ecologically appear to be coexisting. When niches are unoccupied, species will spread out, decreasing niche overlap, and have little changes in fitness differences. Interestingly, there appeared to be few mechanisms for convergence in fitness that were distinct from species diverging in niches simultaneously. Thus, I conclude that fitness differences are most likely to increase as a result of evolution in competition. The degree of conservation in niche overlap and fitness difference would determine the extent to which phylogeny is predictive of competitive outcomes. If niches are not labile, relative to the

78 scale of phylogeny, we would expect phylogeny to be predictive. Additionally, conservation in fitness between closely related taxa would result in strong phylogenetic signal in the strength of competitive interactions. In the phylogeny chapter, we find that phylogeny is not predictive of competitive outcomes in protists, possibly due to the deeply divergent nature of this clade. None the less, phylogeny is not going to be predictive of future evolution in this system. In the experiment chapter, we find it difficult to determine if species are evolving in either fitness differences or niche overlap. But it seems that fitness differences evolve more than niche overlap, suggesting constraints on niche evolution. The difficult question to answer is whether this experiment was successful. Were the response surface experiments sufficiently sampled to capture the population dynamics? Was there enough genetic diversity to select upon? Were five replicates sufficient for dealing with the variance in the system? Challenges in this experimental design mean there is more work to do on this question. In the final chapter, novel techniques in description of the bacterial community resulted in similar findings as the experimental chapter, and suffer from most of the same problems. Again, species don’t appear to be evolving much in their niche, but neither did we find large changes in the consumption rate of protists. Competition rarely happens amongst just two species in most communities. Multiple species within a guild interact, compete and evolve in response to these interactions. The next step would be to ask how these processes scale up to whole communities. For species that have been assembled from some external species pool, dynamics may begin to look like those of a constrained niche environment; there is not enough room for each species to have its own niche, so niche convergence may become inevitable. In these cases, fitness differences should increase with evolution in most cases, resulting in extinction of most species. But if communities are assembled through in situ speciation, niche divergence should occur until all niches are filled. Once this process is completed, any subsequent speciation would begin with species that are very similar, my models and experiments suggest that one species will gain an advantage, fitness differences will increase and one of the sister taxa will go extinct. Evolution tends to exacerbate instability in competition systems, which reinforces the principle that two species cannot coexist in the long-term on any given niche. Ultimately, how evolution effects coexistence depends on the relative rate of changes between niche overlap and fitness differences. I’ve made predictions that fitness differences will change more

79 in constrained environments, but more work is need to truly understand what the concept really entails since niches tend to be multidimensional rather than a single axis.

80 BIBLIOGRAPHY

Aarssen, L. W. 1983. Ecological combining ability and competitive combining ability in plants: Toward a general evolutionary theory of coexistence in systems of competition. The American Naturalist 122:707-731.

Abrams, P. A. 1983. The theory of limiting similarity. Annual Review of Ecology and Systematics 14:359-376.

Abrams, P. A. 1986. Character displacement and niche shift analyzed using consumer-resource models of competition. Theoretical Population Biology 29:107-160.

Abrams, P. A. 2000. The Evolution of Predator-Prey Interactions: Theory and Evidence. Annual Review of Ecology and Systematics 31:79-105.

Abrams, P. A., and M. H. Cortez. 2015. The many potential indirect interactions between predators that share competing prey. Ecological Monographs 85:625-641.

Ackerly, D. D. 2003. Community Assembly, Niche Conservatism, and Adaptive Evolution in Chang- ing Environments. International Journal of Plant Sciences 164:S165-S184.

Ackerly, D. D. 2004. Adaptation, Niche Conservatism, and Convergence: Comparative Studies of Leaf Evolution in the California Chaparral. The American Naturalist 163:654-671.

Adler, P. B., J. HilleRislambers, and J. M. Levine. 2007. A niche for neutrality. Ecology Letters 10:95-104.

Agnarsson, I., N. J. Gotelli, A. D, and M. Kuntner. 2016. Limited role of character displacement in the coexistence of congeneric Anelosimus spiders in a Madagascan montane forest. Ecography 39:743-753.

Amarasekare, P. 2003. Competitive coexistence in spatially structured environments: a synthesis. Ecology Letters 6:1109-1122.

Anacker, B. L., and S. Y. Strauss. 2016. Ecological similarity is related to phylogenetic distance between species in a cross-niche ﬁeld transplant experiment. Ecology 97:1807-1818.

Baselga, A. 2010. Partitioning the turnover and nestedness components of beta diversity. Global Ecology and Biogeography 19:134-143.

Baselga, A., and C. D. L. Orme. 2012. Betapart: An R package for the study of beta diversity. Methods in Ecology and Evolution 3:808-812.

81 Bastolla, U., M. A. Fortuna, A. Pascual-Garca, A. Ferrera, B. Luque, and J. Bascompte. 2009. The architecture of mutualistic networks minimizes competition and increases biodiversity. Nature 458:1018-1020.

Becks, L., S. P. Ellner, L. E. Jones, and N. G. Hairston. 2012. The functional genomics of an eco-evolutionary feedback loop: Linking gene expression, trait evolution, and community dynamics. Ecology Letters 15:492-501.

Blomberg, S. P., T. Garland, and A. R. Ives. 2003. Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution 57:717-745.

Bokulich, N. A., C. M. L. Joseph, G. Allen, A. K. Benson, and D. A. Mills. 2012. Next-generation sequencing reveals signiﬁcant bacterial diversity of botrytized wine. PLoS ONE 7:e36357.

Bono, L. M., C. L. Gensel, D. W. Pfennig, and C. L. Burch. 2013. Competition and the origins of novelty: experimental evolution of niche-width expansion in a virus. Biology letters 9:2012.0616.

Burns, J. H., and S. Y. Strauss. 2011. More closely related species are more ecologically similar in an experimental test. Proceedings of the National Academy of Sciences 108:5302-5307.

Cahill, J. F., S. W. Kembel, E. G. Lamb, and P. A. Keddy. 2008. Does phylogenetic relatedness inﬂuence the strength of competition among vascular plants? Perspectives in Plant Ecology, Evolution and Systematics 10:41-50.

Caporaso, J. G., K. Bittinger, F. D. Bushman, T. Z. DeSantis, G. L. Andersen, and R. Knight. 2010. PyNAST: a ﬂexible tool for aligning sequences to a template alignment. Bioinformatics (Oxford, England) 26:266-7.

Caporaso, J. G., C. L. Lauber, W. a Walters, D. Berg-Lyons, J. Huntley, N. Fierer, S. M. Owens, J. Betley, L. Fraser, M. Bauer, N. Gormley, J. a Gilbert, G. Smith, and R. Knight. 2012. Ultra- high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. The ISME Journal 6:1621-1624.

Cavender-Bares, J., K. H. Kozak, P. V. A. Fine, and S. W. Kembel. 2009. The merging of community ecology and phylogenetic biology. Ecology Letters 12:693-715.

Chantangsi, C., D. H. Lynn, M. T. Brandl, J. C. Cole, N. Hetrick, and P. Ikonomi. 2007. Barcoding ciliates: a comprehensive study of 75 isolates of the genus Tetrahymena. International journal of systematic and evolutionary microbiology 57:2412-25.

Chase, J. M., and M. Leibold. 2003. Ecological niches: linking classical and contemporary approaches. University of Chicage Press.

82 Chave, J., D. Coomes, S. Jansen, S. L. Lewis, N. G. Swenson, and A. E. Zanne. 2009. Towards a worldwide wood economics spectrum. Ecology Letters 12:351-366.

Chesson, P. 1994. Multispecies Competition in Variable Environments. Theoretical Population Biology 45:227-276.

Chesson, P. 2000. Mechanisms of Maintenance of Species Diversity. Annual Review of Ecology and Systematics 31:343-366.

Chesson, P. 2013. Species competition and predation. Pages 223-256Ecological Systems. Springer, New York.

Cole, J. R., Q. Wang, E. Cardenas, J. Fish, B. Chai, R. J. Farris, A. S. Kulam-Syed-Mohideen, D. M. McGarrell, T. Marsh, G. M. Garrity, and J. M. Tiedje. 2009. The Ribosomal Database Project: Improved alignments and new tools for rRNA analysis. Nucleic Acids Research 37.

Cortez, M. H., and J. S. Weitz. 2014. Coevolution can reverse predator-prey cycles. Proceedings of the National Academy of Sciences of the United States of America 111:7486-91.

Cushman, L. H., and J. F. Addicott. 1989. Intra- and interspeciﬁc competition for mutualists: ants as a limited and limiting resource for aphids. Oecologia 79:315-321.

Dieckmann, U., and M. O. Doebeli. 1999. On the origin of species by sympatric speciation. Nature 400:354-357.

Elias, M., Z. Gompert, C. Jiggins, and K. Willmott. 2008. Mutualistic Interactions Drive Ecological Niche Convergence in a Diverse Butterﬂy Community. PLoS Biology 6:e300.

Ellis, C. N., C. C. Traverse, L. Mayo-Smith, S. W. Buskirk, and V. S. Cooper. 2015. Character displacement and the evolution of niche complementarity in a model bioﬁlm community. Evolution 69:283-293.

Falster, D. S., . Brnnstrm, M. Westoby, and U. Dieckmann. 2017. Multitrait successional forest dynamics enable diverse competitive coexistence. Proceedings of the National Academy of Sciences 114:E2719-E2728.

Fiˇser,C., R. Luˇstrik, S. Sarbu, J. Flot, and P. Trontelj. 2015. Morphological evolution of coexisting amphipod species pairs from sulﬁdic caves suggests competitive interactions and character displacement, but no environmental ﬁltering and convergence. PloS one 10:e0123535.

First, M. R., N. Y. Park, M. E. Berrang, R. J. Meinersmann, J. M. Bernhard, R. J. Gast, and J. T. Hollibaugh. 2012. Ciliate ingestion and digestion: Flow cytometric measurements and regrowth of a digestion-resistant Campylobacter jejuni. Journal of Eukaryotic Microbiology 59:12-19.

83 Fox, J. W., and D. a Vasseur. 2008. Character convergence under competition for nutritionally essential resources. The American naturalist 172:667-80.

Gao, F., and L. a Katz. 2014. Phylogenomic analyses support the bifurcation of ciliates into two major clades that diﬀer in properties of nuclear division. Molecular phylogenetics and evolution 70:240-3.

Gause, F. 1934. The Struggle for Existence. The Yale journal of biology and medicine 7:609.

Del Giorgio, P. a., D. F. Bird, Y. T. Prairie, and D. Planas. 1996. Flow cytometric determination of bacterial abundance in lake plankton with the green nucleic acid stain SYTO 13. Limnology and Oceanography 41:783-789.

Godoy, O., N. J. B. Kraft, and J. M. Levine. 2014. Phylogenetic relatedness and the determinants of competitive outcomes. Ecology Letters 17:836-844.

Godoy, O., and J. M. Levine. 2014. Phenology eﬀects on invasion success: Insights from coupling ﬁeld experiments to coexistence theory. Ecology 95:726-736.

Goldberg, D. E. 1996. Competitive ability: deﬁnitions, contingency and correlated traits. Philo- sophical Transactions of the Royal Society B: Biological Sciences 351:1377-1385.

Gomulkiewicz, R., R. D. Holt, R. Gomulkiewicz1and, and R. D. Holt. 1995. When does evolution by natural selection prevent extinction? Evolution 49:201-207.

Hairston, N. G., S. P. Ellner, M. A. Geber, T. Yoshida, and J. A. Fox. 2005. Rapid evolution and the convergence of ecological and evolutionary time. Ecology Letters 8:1114-1127.

Harmon, L. J., and S. Harrison. 2015. Species diversity is dynamic and unbounded at local and continental scales. The American Naturalist 185:584-593.

Harvey, P. H., and M. D. Pagel. 1991. The comparative method in evolutionary biology. Page Oxford University Press, Oxford, England.

Hill, S. B., and P. M. Kotanen. 2009. Evidence that phylogenetically novel non-indigenous plants experience less herbivory. Oecologia 161:581-590.

Hubbell, S. P. 2001. The uniﬁed neutral theory of biodiversity and biogeography. Page Monographs in Population Biology. Princeton University Press, Princeton.

Hubbell, S. P. 2005. Neutral theory in community ecology and the hypothesis of functional equivalence.

Hutchinson, G. E. 1959. Homage to Santa Rosalia or why are there so many kinds of animals? the American Naturalist 93:145-159.

84 Hutchinson, G. E. 1965. The ecological theater and the evolutionary play. Yale University Press.

Inouye, B. D. 2001. Response surface experimental designs for investigating interspeciﬁc competition. Ecology 82:2696-2706.

Jiang, L., J. Tan, and Z. Pu. 2010. An Experimental Test of Darwin’s Naturalization Hypothesis. The American Naturalist 175:415-423.

Jrgens, K., and C. Matz. 2002. Predation as a shaping force for the phenotypic and genotypic composition of planktonic bacteria. Antonie van Leeuwenhoek, International Journal of General and Molecular Microbiology 81:413-434.

Katoh, K., and D. M. Standley. 2013. MAFFT multiple sequence alignment software version 7: Improvements in performance and usability. Molecular Biology and Evolution 30:772-780.

Kokko, H., and A. Lpez-Sepulcre. 2007. The ecogenetic link between demography and evolution: Can we bridge the gap between theory and data?

Kraft, N. J. B., P. B. Adler, O. Godoy, E. James, S. Fuller, and J. M. Levine. 2014. Community assembly, coexistence, and the environmental ﬁltering metaphor. Functional Ecology:1-8.

Kraft, N. J. B., W. K. Cornwell, C. O. Webb, and D. D. Ackerly. 2007. Trait evolution, community assembly, and the phylogenetic structure of ecological communities. American Naturalist 170:271-283.

Kraft, N. J. B., O. Godoy, and J. M. Levine. 2015. Plant functional traits and the multidimensional nature of species coexistence. Proceedings of the National Academy of Sciences 112:797-802.

Kunstler, G., S. Lavergne, B. Courbaud, W. Thuiller, G. Vieilledent, N. E. Zimmermann, J. Kattge, and D. a Coomes. 2012. Competitive interactions between forest trees are driven by species’ trait hierarchy, not phylogenetic or functional similarity: implications for forest community assembly. Ecology Letters 15:831-40.

Lande, R. 1976. Natural Selection and Random Genetic Drift in Phenotypic Evolution. Evolution 30:314.

Lankau, R. A. 2011. Rapid Evolutionary Change and the Coexistence of Species. Annual Review of Ecology, Evolution, and Systematics 42:335-354.

Lankau, R. A. 2012. Coevolution between invasive and native plants driven by chemical competition and soil biota 109.

Lankau, R. A., and S. Y. Strauss. 2011. Newly rare or newly common: evolutionary feedbacks through changes in population density and relative species abundance, and their management implications. Evolutionary Applications 4:338-353.

85 Law, R., and A. R. Watkinson. 1987. Response-surface analysis of two-species competition: an experiment on Phleum arenarium and Vulpia fasciculata. Journal of Ecology 75:871-886.

Leibold, M. A., and M. A. McPeek. 2006. Coexistence of the niche and neutral perspectives in community ecology.

Letten, A. D., P. J. Ke, and T. Fukami. 2017. Linking modern coexistence theory and contemporary niche theory. Ecological Monographs 87:161-177.

Levine, J. M., and J. HilleRisLambers. 2009. The importance of niches for the maintenance of species diversity. Nature 461:254-7.

Loreau, M. 2004. Does functional redundancy exist? Oikos 104:606-611.

Macarthur, R. H. 1958. Population ecology of some warblers of northeastern coniferous forests. Source: Ecology 39:599-619.

Macarthur, R., and R. Levins. 1967. The limiting similarity, convergence, and divergence of coexisting species. American Naturalist 101:377-385.

Maherali, H., and J. N. Klironomos. 2007. Inﬂuence of phylogeny on fungal community assembly and ecosystem functioning. Science 316:1746-8.

Mayﬁeld, M. M., and J. M. Levine. 2010. Opposing eﬀects of competitive exclusion on the phylogenetic structure of communities. Ecology Letters 13:1085-1093.

Medlin, L., H. J. Elwood, S. Stickel, and M. L. Sogin. 1988. The characterization of enzymatically ampliﬁed eukaryotic 16S-like rRNA-coding regions. Gene 71:491-499.

Miller, T. E. 1994. Direct and indirect species interactions in an early old-ﬁeld plant community. The American Naturalist 143:1007-1025.

Miller, T. E., E. S. Gornish, and H. L. Buckley. 2009. Climate and coastal dune vegetation: Disturbance, recovery, and succession. Plant Ecology 206:97-104.

Miller, T. E., E. R. Moran, and C. P. TerHorst. 2014. Rethinking niche evolution: experiments with natural communities of Protozoa in pitcher plants. The American Naturalist 184:277-283.

Mouquet, N., V. Devictor, C. N. Meynard, F. Munoz, L.-F. Bersier, J. Chave, P. Couteron, A. Dalecky, C. Fontaine, D. Gravel, O. J. Hardy, F. Jabot, S. Lavergne, M. Leibold, D. Mouillot, T. Mnkemller, S. Pavoine, A. Prinzing, A. S. L. Rodrigues, R. P. Rohr, E. Thbault, and W. Thuiller. 2012. Ecophylogenetics: advances and perspectives. Biological reviews of the Cambridge Philosophical Society 87:769-85.

86 Pacala, S. W., and J. Roughgarden. 1985. Population experiments with the Anolis lizards of St. Maarten and St. Eustatius.

Pagel, M. 1999. The maximum likelihood approach to reconstructing ancestral character states of discrete characters on phylogenies. Systematic biology 48:612-622.

Paisie, T. K., T. E. Miller, and O. U. Mason. 2014. Eﬀects of a ciliate protozoa predator on microbial communities in pitcher plant (Sarracenia purpurea) leaves. PLoS ONE 9.

Parfrey, L. W., D. J. G. Lahr, A. H. Knoll, and L. A. Katz. 2011. Estimating the timing of early eukaryotic diversiﬁcation with multigene molecular clocks. Proceedings of the National Academy of Sciences 108:13624-13629.

Pastore, A. I., and B. P. Scherer. 2016. Changes in community phylogenetic structure in a North American forest chronosequence. Ecosphere 7:e01592. 10.1002/ecs2.1592.

Paulson, J. N., O. C. Stine, H. C. Bravo, and M. Pop. 2013. Diﬀerential abundance analysis for microbial marker-gene surveys. Nature methods 10:1200-2.

Peay, K. G., M. Belisle, and T. Fukami. 2012. Phylogenetic relatedness predicts priority eﬀects in nectar yeast communities. Proceedings of the Royal Society B: Biological Sciences 279:749- 758.

Penn, O., E. Privman, H. Ashkenazy, G. Landan, D. Graur, and T. Pupko. 2010. GUIDANCE: A web server for assessing alignment conﬁdence scores. Nucleic Acids Research 38. Peterson, M. L., K. J. Rice, and J. P. Sexton. 2013. Niche partitioning between close relatives suggests trade-oﬀs between adaptation to local environments and competition. Ecology and Evolution 3:512-522.

Prinzing, A., W. Durka, S. Klotz, and R. Brandl. 2001. The niche of higher plants: evidence for phylogenetic conservatism. Proceedings of the Royal Society B: Biological Sciences 268:2383- 2389.

Revell, L. J. 2012. phytools: An R package for phylogenetic comparative biology (and other things). Methods in Ecology and Evolution 3:217-223.

Roughgarden, J. D. R. and J. 1985. A Theory of Faunal Buildup for Competition Communities. Evolution 39:26. Sanderson, M. J. 2003. r8s: Inferring absolute rates of molecular evolution and divergence times in the absence of a molecular clock. Bioinformatics 19:301-302.

Sargent, R. D., C. Goodwillie, S. Kalisz, and R. H. Ree. 2007. Phylogenetic evidence for a ﬂower size and number trade-oﬀ. American Journal of Botany 94:2059-2062.

Schreiber, S. J., R. B”urger, and D. I. Bolnick. 2011. The community eﬀects of phenotypic and genetic variation within a predator population. Ecology 92:1582-1593.

87 Sherr, E. B., and B. F. Sherr. 2002. Signiﬁcance of predation by protists in aquatic microbial food webs. Antonie van Leeuwenhoek, International Journal of General and Molecular Microbiol- ogy 81:293-308.

Snyder, R. E., and P. B. Adler. 2011. Coexistence and coevolution in ﬂuctuating environments: can the storage eﬀect evolve- The American Naturalist 178:E76-84.

Stephens, P. R., and J. J. Wiens. 2008. Testing for evolutionary trade-oﬀs in a phylogenetic context: ecological diversiﬁcation and evolution of locomotor performance in emydid turtles. Journal of Evolutionary Biology 21:77-87.

Strauss, S. Y., C. O. Webb, and N. Salamin. 2006. Exotic taxa less related to native species are more invasive. Proceedings of the National Academy of Sciences of the United States of America 103:5841-5.

Stuart, Y. E., T. S. Campbell, P. A. Hohenlohe, R. G. Reynolds, L. J. Revell, and J. B. Losos. 2014. Rapid evolution of a native species following invasion by a congener. Science 346:463-466.

Swenson, N. G. 2009. Phylogenetic Resolution and Quantifying the Phylogenetic Diversity and Dispersion of Communities 4. Swoﬀord, D. L. 2002. PAUP*: Phylogenetic Analysis Using Parsimony (*and Other Methods). Sinauer Associates, Sunderland, Massachusetts.:1-142. terHorst, C. P. 2011. Experimental evolution of protozoan traits in response to interspeciﬁc competition. Journal of evolutionary biology 24:36-46. terHorst, C. P., T. E. Miller, and E. Powell. 2010. When can competition for resources lead to ecological equivalence? Evolutionary Ecology Research 12:843-854. terHorst, C. P., P. C. Zee, K. D. Heath, T. E. Miller, A. I. Pastore, S. Patel, S. J. Schreiber, M. J. Wade, and M. R. Walsh. 2017. Evolution in a Community Context. The American Naturalist.

Tilman, D. 1980. Resources: A Graphical-Mechanistic Approach to Competition and Predation. The American Naturalist 116:362-393.

Tilman, D. 1990. Mechanisms of plant competition for nutrients: the elements of a predictive theory of competition. Perspectives on Plant Competition.:117-141.

Verd, M., P. J. Rey, J. M. Alcntara, G. Siles, and A. Valiente-Banuet. 2009. Phylogenetic signatures of facilitation and competition in successional communities. Journal of Ecology 97:1171-1180.

Violle, C., Z. Pu, and L. Jiang. 2010. Experimental demonstration of the importance of competition under disturbance. Proceedings of the National Academy of Sciences of the United States of America 107:12925-9.

88 Webb, C. O. 2000. Exploring the Phylogenetic Structure of Ecological Communities: An Example for Rain Forest Trees. The American Naturalist 156:145-155.

Webb, C. O., D. D. Ackerly, M. a. McPeek, and M. J. Donoghue. 2002. Phylogenies and Commu- nity Ecology. Annual Review of Ecology and Systematics 33:475-505.

Weber, M. G., C. E. Wagner, R. J. Best, L. J. Harmon, and B. Matthews. 2017. Evolution in a community context: on integrating ecological interactions and macroevolution. Trends in Ecology and Evolution 32:291-304.

Westoby, M., D. S. Falster, A. T. Moles, P. A. Vesk, and I. J. Wright. 2002. Plant ecological strategies: some leading dimensions of variation between species. Annual Review of Ecology and Systematics 33:125-159.

Yoshida, T., L. E. Jones, S. P. Ellner, G. F. Fussmann, and N. G. H. Jr. 2003. Rapid evolution drives ecological dynamics in a predator ? prey system 424:303-306.

Zhang, L., X. Mi, and H. Shao. 2016. Phylogenetic relatedness inﬂuences plant interspeciﬁc interactions across stress levels in coastal ecosystems: a meta-analysis. Estuaries and Coasts 39:1669-1678.

Zhao, L., Q. G. Zhang, and D. Y. Zhang. 2016. Evolution alters ecological mechanisms of coexistence in experimental microcosms. Functional Ecology 30:1440-1446.

Zuppinger-Dingley, D., B. Schmid, J. S. Petermann, V. Yadav, G. B. De Deyn, and D. F. B. Flynn. 2014. Selection for niche diﬀerentiation in plant communities increases biodiversity eﬀects. Nature 515:108-111.

89 BIOGRAPHICAL SKETCH

Abigail Ilona Pastore was born in Wichita, Kansas where the wonders of nature swept through empty prairies in milky starry nights, the complexity and chaotic nature of the world bloomed from prairie flowers and galaxies in kind. Abigail was embraced by the intellectual community at the University of Kansas, which encouraged her to walk an alternate path as well as granted her a Bachelor of Science in Mathematics advised by Dr. Saul Stahl. While watching dandelions set seed and disperse on the hillside of Mount Oread, Abigail was compelled to study the distribution and habit of plants. Then followed biology coursework and her first research project at Wichita State University, advised by Dr. Leland Russell, where she first toiled in the dirt of the prairies in order to manipulate and measure nature. Subsequently, Abigail found my way to Florida State under the mentorship of Dr. Tom Miller, where the work described here, and many other projects took place. At the time of publication of this work Abigail was at the University of Queensland, working as a postdoctoral fellow with Dr. Margie Mayfield.