The eco-evolutionary origins of life: Experimental evolution of biological innovation A DISSERTATION SUBMITTED TO THE FACULTY OF THE UNIVERSITY OF MINNESOTA BY Maria Kalambokidis IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY Michael Travisano August 2025 © Maria Kalambokidis 2025 i Acknowledgments This dissertation would not exist without the contributions of many people. First, I want to thank my advisor, Mike, whose support, perspective, and expertise made this work possible. Our weekly discussions were the most valuable part of my time in graduate school – they not only shaped all of the ideas presented in this dissertation but also provided encouragement which built my confidence as a scientist. Likewise, I am grateful to my committee members Kate, Will, and Mark for their support throughout each stage of this research – I am lucky to have such an inspiring team of mentors. Thank you also to my funding sources: the Future Investigators in NASA Earth and Space Science and Technology (FINESST) award, the University of Minnesota (EEB Excellence Fellowship, Research & Travel Awards), the Biotechnology Institute, and the Santa Fe Institute. In addition to supporting my research, this funding allowed me to attend several international conferences, each of which had a big influence on my ideas. Scientific research comes with a lot of uncertainty and disappointment; it really helps to have a supportive lab group to weather the storm, so thank you to the Travisano Lab for the years of feedback, troubleshooting, and camaraderie. In particular, I am so grateful to Bea and Nahui for their friendship and support throughout this experience – I admire and am inspired by both of you. To the EEB community, thank you for being a welcoming and exciting place to work. To my cohort – who I met first online, then in masks, and eventually out in the world – thanks for being such a cool, inspiring group of people. Thank you to my officemates Aarcha, Martha, and Katie for all the laughter and wisdom over the years – you made the office a fun place to be! And thank you to Kate and Neal for keeping the ship afloat by supporting and advocating for all EEB graduate students. ii Beyond EEB, I want to acknowledge the Biology Interest Group (BIG) community at the Minnesota Center for the Philosophy of Science (MCPS). Participating in our weekly seminars was such an important part of my education in graduate school, and I am especially grateful for the incredible academic examples I interacted with there, including Alan, Mark, and Ruth. Of course, nothing is possible without my friends and family. Having grown up in Minnesota, it was such a privilege to return here for my PhD and be near so many family members, including several aunts, uncles, cousins, and both of my grandmas. Thank you especially to my mom (the original Dr. Kalambokidis) for always believing in me – I can’t imagine going through this experience without you. And to my friends – from childhood, college, and beyond – thanks for keeping me going. To Alec, there aren’t quite words to describe it, but thank you for everything. You make the best parts of life even better, and you’re what gets me through the hardest parts. Finally, to all of my teachers and mentors who have supported me through each stage of my life. You encouraged every version of myself – from the seventh grader who couldn’t stop asking questions to the current version writing these words. Thanks to you, I got to explore the world. iii Dedication To my grandma, whom I saw every week during these past five years but who passed away just three weeks before my defense. Every starry sky, I think of you. iv Abstract The evolution of novelty is central to understanding the diversity and complexity of life. Pivotal moments of adaptation in Earth’s history led to leaps in biological complexity, such as during the origins of life, multicellularity, and eukaryotes, opening diverse avenues for further innovation. These transitions in individuality rely on the interplay of community-level interactions, yet most studies have not addressed this ecological context, in particular due to limits in our ability to examine these dynamics retrospectively. Here, I explore experimentally which selective contexts facilitate the emergence of innovation, introducing ecological complexity to three empirical systems: experimental evolution of multicellularity in two yeast species (Saccharomyces cerevisiae and Kluyveromyces lactis), in vitro selection of single-stranded DNA, and multidimensional niche breadth evolution of a halophilic archaeon (Halobacterium salinarum) and bacterium (Salinibacter ruber). I found that interspecific interactions shape the transition to multicellularity in each yeast species, which then shapes the nature of the species’ interactions. Similarly, the maintenance of diversity among single-stranded DNA provides opportunities for novel interactions among sequences, promoting an open-ended evolutionary response in a prebiotic context. By leveraging a comparative experimental approach, I also demonstrate that selection of halophiles in low and high-resource environments leads to species-specific responses observed across niche dimensions. Together, these results highlight how ecological and evolutionary constraints shape the emergence of innovation. I develop an “eco-evolutionary” approach to the origins of life, which is agnostic towards particular chemistries and instead explores the several ways an evolvable system may emerge and gain complexity. v Table of Contents Acknowledgements i Dedication ii Abstract iv List of Tables vii List of Figures viii Dissertation Outline x Introduction: The eco-evolutionary origins of life 1 CHAPTER 1: Multispecies interactions shape the transition to multicellularity 36 Abstract……………………………………………………………………………36 Introduction………………………………………………………………………..37 Materials and Methods…………………………………………………………….39 Results……………………………………………………………………………..43 Discussion…………………………………………………………………………52 Conclusion…………………………………………………………………………56 References…………………………………………………………………………57 Supplemental Materials……………………………………………………………63 CHAPTER 2: The maintenance and emergence of diversity promotes open-ended evolution in a pre-cellular system 72 Abstract……………………………………………………………………………72 Introduction………………………………………………………………………..73 Results……………………………………………………………………………..77 Discussion…………………………………………………………………………88 Conclusion…………………………………………………………………………95 Materials and Methods…………………………………………………………….96 References………………………………………………………………………..102 Supplemental Materials…………………………………………………………..112 vi CHAPTER 3: Experimental evolution of halophiles: rapid divergence along a multidimensional niche 122 Abstract…………………………………………………………………………..122 Introduction………………………………………………………………………123 Materials and Methods…………………………………………………………...128 Results……………………………………………………………………………134 Discussion………………………………………………………………………..143 Conclusion………………………………………………………………………..149 References………………………………………………………………………..150 Supplemental Materials…………………………………………………………..157 vii List of Tables Introduction: The eco-evolutionary origins of life Table 1. Summary of how three insights from ecology and evolution can guide investigations on the origins of life (OoL)………………………………………………...19 CHAPTER 1: Multispecies interactions shape the transition to multicellularity Table 1. Selection coefficients during multicellular sweep of K. lactis for each initial frequency…………………………………………………………………………………..48 CHAPTER 3: Experimental evolution of halophiles: rapid divergence along a multidimensional niche Table. 1. Abundance of salts and other resources in the rich and poor medias used during evolution experiment.…………………………………………………………………….129 Table 2. Media components for all possible medias………………………………...…...142 viii List of Figures Introduction: The eco-evolutionary origins of life Figure 1. Eco-evolutionary feedback……………………………………………………….9 Figure 2. Hypothetical scenario for how prebiotic EEFs can be conceptualized………….12 Figure 3. Evolvability as a function of energy potential differs for ancestral and derived metabolic systems. ………………………………………………………………………..16 Figure 4. Two conceptual views of the origins of life. …………………………………...22 CHAPTER 1: Multispecies interactions shape the transition to multicellularity Figure 1. Unicellular (a) and multicellular (b) colonies of K. lactis and S. cerevisiae on YPL Plates…………………………………………………………………………………40 Figure 2. Relative frequencies of unicellular and multicellular K. lactis and S. cerevisiae in monocultures during 19 days with and without settling selection………………………...44 Figure 3. Relative frequencies of K. lactis and S. cerevisiae in cocultures during 19 days with and without settling selection.……………………………………………………….45 Figure 4. Density of cells after 24 hours of growth on spent media………………………46 Figure 5. Frequency of K. lactis multicellular clusters during 19 days of selection………47 Figure 6. Differential growth rate of multicellular K. lactis and S. cerevisiae……………49 Figure 7. Invasion dynamics of S. cerevisiae and K. lactis………………………………..51 CHAPTER 2: The maintenance and emergence of diversity promotes open-ended evolution in a pre-cellular system Figure 1. Adaptation during static selection with beads and cells………………………...79 Figure 2. Adaptation during variable selection with beads and cells……………………...82 Figure 3. Characterization of derived populations………………………………………...85 Figure 4. Innovation emerges during selection with cells…………………………………87 Figure 5. Alternative adaptive pathways during in vitro selection………………………..95 CHAPTER 3: Experimental evolution of halophiles: rapid divergence along a multidimensional niche Figure. 1. Experimental design for selection and niche breadth assays………………….130 ix Figure 2. Fitness of S. ruber and H. salinarum grown in rich and poor media…………..135 Figure 3. Absolute fitness values for archaeal and bacterial populations across a gradient of yeast extract abundance…………………………………………………………………..138 Figure 4. Fitness of bacterial and archaeal populations grown in low and high resources with varying salinity……………………………………………………………………...140 x Dissertation Outline This dissertation explores the interplay of ecological and evolutionary dynamics during the emergence of biological innovation in experimental populations. The work is organized into three chapters: • Chapter 1 examines the role of interspecific interactions during the transition to multicellularity in co-cultures of the yeast species S. cerevisiae and K. lactis, where multicellularity evolves in response to selection for faster settling ability. I document an eco-evolutionary feedback, where cooperation and competition across several biological scales shapes the origin and persistence of multicellularity. • In Chapter 2, I explore which eco-evolutionary contexts promote the emergence of innovation in an empirical model for pre-cellular evolution: in vitro selection of single-stranded DNA, where individuals are selected for their ability to bind to target substrates (e.g. cells or beads) as well as replicate. I find that reduced selection stringency maintains sequence diversity, which provides opportunities for cooperative interactions and the generation of novel, complex phenotypes. This open-ended evolutionary response highlights the roles of diversity, cooperation, and environmental variability in facilitating innovative outcomes in a prebiotic context. • While the first two chapters explore the role of biotic interactions in shaping novel phenotypes, Chapter 3 interrogates the complexity of the abiotic environment. I develop a new experimental system of the halophilic archaeon H. salinarum and the bacterium S. ruber selected in both low and high resources, measuring their growth across a range of non-selected environments. By disentangling the relative effects of resources versus salinity on their niche breadth evolution, I find that tradeoffs and pleiotropic improvements rapidly emerge and are multidimensional in xi nature. The work highlights a comparative experimental approach, where ancestral niche breadth appears to underlie instances of divergence versus parallelism, providing a predictive lens for anticipating pleiotropy. 1 Introduction: The eco-evolutionary origins of life Abstract The origin of life remains one of the greatest enigmas in science. The immense leap in complexity between prebiotic soup and cellular life challenges historically “chemistry- forward” and “biology-backwards” approaches. Evolution must have bridged this gap in complexity, so understanding factors that influence evolutionary outcomes is critical for exploring life’s emergence. Here, we review insights from ecology and evolution and their application throughout abiogenesis. In particular, we discuss how ecological and evolutionary constraints shape the evolution of biological innovation. We propose an “eco- evolutionary” approach, which is agnostic towards particular chemistries or environments and instead explores the several ways that an evolvable system may emerge and gain complexity. 2 Main Text: “But if (and oh what a big if) we could conceive in some warm little pond…” wrote Charles Darwin in 1871. In this pond, he imagined, life could chemically form, “ready to undergo still more complex changes […]” (Darwin & Hooker, 1871). Today, the origin of life remains one of the biggest open questions in biology. How did life emerge and gain complexity? For over a century, evolutionary biologists have explored the history of life on Earth, using comparative methods to disentangle the origin of novel traits, innovation, and major transitions in complexity (Szathmáry & Smith, 1995; Weiss et al., 2016). The origin of life itself, however, has eluded this comparative approach, paving the way for chemists to characterize Darwin’s “warm little pond” and build a repertoire of prebiotic processes. As a result, most investigations have focused on life’s initial chemistry rather than its biology. These “chemistry-forward” approaches have provided many insights into the suite of organic compounds present on early Earth and the chemistries likely to facilitate life-like processes, such as energy acquisition (Sousa et al. 2013) and autocatalysis (Kauffman, 1986). However, the primary challenge of this work has been to explain the immense leap in molecular complexity between the first steps in life’s emergence and the Last Universal Common Ancestor (LUCA). How does chemistry transition to biology? To fill this gap, “biology-backward” approaches have used phylogenetic methods to explore simpler precursors to the processes and structures found in extant life (Becerra et al., 2014). However, there is little expectation that even these precursors could arise spontaneously without a prior adaptive process. Further work, therefore, has recognized the role of evolvability in passing through a series of “mesobiotic” entities toward LUCA (Baum, 2015; Shenhav et al., 2003), although it remains unlikely that “chemistry-forward” and “biology-backward” approaches will someday meet in the middle. In this paper, we consider an “eco-evolutionary” approach, where eco-evolutionary processes are considered 3 at the outset, and insights from biological disciplines guide our exploration of life’s origins. Insights from alternative programs Chemistry-forward approach While the transition from nonliving matter into life is yet unknown, progress has been made towards a solution. Seventy years ago, the Miller-Urey spark-discharge experiment synthesized amino acids under what was considered a model atmosphere for primitive Earth (Miller, 1953). This landmark experiment ushered in a new era of experimental studies on prebiotic chemistry, where researchers developed an inventory of organic compounds thought to be necessary for life (McCollom, 2013). The synthesis of life’s building blocks, however, does not elucidate a pathway toward even the simplest cellular processes. Hence, another “chemistry-forward” approach emerged through the study of macromolecular self-organization and autocatalysis, the process in which reaction products are also catalysts in the same or coupled reactions (Hordijk et al., 2012; Kauffman, 1993). In particular, autocatalytic sets demonstrate how self-sustaining, evolvable systems may emerge spontaneously in the environment – a critical step toward the evolution of greater complexity (Hordijk & Steel, 2014). Yet, a vast gulf remains between “chemistry-forward” approaches and derived biological processes. Biology-backward approach Modern biology can tell us a great deal about LUCA and its closest predecessors. By exploring precursors to cellular life, “biology-backward” approaches have constructed minimal cells (Hutchison et al., 2016), protocells (Adamala & Szostak, 2013), and inferred the evolution of cellular processes like protein catalysis and genetic transmission (Guseva et al., 2017). However, anticipating precise evolutionary trajectories before genetic transmission challenges the scope of phylogenetic methods (Becerra et al., 2014). To add 4 to this challenge, our ability to predict evolution even in extant life is severely limited by the roles of chance and history (Gould, 1990; Travisano et al., 1995). While studies of parallel and convergent evolution have shown that similar phenotypes can evolve in response to similar environmental challenges (Losos et al., 1998), they may take substantially different paths toward that outcome. In the Long-Term Evolution Experiment (LTEE), for example, 12 separate populations of E. coli were founded from a single clone, evolving in controlled conditions for over 65,000 generations (Lenski et al., 1991). While the derived populations resembled one another in several ways (e.g., each population evolving greater fitness, growth rate, and cell size), the populations also diverged considerably, with each population accumulating a unique set of mutations and achieving different degrees of fitness under other conditions (Blount et al., 2018). Parallel replay experiments such as this demonstrate the sensitivity of outcomes to chance events over evolutionary history, known as “historical contingencies.” If evolutionary outcomes are contingent upon idiosyncratic events – rendering them unpredictable even in highly simplified and controlled settings where initial conditions are precisely known – how can we hope to predict evolutionary trajectories that occurred 4 billion years in the past? How can we peer behind the phylogenetic curtain and determine the path life took prior to genetic transmission? What is more, many hypotheses on the origins of life are non- falsifiable, leading investigations down separate, inconclusive paths. Thus, we need an approach to investigating the origins of life that is not wedded to resolving each step pre- cellular life took on primordial Earth and which incorporates insights on processes we know were present. What’s missing? Consistent with each proposed scenario for the origins of life is the relevance of evolutionary processes as they played out in an ecological theater. Studies of eco- evolutionary dynamics have explored the generation and maintenance of biological diversity in a range of contemporary contexts, incorporating theory on environmental 5 heterogeneity, organism function, and niche construction (Fussmann et al., 2007). In particular, the reciprocal interaction between the environment and evolving populations – known as eco-evolutionary feedbacks (EEFs) – sits at the core of ecology and evolutionary biology today (Post & Palkovacs, 2009). At the origins of life, eco-evolutionary feedbacks were undoubtedly central in driving prebiotic complexity, yet they are rarely labeled as such in the literature. For example, vast amounts of work have explored how vesicles provide mechanisms for localization and microenvironments favorable for prebiotic reaction networks (Toparlak & Mansy, 2019) or how encapsulation may provide stability necessary for information transmission (Peng et al., 2022). Other work includes theory on environmental alteration without encapsulation (e.g., localization on a mineral surface) that drives subsequent selection (e.g., colonizing the surface) (Baum, 2018; Wächtershäuser, 1988). Despite the ubiquity of eco-evolutionary dynamics in a prebiotic world, ecologists and evolutionists have largely remained silent on the subject. Since eco-evolutionary theory may be adapted to a variety of prebiotic contexts, allowing for the exploration of multiple paths toward life, biological approaches can greatly inform our understanding of pre-cellular evolution. Indeed, with evolvability as a requisite to life, there is a clear role for ecological and evolutionary perspectives in exploring life’s emergence. The goal of this paper is to demonstrate how foundational insights from ecology and evolution can be applied throughout the process of abiogenesis – from prebiotic soup to cells. We will explore three insights in particular, using each to inform our understanding of the eco-evolutionary dynamics present at the origins of life, beginning with, 1) spatial and temporal dynamics of eco-evolutionary feedbacks, 2) ecophysiological constraints on evolvability, and 3) determining ancestral function from derived traits. We will survey literature that exemplify this eco-evo approach, concluding with how these perspectives can guide future work on the origins of life. 6 Box. 1 Defining life and evolution. Here we briefly discuss definitions of life and prebiotic evolution as used in this paper. Defining life has been subject to tremendous debate. For practical purposes, we employ a working definition lately used by NASA: “a self- sustaining system capable of Darwinian evolution” (Joyce et al., 1994). Our consideration of life, in the context of life’s origin, is chemically agnostic and in general pertains to a variety of systems that could exist prior to cellular life as we know it. Our use of the term “pre-cellular” refers to systems that existed before – or are simpler than – LUCA (e.g., bacterial or archaeal life). Our requirements for “Darwinian evolution” are as described by Lewontin (1970) (Lewontin, 1970): phenotypic variation, fitness differences, and heritability of fitness. This scheme of evolution does not assume a particular mechanism of inheritance, so we apply it generally throughout the emergence of life. “Darwinian evolution,” also known as adaptive evolution, results specifically from the process of selection, but we assume that pre-cellular evolutionary outcomes could also be subject to processes governed by chance and history (e.g. drift and the stochastic generation of variation). “Evolvability” here is considered the ability to undergo adaptive and neutral evolution. A few evolvable systems used in the literature: 7 Eco-evolutionary feedbacks I cannot consider the organism without its environment… (Mitchell, 1959) How early cells interacted with their environment has been a central concern for several decades, leading to a search for environments that mimic life’s features (e.g., proton gradients; Sojo et al., 2016). On Earth today, the converse to the above quote is also true: it’s difficult to consider environments without organisms. Eco-evolutionary feedbacks have become critical to our understanding of biological systems, playing an important role in processes like the emergence of adaptive radiations (Habets et al., 2006), species coexistence (Loeuille, 2010), and the maintenance of community stability (Patel et al., 2018). Key parameters governing EEFs today can therefore inform the eco-evolutionary context when life began. Eco-evolutionary feedbacks are local in time and space There is accumulating evidence that ecological and evolutionary processes can occur at compatible timescales. Temporal constraints on EEFs require that the evolutionary response to ecological change is congruent with the subsequent ecological response, and so forth. Despite these constraints, EEFs are well-documented throughout natural and experimental systems, over long and short timescales (Post & Palkovacs, 2009). Even in highly simple systems – a single species in a constant environment – feedbacks can occur between changing population densities and corresponding strengths of selection (i.e., density-dependent selection) (Clarke, 1972; Travis et al., 2013). In more complex systems, EEFs occur across several levels of biological organization, from populations to communities or even ecosystems. For example, recent empirical work with three-spined sticklebacks observed ecosystem modifications caused by population-level changes in response to parasites, which then influenced the next generation of sticklebacks, highlighting the overlapping timescales of host-parasite and host-ecosystem interactions (Brunner et al., 2017). Thus, while EEFs may occur across several biological scales, temporal constraints remain local (Fig. 1). 8 As EEFs occur locally, they are also spatially constrained. Even in a nonstructured single-species environment, where resources are uniformly available, EEFs can occur between population size and trait values. In a structured environment, where resources are localized or populations exist in patches, EEFs may occur among patches between regional population sizes and local trait values (Govaert et al., 2019). In doing so, EEFs can link eco-evolutionary processes at different spatial scales; for example, a selective increase in patch size can select against dispersal, which decreases the likelihood of re-colonization (Poethke et al., 2011). Spatial selection can also act as an ecological filter, where faster individuals who arrive first benefit from reduced competition (Fronhofer & Altermatt, 2015). In each instance, the ecological processes at a regional scale among several patches (e.g., dispersal) can influence within-patch dynamics (e.g., competition), and likewise for evolutionary processes, where gene flow can reduce the effects of drift in a local population. In structured environments, or systems with discrete patches, dispersal is therefore a central trait, acting as both an ecological mechanism impacting densities and an evolutionary process mediating gene flow. Without dispersal, the evolutionary response to ecological change is more likely to be influenced by stochasticity (Gomulkiewicz & Holt, 1995). At the origins of life, where pre-cellular life faced a heterogeneous abiotic environment, these mechanisms would have been crucial for system stability. As life began to construct its environment, EEFs would have further generated spatial variation in biotic interactions, making spatiotemporal constraints even more important for understanding evolutionary outcomes. 9 Fig. 1. Eco-evolutionary feedback. EEFs are cyclical interactions between ecology and evolution. (1) Ecological interactions affect evolutionary change in a population’s traits, which then, (2) modify the nature of ecological interactions. The origins of life as an eco-evolutionary feedback Since life must have emerged from an abiotic environment, it is reasonable to consider the geochemical conditions favorable for the transition from non-living to living matter. As soon as an evolvable system emerged, however, it would have interacted with its environment (and other individuals) in a way that impacted its likelihood of persistence. For example, the adsorption of self-replicating molecules to a mineral surface would have stabilized the local environment, making it more likely that further autocatalysis could occur there, similar to primary succession in ecology (Baum, 2018). It is therefore important to not only consider which prebiotic environments led rise to life, but which constructed environments enabled its persistence. Indeed, this idea appears in many hypotheses on the origins of life. The RNA World Theory, for instance, relies on the insight that RNA molecules can both undergo replication (without DNA) and catalyze their replication (without proteins), leading to a proposed system of self-replicating RNA molecules that preceded the flow of genetic information from DNA to RNA (Joyce, 2002). 10 This scenario requires the eventual takeover of DNA as the primary means of heredity and control – an event that would have occurred in an RNA-constructed world, such as the microenvironment of a protocell (Szostak et al., 2001). In several proposed scenarios, encapsulation is required for this genetic takeover, demonstrating the interplay of niche construction and subsequent evolution before cellular life. Another key consideration for EEFs at the origins of life is the role of dispersal as both an ecological and evolutionary mechanism, as discussed above. An excellent demonstration of these dynamics is a model proposed by David Baum and colleagues, explored using “chemical ecosystem selection” (Baum, 2015; Baum & Vetsigian, 2017; Vincent et al., 2019). In this system, an “ecosystem” of mutually-catalytic molecules colonize a mineral surface in an aqueous solution (e.g., the ocean). As they reproduce, they spread across the surface, competing with neighboring evolvers for space. Eventually, parts of the surface-bound networks could break away and be released into the water column. These propagules, like spores or seeds, can land on another surface and grow, enabling the colonization of more surfaces via dispersal. In this system, the ability to persist in the water column (i.e., migration) offers a competitive advantage over other surface evolvers and is selected for, offering a pathway toward pre-cells where life may reproduce without adherence to a surface. Another model demonstrating the eco-evolutionary dynamics of dispersal is the Hot Spring Hypothesis, which proposes a pathway for the synthesis of lipid-encapsulated polymers (Damer & Deamer, 2020). In this system, cycles of rehydration and dehydration are thought to occur in “chemically optimal” freshwater pools, where selection favors persistence throughout the cycles. Following evolution in these pools, life-like systems would disperse to other “more extreme” streams and marine systems, where they would evolve tolerance to this new suite of conditions. Similar to surface evolvers, selection in this system favors persistence through dilution and stability in a new environment. In other words, the system selects for colonization ability. As the ability to colonize increases, life 11 facilitates its evolution. While proposed to occur in different environments, the commonalities between the Hot Spring Hypothesis and Baum’s surface evolvers demonstrate how eco-evolutionary processes like dispersal can generally be applied to research on the origins of life. Large-scale EEFs have long been recognized for their role in the evolution of life, such as the Great Oxidation Event, where the rise of oxygenic photosynthesis transformed Earth’s eco-evolutionary landscape. At the origins of life, EEFs would have occurred locally. The local spatiotemporal constraints on feedbacks are therefore critical for understanding the evolutionary context of early life. Our understanding of abiogenesis should not be a linear progression of chemical systems in an unaltered environment, but instead the evolution of EEFs themselves (Fig. 2). Future work should allow for the “deformation” of fitness landscapes, exploring how biology-induced environmental changes could have consequences for the fitness and coexistence of descendants, a phenomena observed throughout life in variable environments today (Suvorov et al., 2023; Ogbunugafor et al., 2016; Lindsey et al., 2013). 12 Fig. 2. Hypothetical scenario for how EEFs can be conceptualized in a prebiotic system. As time passes, increased stability of life-like systems promotes the strength of the feedback. The origin of biological function can be understood through change in these cyclical interactions. (1) A diverse pool of self-replicating polymers exist in solution with a mineral surface. Polymers have varying abilities to bind to the surface. (2) Some polymers attach to the surface, which provides a concentration mechanism to help molecules react, grow, and propagate. (3) The localized reactants at the surface provides a selection pressure to bind to the surface. Better binders are more likely to persist and outcompete polymers in solution. (4) Bound polymers provide new structures, where other polymers can attach. More opportunities to bind to molecules results in the evolution of cooperation between different polymer ‘species.’ (5) Networks of self-replicating polymers populate a mineral surface. Catalyzing the production of neighboring molecules provides new 13 surfaces to bind to. (6) Some networks of polymers grow faster than others and spread across the surface. (7) Faster growers outcompete other networks and evolve competitive ability to grow and colonize surfaces. (8) Evolution of new networks provides opportunities for novel interactions with other (perhaps larger) compounds. Our description of EEFs at the origins of life (Fig. 2) emphasizes adaptation with the pervasive effects of drift during the evolution of complexity. Indeed, random and neutral processes like drift have been studied extensively (Kimura, 1983; Suvorov et al., 2023; Wright, 1931), and evolution cannot be understood without them (Lynch, 2007; Travisano et al., 1995). Natural selection alone is insufficient to explain the emergence of complexity, and in some cases, diversity and complexity increase via a series of non- adaptive steps (Muñoz-Gómez et al., 2021; Stoltzfus, 1999). Neither drift nor selection occur in isolation, complexity arising spontaneously as variation accumulates (Brandon & McShea, 2020) and as a result of adaptive benefits (Szathmáry & Smith, 1995). Some researchers have highlighted the contributions of random and neutral processes to the spread of complex traits, such as the origins of life, eukaryotes and multicellularity, that are not necessarily more adaptive than simpler alternatives (Lynch et al., 2022; Muñoz- Gómez et al., 2021). This body of work, which includes the role of bioenergetics and evolutionary cell biology (Lynch et al., 2022; Lynch & Trickovic, 2020; Yang et al., 2021) further clarifies the mechanisms by which neutral processes can generate complexity and their relevance for selection. This interplay underscores the need for an eco-evolutionary approach to the origins of life, where opportunities for stochastic effects are embraced alongside adaptive consequences. Ecophysiological constraints on evolvability Living cells have evolved the ability to take energy from their surroundings and transform 14 it into forms usable inside a cell. Ecophysiological constraints on evolvability are important for understanding early forms of energy acquisition. Modern metabolic systems follow a series of small leaps in energy For all life on Earth, metabolism is the summation of incremental changes. Through a series of stepwise chemical reactions, electrons flow from donors to acceptors to generate ATP and establish a charge on the cell membrane. Cells then couple this flow of electrons to essential physiological activities, using enzymes to catalyze each reaction step (i.e. lower activation energies). This makes a relatively direct association between enzymatic activities and reproductive success, particularly for microorganisms. Because survival, stability, and fitness all depend on patterns of metabolism, microorganisms have evolved remarkable metabolic capacities that capture significant amounts of free energy from a substrate and use it to reproduce. This multi-step process of lowering activation energies through a series of intermediates is clearly derived – an enigma recognized by researchers of the origins of life. How did the first life-like systems acquire energy before the coordination of enzymatic machinery? Early life-like systems lacked the metabolic capacity to make big steps small The energy potential of a prebiotic environment must have been harnessed without incremental steps. This means that the activation energies needed to tap into the chemical energy harbored in prebiotic “food” must have been sufficiently low that they were surpassed in one or a few steps. Additionally, since homeostasis is derived, the environmental fluctuations and extremes present at life’s outset must not have been so high as to disrupt an individual’s integrity. This is why life today can exist in environments that would not have been hospitable for the earliest biological systems. For example, some microorganisms can survive the harsh and highly dynamic conditions of an active hydrothermal vent chimney, due to the versatility in their energy and carbon exploitation – but this capacity evolved over time (Le Bris et al., 2019). Thus, while it is tempting to look for environments with high energy potentials to “jumpstart” life, early biological systems 15 lacked the metabolic machinery to make these big leaps small. Instead, ancestral systems would only have had the ability to co-opt small leaps in energy. The physiochemical gradients surrounding vents – or other more moderate environments - may have provided energy potentials that prebiotic metabolisms could overcome. From an eco-evolutionary perspective, the contexts favorable for early biological processes would also have provided enough stability for persistence yet the capacity to change and eventually disperse to other locations. The canonical “chicken-and-egg” paradox at the origins of life has led to a variety of proposed mechanisms for self-replication without coordinated enzymes. “Metabolism- first” scenarios generally begin with life as a collective self-reproducing metabolism that emerges in a space of possible organic reactions. From this foundation, the ability to accurately transmit information via replication must emerge, a constraint that is difficult to overcome in high-energy environments (Vasas et al., 2010). This is because “metabolism- first” models often rely on information storage via composition of molecular assemblies, where compositional information is passed on following fission of vesicles (Segré et al., 2000, Segré et al., 2001). The evolvability of these “ensemble replicators” is limited by the inaccuracy of replicating compositional information such that fitter genomes are not necessarily maintained by selection (Vasas et al., 2010). Given this challenge – and the derived nature of modern metabolisms – an “eco-evolutionary” approach prioritizes evolvability as a requisite to harnessing large energy potentials (Fig. 3). 16 Fig. 3. Evolvability as a function of energy potential differs for ancestral and derived metabolic systems. Early life-like systems could harness small energy potentials in order to self-propagate and pass on fitness differences. As the environment’s energy potential increases, the ability of ancestral metabolisms to harness energy decreases. Modern life, however, uses enzymes to lower activation energies in a series of incremental steps, enabling adaptive evolution at high energy potentials. Arrow indicates the difference in evolvability between early and modern life, at the point where energy potential is too high to be overcome without derived coordination of enzymatic machinery. Environments favorable for cellular life cannot be mapped onto ancestral systems. Evolution of genetic inheritance Modern genetic systems are incredibly good at expressing and transmitting biological information. How did template-based inheritance begin? Determining the origin of information transmission poses several challenges inherent in exploring ancestral states of complex traits. Adaptation cannot be inferred from current function 17 Biologists have long recognized the difficulty in disentangling evolutionary trajectories, both retroactively and in predicting future trends. This is largely due to the roles of chance and history in determining evolutionary outcomes, where contingencies obscure reasons why life evolved the way it did (Gould, 1990; Travisano et al., 1995). Beyond chance and history, there are also challenges in inferring the ancestral state of a derived trait when a trait’s current function did not serve as its causal basis (Gould & Lewontin, 1979). In these cases, a trait may have originally evolved due to selection for a different function than is observed today (Gould & Vrba, 1982). For example, feathers have typically been seen as adaptations for flight, as their flight-enhancing features appear to be the product of recent evolutionary history. The origin of feathers, however, may have occurred due to selection for a different function, such as thermal regulation, as there are early fossils of flightless – yet feathered – animals (Benton et al., 2019). At some point in its evolutionary history, selection may have switched such that feathers evolved into structures that could support flight. Hence, the function of a trait at its emergence may be different from its function today. Evolution of pre-genetic inheritance In the case of information transmission, double-stranded DNA and single-stranded RNA have a remarkable capacity for translating information and passing that information to offspring. Since inheritance is essential for evolution, an incredible amount of work has explored how these polynucleotide chains emerged (Bhowmik & Krishnamurthy, 2019; Crick, 1968; Dounce, 1981; Robertson & Joyce, 2012). Much of this work has assumed that ancestral genes – or simpler precursors– were still comprised of linear strands of nucleotides. This assumption stems from our understanding of contemporary genetic inheritance, for which linear chains serving as templates are essential for transmitting information. In other words, researchers assume similar ancestral and derived functions for nucleic acids. These hypotheses imagine linear strands of DNA or (more commonly) RNA 18 performing template-based replication in a prebiotic context, often as the first evolvable unit (Lincoln & Joyce, 2009). However, it is difficult to conceive how something as complex as a strand of RNA – with sufficient stability and replication fidelity – could appear without a prior adaptive process. Earlier adaptation must have occurred without digital inheritance. To fill this gap, several hypotheses have proposed a means for non- genetic inheritance, such as analog inheritance, where varying chemical compositions are passed from generation to generation (Segré et al., 2000). If this is the case – and the earliest life-forms evolved prior to genetic transmission – then how and when did nucleic acids emerge? Looking to the origin of feathers for inspiration, nucleotides may have performed a different function before they were co-opted for replication. As a consequence, it is difficult to explore this prior function based on contemporary genetic structures. Just as is the case with other co-opted adaptations, modern genetics should not be a guide for what heritability looked like at the origins of life. A phylogenetic approach that extrapolates from modern biology is very limited in its ability to infer past functions. Detailing the steps in the origins of life is not a case of piecing together a string of increasingly complex structures. Instead, evolutionary processes create paths that twist, turn, and pass through a series of non-intuitive entities. Nature itself has been described as a “crafty backwoods mechanic,” reconditioning and redesigning old machines, or fashioning new ones with whatever is at hand (Wimsatt & Wimsatt, 2007). This process of re-engineering has profound implications for the products of evolutionary change. Rather than examine the path toward a certain structure, exploring the many ways of enabling biological function will help us better understand how life can emerge. 19 Eco-evolutionary insight Significance for OoL Approach to OoL studies with this perspective Examples of work using this approach in non-OoL systems Eco-evolutionary feedbacks are local in time and space Prebiotic evolution cannot be understood without considering the spatiotemporal constraints on reciprocal interactions between organisms and their environment Design experiments that allow evolving populations to reshape (or “deform”) fitness landscapes • Andrade- Domínguez et al., 2014 • Fronhofer & Altermatt, 2015 • Bajicć et al., 2018 Ecophysiological factors constrain evolvability Pre-cellular life had to evolve the metabolic capacity to adapt to environments with large energy potentials Explore ecological contexts that promote evolvability in the emergence of life • Goldsby et al., 2012 • Crombach & Hogeweg, 2008 • Review: Payne & Wagner, 2019 Adaptation cannot be inferred from current function The structures found in cellular life (e.g. ribosomes, nucleic acids) are limited guides for determining their origin Study the interplay of evolutionary processes (e.g. selection, drift, migration) in a variety of evolvable systems • Rebolleda- Gómez & Travisano, 2019 • Meyer et al., 2012 • Losos et al., 1998 Table 1. Summary of how three insights from ecology and evolution can guide investigations on the origins of life (OoL). Given evolvability was a requisite for 20 transitioning from prebiotic soup to LUCA, current understanding of the ecology and evolution of living systems should guide research on the emergence of life. This “eco- evolutionary” approach explores how eco-evolutionary dynamics in extant life inform evolution before cells and is not restricted to particular chemistries, geochemical environments, or molecular entities that may have influenced the emergence of life on early Earth. Since most hypotheses on the origins of life are non-falsifiable (e.g., RNA World, Hot Spring Hypothesis, etc.), the overarching goal of this approach is to explore the several ways that an evolvable system may emerge and gain complexity. An “eco-evolutionary” approach to the origins of life Darwin’s “tangled bank” appreciates life's remarkable complexity and co-dependency that is nevertheless guided by the same natural laws (Darwin, 1859). Rather than disentangling or recreating the bank, efforts to understand the origin of life should rely on understanding these natural processes. An “eco-evolutionary” approach asks: How do eco-evolutionary processes occur throughout the emergence of life? Do expectations shift for evolvable systems before cells? Goals of an “eco-evolutionary” approach The overarching goal of this approach is to explore the several ways that an evolvable system may emerge and gain complexity. This perspective remains agnostic towards particular chemistries or geochemical environments that may have influenced the emergence of life on early Earth and instead aims to understand eco-evolutionary contexts that facilitate multiple paths toward life. Drawing inspiration from the LTEE, where general trends were observable but pathways towards these outcomes were unpredictable (Blount et al., 2018, Travisano & Lenski, 1996), we conclude that retracing the precise trajectory life took on early Earth is unknowable, but understanding relevant processes – and their influence on the emergence of biological innovation – is possible. 21 This perspective is compatible with a “piecewise approximations to reality” approach to exploring complex systems (Wimsatt & Wimsatt, 2007). This view embraces the reality that even the best theories make idealizations or assumptions that fail as correct descriptions of the world. In the study of the origins of life, this is particularly poignant, as there are no “natural” systems or observational studies. Rather, every system is a model aimed at understanding historical events but ultimately failing to recapitulate the events themselves. Even the most “prebiotically plausible” or “realistic” systems will fall short at recreating primordial environments and evolutionary events, due to the limitations both inherent in the geologic record and of researchers trying to predict evolution’s capacity to tinker and re-engineer. Instead, researchers could employ what is described as “local realism,” where scientists argue on certain terms that a phenomenon is real and that their approach requires them to presuppose the existence of that phenomenon. For example, studies on the origins of life argue that there must have been adaptation before cells; therefore, we can presuppose the existence of an evolvable system. In this way, we hope to focus on biological processes unfolding throughout life’s emergence, rather than connecting a stepwise series of chemical entities to cellular life as we know it (Fig. 4). 22 Fig. 4. Two conceptual views of the origins of life. The “traditional view” (dark line) depicts a series of stepwise increases in life-like properties. Each step represents a discrete transition in complexity that follows a particular hypothesis for how life emerged on early Earth, in this case the RNA World Theory. Researchers with this view of life aim to determine how each step could have occurred under “prebiotically plausible” conditions, with the overarching goal to string together each step from prebiotic soup to simple cellular life. Importantly, this perspective follows one lineage, where each step proposes particular chemical entities that fit the same narrative. Given the role of stochasticity and contingency in evolution, piecing together each step in the emergence of life is not only impossible but also an unrealistic depiction of evolution as progress. An “eco- evolutionary” view (light lines) envisages that there are many paths towards life and explores several ways life-like systems may gradually evolve. In this view, eco- evolutionary dynamics like convergence, coexistence, and competitive exclusion play important roles throughout life’s trajectory. While “life-likeness” must have generally increased, this progression was not uniform, with some populations reverting to ancestral states or remaining stable. Tips of branches indicate extinction events, and branching nodes represent instances of divergence, where EEFs facilitate biological innovation and 23 the evolution of greater complexity in some lineages (see Fig. 2). Researchers with this view do not aim to determine each step in the origin of life but instead explore how the process of evolution can play out in a variety of systems (figure adapted from Shenhav et al. 2003) (Shenhav et al., 2003). Evolution of complexity The origin of life requires the emergence of biological innovation and complexity. From prebiotic soup to LUCA, the production of novel organisms may not have been continuous, but complexity must have generally increased. This challenge poses several relevant questions for evolutionary biologists: How do populations avoid evolutionary plateaus or stable states? How can innovation facilitate diversification? What factors promote evolvability? Answers to these questions rely on concepts like competition, cooperation, and niche construction to help generate systems capable of open-ended evolution (Channon, 2006; Leon et al., 2018; Turner et al., 2015). For example, competition within a species can promote novel phenotypes that alter fitness landscapes to provide qualitatively different peaks (Rainey & Travisano, 1998), and cooperation within groups can harbor diversity necessary for adaptation to new environments (Preisner et al., 2016). Evolvability can even emerge as a byproduct of selection for other traits, such as modularity in the beaks of Darwin’s finches. In this case, a two-module developmental program independently regulating beak depth and width induced multi-dimensional shifts in beak morphology, enabling radiations to new niches (Mallarino et al., 2011). These and similar studies show the connections to longstanding evolutionary theory on quantitative genetic control of traits (Andersson & Shaw, 1994; Hansen & Pélabon, 2021) and the complex causal connections between plasticity and evolvability (Crozier et al., 2008; Draghi & Whitlock, 2012; Murren et al., 2015). Devising experiments that allow open-endedness has been a central concern for artificial life researchers and can be translated to research on the origins of life (Dittrich et al., 2000; Taylor et al., 2016). For instance, the digital evolution 24 system, AVIDA, has enabled the evolution of novel, complex features (Lenski et al., 2003) and ecological interactions with greater levels of complexity (Zaman et al., 2014). An “eco-evolutionary” approach to the origins of life capitalizes on our understanding of these processes to design experiments with multiple solutions to a given selective pressure. These studies should not rely on the imagination of the researcher to predetermine the desired outcome but instead provide ecological contexts that allow several pathways for innovation to emerge. We’ve emphasized the role of selection here against a backdrop of ever-present random processes, which are essential for opening avenues for further innovation. The ability for drift and selection to act together during adaptive evolution has long been recognized in evolutionary theory (Ishida, 2017; Provine, 1986; Svensson, 2023), albeit with contention over their relative contributions (Dobzhansky & Pavlovsky, 1957; Fisher, 1930; Jensen et al., 2019; Kern & Hahn, 2018; Lynch et al., 2016). There is strong experimental evidence for how drift and selection can each act as primary drivers of evolutionary outcomes (Hochberg et al., 2020; Marques et al., 2018; Wong et al., 2012), yet entanglement of the two specifically can have profound consequences for the emergence of innovation. For instance, allowing populations to randomly explore genotype space via drift may provide opportunities to cross fitness valleys. Similarly, drift can work to maintain within-population variation upon which selection may then act. Experimental evidence for this entanglement was found during the LTEE when one population evolved the ability to use citrate, an abundant but previously unusable energy source (Blount et al., 2008). This radical innovation was made possible through long periods of drift before the population could climb a new adaptive peak (Blount et al., 2008; Burnham & Travisano, 2021). For these reasons, experiments that allow several possible outcomes are more likely to provide insights into the emergence of complexity. 25 Conclusion Research on the origins of life has traditionally been conducted by chemists. We recast these questions so that biologists study them as well. The dichotomy between “chemistry- forward” and “biology-backward” approaches neglects the role of eco-evolutionary dynamics throughout the process of abiogenesis, thus our goal is to use insights from biological disciplines to guide future work on the origins of life. Hypotheses claiming particular environments or the stepwise emergence of cellular structures are non-falsifiable and should not limit the scope of investigations. Instead, our inability to recreate primordial environments, as well as anticipate stochastic evolutionary trajectories, means that experiments should explore the processes we know were present when life began: evolution as influenced by adaptation, chance, and history. An “eco-evolutionary” approach harnesses these processes in a variety of systems to explore the many ways innovation can emerge, in a warm little pond or otherwise. References Adamala, K., & Szostak, J. W. (2013). Competition between model protocells driven by an encapsulated catalyst. Nature chemistry, 5(6), 495-501. Adami, C., & Brown, C. T. (1994). Evolutionary learning in the 2D artificial life system" Avida". arXiv preprint adap-org/9405003. Andersson, S., & Shaw, R. G. (1994). Phenotypic plasticity in Crepis tectorum (Asteraceae): genetic correlations across light regimens. Heredity, 72(2), 113-125. Andrade-Domínguez, A., Salazar, E., del Carmen Vargas-Lagunas, M., Kolter, R., & Encarnacion, S. (2014). Eco-evolutionary feedbacks drive species interactions. The ISME journal, 8(5), 1041-1054. Bajić, D., Vila, J. C., Blount, Z. D., & Sánchez, A. (2018). On the deformability of an empirical fitness landscape by microbial evolution. Proceedings of the National Academy of Sciences, 115(44), 11286-11291. 26 Baum, D. A. (2015). Selection and the origin of cells. Bioscience, 65(7), 678-684. Baum, D. A. (2018). The origin and early evolution of life in chemical composition space. Journal of theoretical biology, 456, 295-304. Baum, D. A., & Vetsigian, K. (2017). An experimental framework for generating evolvable chemical systems in the laboratory. Origins of Life and Evolution of Biospheres, 47(4), 481-497. Becerra, A., Rivas, M., García-Ferris, C., Lazcano, A., & Peretó, J. (2014). A phylogenetic approach to the early evolution of autotrophy: the case of the reverse TCA and the reductive acetyl-CoA pathways. Int Microbiol, 17(2), 91-97. Benton, M. J., Dhouailly, D., Jiang, B., & McNamara, M. (2019). The early origin of feathers. Trends in ecology & evolution, 34(9), 856-869. Bhowmik, S., & Krishnamurthy, R. (2019). The role of sugar-backbone heterogeneity and chimeras in the simultaneous emergence of RNA and DNA. Nature chemistry, 11(11), 1009-1018. Blount, Z. D., Borland, C. Z., & Lenski, R. E. (2008). Historical contingency and the evolution of a key innovation in an experimental population of Escherichia coli. Proceedings of the National Academy of Sciences, 105(23), 7899-7906. Blount, Z. D., Lenski, R. E., & Losos, J. B. (2018). Contingency and determinism in evolution: Replaying life’s tape. Science, 362(6415), eaam5979. Brandon, R. N., & McShea, D. W. (2020). The missing two-thirds of evolutionary theory. Cambridge University Press. Brunner, F. S., Anaya-Rojas, J. M., Matthews, B., & Eizaguirre, C. (2017). Experimental evidence that parasites drive eco-evolutionary feedbacks. Proceedings of the National Academy of Sciences, 114(14), 3678-3683. Burnham, T. C., & Travisano, M. (2021). The landscape of innovation in bacteria, battleships, and beyond. Proceedings of the National Academy of Sciences, 118(26), e2015565118. 27 Channon, A. (2006). Unbounded evolutionary dynamics in a system of agents that actively process and transform their environment. Genetic Programming and Evolvable Machines, 7(3), 253-281. Clarke, B. (1972). Density-dependent selection. The American Naturalist, 106(947), 1-13. Crick, F. H. (1968). The origin of the genetic code. Journal of molecular biology, 38(3), 367-379. Crombach, A., & Hogeweg, P. (2008). Evolution of evolvability in gene regulatory networks. PLoS computational biology, 4(7), e1000112. Crozier, L. G., Hendry, A. P., Lawson, P. W., Quinn, T. P., Mantua, N. J., Battin, J., ... & Huey, R. (2008). Potential responses to climate change in organisms with complex life histories: evolution and plasticity in Pacific salmon. Evolutionary Applications, 1(2), 252-270. Damer, B., & Deamer, D. (2020). The hot spring hypothesis for an origin of life. Astrobiology, 20(4), 429-452. Darwin, C. (1859). The Origin of Species by Means of Natural Selection, Or, The Preservation of Favoured Races in the Struggle for Life. Books, Incorporated, Pub. Darwin, C., & Hooker, J. D. (1871). Darwin Correspondence Project, “Letter No. 7471.” Dittrich, P., Liljeros, F., Soulier, A., & Banzhaf, W. (2000). Spontaneous group formation in the seceder model. Physical Review Letters, 84(14), 3205. Dobzhansky, T., & Pavlovsky, O. (1957). An experimental study of interaction between genetic drift and natural selection. Evolution, 311-319. Dounce, A. L. (1981). Origin of life-proposed mechanisms for primeval polynucleotide and peptide chain synthesis. Journal of Theoretical Biology, 90(1), 63-79. Draghi, J. A., & Whitlock, M. C. (2012). Phenotypic plasticity facilitates mutational variance, genetic variance, and evolvability along the major axis of environmental variation. Evolution, 66(9), 2891-2902. Eigen, M., & Schuster, P. (1977). A principle of natural self-organization. Naturwissenschaften, 64(11), 541-565. 28 Fisher, R. A. (1930). The genetical theory of natural selection. Clarendon Press, Oxford. Fronhofer, E. A., & Altermatt, F. (2015). Eco-evolutionary feedbacks during experimental range expansions. Nature communications, 6(1), 1-9. Fussmann, G. F., Loreau, M., & Abrams, P. A. (2007). Eco-evolutionary dynamics of communities and ecosystems. Functional ecology, 465-477. Goldsby, H. J., Dornhaus, A., Kerr, B., & Ofria, C. (2012). Task-switching costs promote the evolution of division of labor and shifts in individuality. Proceedings of the National Academy of Sciences, 109(34), 13686-13691. Gomulkiewicz, R., & Holt, R. D. (1995). When does evolution by natural selection prevent extinction?. Evolution, 201-207. Gould, S. J. (1990). Wonderful life: the Burgess Shale and the nature of history. WW Norton & Company. Gould, S. J., & Vrba, E. S. (1982). Exaptation—a missing term in the science of form. Paleobiology, 8(1), 4-15. Gould, S. J., Lewontin, R.C. (1979). The spandrels of San Marco and the Panglossian paradigm. Proc. R. Soc, Lond. B, 205, 581-598. Govaert, L., Fronhofer, E. A., Lion, S., Eizaguirre, C., Bonte, D., Egas, M., ... & Matthews, B. (2019). Eco‐evolutionary feedbacks—Theoretical models and perspectives. Functional Ecology, 33(1), 13-30. Guseva, E., Zuckermann, R. N., & Dill, K. A. (2017). Foldamer hypothesis for the growth and sequence differentiation of prebiotic polymers. Proceedings of the National Academy of Sciences, 114(36), E7460-E7468. Habets, M. G., Rozen, D. E., Hoekstra, R. F., & de Visser, J. A. G. (2006). The effect of population structure on the adaptive radiation of microbial populations evolving in spatially structured environments. Ecology letters, 9(9), 1041-1048. Hansen, T. F., & Pélabon, C. (2021). Evolvability: a quantitative-genetics perspective. Annual Review of Ecology, Evolution, and Systematics, 52, 153-175. 29 Hochberg, G. K., Liu, Y., Marklund, E. G., Metzger, B. P., Laganowsky, A., & Thornton, J. W. (2020). A hydrophobic ratchet entrenches molecular complexes. Nature, 588(7838), 503-508. Hordijk, W., & Steel, M. (2014). Conditions for evolvability of autocatalytic sets: A formal example and analysis. Origins of Life and Evolution of Biospheres, 44(2), 111-124. Hordijk, W., Steel, M., & Kauffman, S. (2012). The structure of autocatalytic sets: Evolvability, enablement, and emergence. Acta biotheoretica, 60(4), 379-392. Hutchison III, C. A., Chuang, R. Y., Noskov, V. N., Assad-Garcia, N., Deerinck, T. J., Ellisman, M. H., ... & Venter, J. C. (2016). Design and synthesis of a minimal bacterial genome. Science, 351(6280), aad6253. Ishida, Y. (2017). Sewall Wright, shifting balance theory, and the hardening of the modern synthesis. Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences, 61, 1-10. Jensen, J. D., Payseur, B. A., Stephan, W., Aquadro, C. F., Lynch, M., Charlesworth, D., & Charlesworth, B. (2019). The importance of the neutral theory in 1968 and 50 years on: a response to Kern and Hahn 2018. Evolution, 73(1), 111-114. Joyce G.F. Deamer D.W. Fleischaker G. Origins of Life: The Central Concepts. Jones and Bartlett; Boston: 1994. Foreword. Joyce, G. F. (2002). The antiquity of RNA-based evolution. nature, 418(6894), 214-221. Kauffman, S. A. (1986). Autocatalytic sets of proteins. Journal of theoretical biology, 119(1), 1-24. Kauffman, S. A. (1993). The origins of order: Self-organization and selection in evolution. Oxford University Press, USA. Kern, A. D., & Hahn, M. W. (2018). The neutral theory in light of natural selection. Molecular biology and evolution, 35(6), 1366-1371. Kimura, M. (1983). The neutral theory of molecular evolution. Cambridge University Press. 30 Le Bris, N., Yücel, M., Das, A., Sievert, S. M., LokaBharathi, P., & Girguis, P. R. (2019). Hydrothermal energy transfer and organic carbon production at the deep seafloor. Frontiers in Marine Science, 5, 531. Lenski, R. E., Ofria, C., Pennock, R. T., & Adami, C. (2003). The evolutionary origin of complex features. Nature, 423(6936), 139-144. Lenski, R. E., Rose, M. R., Simpson, S. C., & Tadler, S. C. (1991). Long-term experimental evolution in Escherichia coli. I. Adaptation and divergence during 2,000 generations. The American Naturalist, 138(6), 1315-1341. Leon, D., D'Alton, S., Quandt, E. M., & Barrick, J. E. (2018). Innovation in an E. coli evolution experiment is contingent on maintaining adaptive potential until competition subsides. PLoS genetics, 14(4), e1007348. Lewontin, R. C. (1970). The units of selection. Annual review of ecology and systematics, 1-18. Lincoln, T. A., & Joyce, G. F. (2009). Self-sustained replication of an RNA enzyme. Science, 323(5918), 1229-1232. Lindsey, H. A., Gallie, J., Taylor, S., & Kerr, B. (2013). Evolutionary rescue from extinction is contingent on a lower rate of environmental change. Nature, 494(7438), 463-467. Loeuille, N. (2010). Influence of evolution on the stability of ecological communities. Ecology letters, 13(12), 1536-1545. Losos, J. B., Jackman, T. R., Larson, A., Queiroz, K. D., & Rodrı́guez-Schettino, L. (1998). Contingency and determinism in replicated adaptive radiations of island lizards. Science, 279(5359), 2115-2118. Losos, J. B., Jackman, T. R., Larson, A., Queiroz, K. D., & Rodrı́guez-Schettino, L. (1998). Contingency and determinism in replicated adaptive radiations of island lizards. Science, 279(5359), 2115-2118. 31 Lynch, M. (2007). The frailty of adaptive hypotheses for the origins of organismal complexity. Proceedings of the National Academy of Sciences, 104(suppl_1), 8597- 8604. Lynch, M., & Trickovic, B. (2020). A theoretical framework for evolutionary cell biology. Journal of molecular biology, 432(7), 1861-1879. Lynch, M., Ackerman, M. S., Gout, J. F., Long, H., Sung, W., Thomas, W. K., & Foster, P. L. (2016). Genetic drift, selection and the evolution of the mutation rate. Nature Reviews Genetics, 17(11), 704-714. Lynch, M., Trickovic, B., & Kempes, C. P. (2022). Evolutionary scaling of maximum growth rate with organism size. Scientific reports, 12(1), 22586. Mallarino, R., Grant, P. R., Grant, B. R., Herrel, A., Kuo, W. P., & Abzhanov, A. (2011). Two developmental modules establish 3D beak-shape variation in Darwin's finches. Proceedings of the National Academy of Sciences, 108(10), 4057-4062. Marques, D. A., Jones, F. C., Di Palma, F., Kingsley, D. M., & Reimchen, T. E. (2018). Experimental evidence for rapid genomic adaptation to a new niche in an adaptive radiation. Nature ecology & evolution, 2(7), 1128-1138. McCollom, T. M. (2013). Miller-Urey and beyond: what have we learned about prebiotic organic synthesis reactions in the past 60 years?. Annual Review of Earth and Planetary Sciences, 41, 207-229. Meyer, J. R., Dobias, D. T., Weitz, J. S., Barrick, J. E., Quick, R. T., & Lenski, R. E. (2012). Repeatability and contingency in the evolution of a key innovation in phage lambda. Science, 335(6067), 428-432. Miller, S. L. (1953). A production of amino acids under possible primitive earth conditions. Science, 117(3046), 528-529. Mitchell, P. (1959). The origin of life and the formation and organizing functions of natural membranes. In The origin of life on the earth (pp. 437-443). Pergamon. 32 Muñoz-Gómez, S. A., Bilolikar, G., Wideman, J. G., & Geiler-Samerotte, K. (2021). Constructive neutral evolution 20 years later. Journal of molecular evolution, 89, 172- 182. Murren, C. J., Auld, J. R., Callahan, H., Ghalambor, C. K., Handelsman, C. A., Heskel, M. A., ... & Schlichting, C. D. (2015). Constraints on the evolution of phenotypic plasticity: limits and costs of phenotype and plasticity. Heredity, 115(4), 293-301. Ogbunugafor, C. B., Wylie, C. S., Diakite, I., Weinreich, D. M., & Hartl, D. L. (2016). Adaptive landscape by environment interactions dictate evolutionary dynamics in models of drug resistance. PLoS computational biology, 12(1), e1004710. Patel, S., Cortez, M. H., & Schreiber, S. J. (2018). Partitioning the effects of eco- evolutionary feedbacks on community stability. The American Naturalist, 191(3), 381- 394. Payne, J. L., & Wagner, A. (2019). The causes of evolvability and their evolution. Nature Reviews Genetics, 20(1), 24-38. Peng, H., Lelievre, A., Landenfeld, K., Müller, S., & Chen, I. A. (2022). Vesicle encapsulation stabilizes intermolecular association and structure formation of functional RNA and DNA. Current Biology, 32(1), 86-96. Poethke, H. J., Dytham, C., & Hovestadt, T. (2011). A metapopulation paradox: partial improvement of habitat may reduce metapopulation persistence. The American Naturalist, 177(6), 792-799. Post, D. M., & Palkovacs, E. P. (2009). Eco-evolutionary feedbacks in community and ecosystem ecology: interactions between the ecological theatre and the evolutionary play. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1523), 1629-1640. Preisner, E. C., Fichot, E. B., & Norman, R. S. (2016). Microbial mat compositional and functional sensitivity to environmental disturbance. Frontiers in microbiology, 7, 1632. Provine, W. B. (1986). Sewall Wright and evolutionary biology. Chicago: University of Chicago Press. 33 Rainey, P. B., & Travisano, M. (1998). Adaptive radiation in a heterogeneous environment. Nature, 394(6688), 69-72. Rebolleda‐Gómez, M., & Travisano, M. (2019). Adaptation, chance, and history in experimental evolution reversals to unicellularity. Evolution, 73(1), 73-83. Robertson, M. P., & Joyce, G. F. (2012). The origins of the RNA world. Cold Spring Harbor perspectives in biology, 4(5), a003608. Segré, D., Ben-Eli, D., & Lancet, D. (2000). Compositional genomes: prebiotic information transfer in mutually catalytic noncovalent assemblies. Proceedings of the National Academy of Sciences, 97(8), 4112-4117. Segré, D., Shenhav, B., Kafri, R., & Lancet, D. (2001). The molecular roots of compositional inheritance. Journal of Theoretical Biology, 213(3), 481-491. Shenhav, B., Segrè, D., & Lancet, D. (2003). Mesobiotic emergence: Molecular and ensemble complexity in early evolution. Advances in Complex Systems, 6(01), 15-35. Sojo, V., Herschy, B., Whicher, A., Camprubi, E., & Lane, N. (2016). The origin of life in alkaline hydrothermal vents. Astrobiology, 16(2), 181-197. Sousa, F. L., Thiergart, T., Landan, G., Nelson-Sathi, S., Pereira, I. A., Allen, J. F., ... & Martin, W. F. (2013). Early bioenergetic evolution. Philosophical Transactions of the Royal Society B: Biological Sciences, 368(1622), 20130088. Stoltzfus, A. (1999). On the possibility of constructive neutral evolution. Journal of molecular evolution, 49, 169-181. Suvorov, V., Saakian, D. B., & Lynch, M. (2023). A solution of the Crow-Kimura evolution model on fluctuating fitness landscape. Europhysics Letters, 142(5), 57003. Svensson, E. I. (2023). The structure of evolutionary theory: beyond Neo-Darwinism, Neo- Lamarckism and biased historical narratives about the Modern Synthesis. In Evolutionary biology: contemporary and historical reflections upon core theory (pp. 173-217). Cham: Springer International Publishing. Szathmáry, E., & Smith, J. M. (1995). The major evolutionary transitions. Nature, 374(6519), 227-232. 34 Szostak, J. W., Bartel, D. P., & Luisi, P. L. (2001). Synthesizing life. Nature, 409(6818), 387-390. Taylor, T., Bedau, M., Channon, A., Ackley, D., Banzhaf, W., Beslon, G., ... & Wiser, M. (2016). Open-ended evolution: Perspectives from the OEE workshop in York. Artificial life, 22(3), 408-423. Toparlak, O. D., & Mansy, S. S. (2019). Progress in synthesizing protocells. Experimental Biology and Medicine, 244(4), 304-313 Travis, J., Leips, J., & Rodd, F. H. (2013). Evolution in population parameters: density- dependent selection or density-dependent fitness?. The American Naturalist, 181(S1), S9-S20. Travisano, M., & Lenski, R. E. (1996). Long-term experimental evolution in Escherichia coli. IV. Targets of selection and the specificity of adaptation. Genetics, 143(1), 15-26. Travisano, M., Mongold, J. A., Bennett, A. F., & Lenski, R. E. (1995). Experimental tests of the roles of adaptation, chance, and history in evolution. Science, 267(5194), 87-90. Turner, C. B., Blount, Z. D., Mitchell, D. H., & Lenski, R. E. (2015). Evolution and coexistence in response to a key innovation in a long-term evolution experiment with Escherichia coli. bioRxiv, 020958. Vasas, V., Szathmáry, E., & Santos, M. (2010). Lack of evolvability in self-sustaining autocatalytic networks constraints metabolism-first scenarios for the origin of life. Proceedings of the National Academy of Sciences, 107(4), 1470-1475 Vincent, L., Berg, M., Krismer, M., Saghafi, S. T., Cosby, J., Sankari, T., ... & Baum, D. A. (2019). Chemical ecosystem selection on mineral surfaces reveals long-term dynamics consistent with the spontaneous emergence of mutual catalysis. Life, 9(4), 80. Wächtershäuser, G. (1988). Before enzymes and templates: theory of surface metabolism. Microbiological reviews, 52(4), 452-484. 35 Weiss, M. C., Sousa, F. L., Mrnjavac, N., Neukirchen, S., Roettger, M., Nelson-Sathi, S., & Martin, W. F. (2016). The physiology and habitat of the last universal common ancestor. Nature microbiology, 1(9), 1-8. Wimsatt, W. C., & Wimsatt, W. K. (2007). Re-engineering philosophy for limited beings: Piecewise approximations to reality. Harvard University Press. Wong, A., Rodrigue, N., & Kassen, R. (2012). Genomics of adaptation during experimental evolution of the opportunistic pathogen Pseudomonas aeruginosa. Wright, S. (1931). Evolution in Mendelian populations. Genetics, 16(2), 97. Yang, X., Heinemann, M., Howard, J., Huber, G., Iyer-Biswas, S., Le Treut, G., ... & Foster, P. J. (2021). Physical bioenergetics: Energy fluxes, budgets, and constraints in cells. Proceedings of the National Academy of Sciences, 118(26), e2026786118. Zaman, L., Meyer, J. R., Devangam, S., Bryson, D. M., Lenski, R. E., & Ofria, C. (2014). Coevolution drives the emergence of complex traits and promotes evolvability. PLoS Biology, 12(12), e1002023. 36 Chapter 1: Multispecies interactions shape the transition to multicellularity Abstract The origin of multicellularity transformed the adaptive landscape on Earth, opening diverse avenues for further innovation. The transition to multicellular life is understood as the evolution of cooperative groups which form a new level of individuality. Despite the potential for community-level interactions, most studies have not addressed the competitive context of this transition, such as competition between species. Here, we explore how interspecific competition shapes the emergence of multicellularity in an experimental system with two yeast species, Saccharomyces cerevisiae and Kluyveromyces lactis, where multicellularity evolves in response to selection for faster settling ability. We find that the multispecies context slows the rate of the transition to multicellularity, and the transition to multicellularity significantly impacts community composition. Multicellular K. lactis emerges first and sweeps through populations in monocultures faster than in cocultures with S. cerevisiae. Following the transition, the between-species competitive dynamics shift, likely in part to intraspecific cooperation in K. lactis. Hence, we document an eco-evolutionary feedback across the transition to multicellularity, underscoring how ecological context is critical for understanding the causes and consequences of innovation. By including two species, we demonstrate that cooperation and competition across several biological scales shapes the origin and persistence of multicellularity. 37 I. Introduction All life on Earth was shaped by cooperation and competition among individuals. Competition in particular has been emphasized for its role driving diversification and adaptation (Pfennig and Pfennig 2012, Baldauf et al. 2014, Shinen and Morgan 2009). In nature, several well-known examples like the adaptive radiation of finches on the Galápagos Islands (Grant and Grant 2002) or the global invasions of Mediterranean bay mussels (Shinen and Morgan 2009) illustrate the significance of these competitive dynamics. In the laboratory, competition among mutants demonstrates striking instances of parallelism and stochasticity, as high-fitness mutations sweep through replicate populations of E. coli (Blount et al. 2018, Travisano et al. 1995). Nevertheless, several decades of work have also brought greater recognition to the ubiquity of cooperation among evolving populations (Rainey and Rainey 2003, Griffin et al. 2004, Nadell et al. 2016), with the interplay of cooperation and conflict facilitating diversification events (Rainey and Travisano 1998, Friesen et al. 2004). Since cooperative behavior has been shown to provide competitive advantages against non-cooperators (Ghoul et al. 2014, Nowak 2006) – and cooperation often occurs among individuals also in competition (Aktipis and Maley 2017, Rius and McQuaid 2009) – it is clear that the two do not occur in isolation. Both cooperation and competition play important roles in natural systems (Crombie 1947, Sachs et al. 2004, Nowak 2012), and experimental studies have been key to disentangling their relative influences (Harcombe et al. 2018, Nahum et al. 2011, Majeed et al. 2011). The most pivotal moments of cooperation among groups are considered Major Evolutionary Transitions (METs) (Szathmáry and Smith 1995; West et al. 2015), more specifically referred to as Evolutionary Transitions in Individuality (Michod and Roze 1997, Michod 2011), where separately replicating units (e.g. single cells) formed new levels of individuality (e.g. multicellular organisms). When first described, the primary challenge for maintaining cooperative benefits across METs was thought to be the potential for genetic conflict within groups (Queller and Strassmann 2013). A central 38 question was therefore whether mechanisms for suppressing within-group conflict could invade when rare (Szathmáry and Smith 1995, Buss 1987, Michod 1996). Recent advances, however, have demonstrated the significance of cooperation and competition not only within but between groups, with the competitive benefits of group cooperation rapidly realized experimentally (Driscoll and Travisano 2017; Queller and Strassmann 2009; Bastiaans et al. 2015). These results indicate that in order to understand the emergence of within-group cooperation across METs, the potential for cooperation and conflict between groups also needs to be explored. In nature, competition among species is central to understanding evolutionary outcomes, and interspecific interactions have shaped the evolution of cooperation in many natural systems (Sachs et al. 2004). Thus far, the vast majority of experiments examining METs involve a single species (Ratcliff et al. 2012, Ratcliff et al. 2013, Hammerschmidt et al. 2014, Koschwanez et al. 2013), constraining our ability to explore the role of ecological context during the evolution of innovation. Analysis of eco-evolutionary dynamics across METs thus requires a system that incorporates these biological scales. How does competition between species shape the emergence of cooperative groups? The origin of multicellularity paved the way for adaptations of complex structures and functions and has been readily observed in the laboratory (Ratcliff et al. 2012; Baselga-Cervera et al. 2022). In particular, parallel kinds of multicellularity evolve in two different yeast species: Saccharomyces cerevisiae and Kluyveromyces lactis, in response to selection for settling through liquid media (Ratcliff and Travisano 2014; Driscoll and Travisano 2017). Despite having diverged phylogenetically ~ 100 million years ago, these distinct species evolved multicellularity via the same mechanism (‘staying together’ of daughter cells with mother cells) (Ratcliff et al. 2014; Tarnita et al. 2013). Unlike S. cerevisiae, however, K. lactis also cooperated with non-relatives to form even larger clusters via floccing (Driscoll and Travisano 2017). This cooperative behavior was observed between multicellular clusters, as well as through the attachment of unicellular ‘free riders’ to multicellular neighbors. Since floccing between groups likely provides 39 further competitive advantages when settling, it’s clear that key parameters facilitating the transition to multicellularity go beyond the suppression of within-group conflict. Rather, the emergence of multicellularity may depend on competitive benefits conferred via cooperation at the cellular, individual, and community levels. To explore the interplay of cooperation and competition across the transition to multicellularity, we evolved cocultures of S. cerevisiae and K. lactis with settling selection in separate experiments, starting with either unicellular or multicellular ancestors, i.e. beginning before and after both species had transitioned to multicellularity. Since species are rarely isolated in nature, interspecific competition during the emergence of biological innovation could be critical for claims about transitions in complexity. In particular, competition between species has important adaptive consequences, with evolutionary trajectories dependent on population densities (Clarke 1972; Travis et al. 2013), the tempo of adaptation (Ratcliff et al. 2013), and competitive ability for essential resources (Travisano and Lenksi 1996). Thus, in this system, we explored how interspecific competition influences the transition to multicellularity, as well as how the transition shifts the competitive ability among species. We found that the transition to multicellularity dramatically affected evolutionary outcomes, with K. lactis emerging first and sweeping through the populations in cocultures. The presence of both species significantly affected this transition, and the nature of interspecific interactions shifted once multicellular. Multicellular responses to selection arises from competition for settling ability both within and between species, in addition to any preexisting competitive and cooperative interactions. Together, this work documents an eco-evolutionary feedback across the transition to multicellularity, where interactions among species shape the emergence of cooperative groups, which in turn shapes interactions between species. We expand on single-species work, demonstrating how ecological context is critical for understanding transitions in individuality. II. Materials and Methods 40 (a) Strains and media The unicellular isolates used were S. cerevisiae strain Y55 and K. lactis strain NRLY-1140, both referred to as ‘ancestors’ in this study. Cultures were grown in 10 ml Yeast Peptone Dextrose media (YPD; 1% yeast extract, 2% peptone, 2% D-glucose) in 25 mm x 150 mm glass culture tubes, at 30°C and shaking at 250 r.p.m. Solid plates were prepared using Yeast Peptone Lactose media (YPL; 1% yeast extract, 2% peptone, 2% lactose, 1.5% agar), which we developed through pilot testing to allow differentiation between species. Clonal populations were established from single colonies. (b) Characterization of populations. Standard dilution plating techniques were used to measure colony-forming units (CFUs) for each population. To distinguish between species, plating on YPL media allowed K. lactis to form larger colonies than S. cerevisiae, which were easily identifiable after 48 hours of growth at 30°C. Presence of multicellular phenotypes was determined based on colony morphology (smooth unicellular colonies and rugose multicellular colonies) (Fig. 1). Figure 1. Unicellular (a) and multicellular (b) colonies of K. lactis and S. cerevisiae on YPL plates. Multicellularity is easily distinguished by the formation of rugose, rather than 41 smooth, colonies. Species are distinguished by the formation of large K. lactis colonies and small S. cerevisiae colonies, following 48 hours of growth in 30°C. (c) Settling selection experiments. Unicellular ancestors. To evaluate the effects of interspecific competition on the transition to multicellularity, we performed a selection experiment with co-cultures of both unicellular species at three different initial frequencies: 25%, 50%, and 75%, as well as monocultures of K. lactis and S. cerevisiae, separately. Six initially isogenic replicate populations per treatment were propagated with benchtop settling selection for 19 days. Every 24 hours, 1.0 ml aliquots of well-mixed culture were transferred to 1.5 ml microfuge tubes, where they sat undisturbed for 7 minutes. The bottom 100 µl was then carefully transferred to 10 ml of fresh YPD media. Six replicate control populations per treatment were propagated via standard daily transfers (without settling). Relative frequencies of each species and morphotype (smooth vs. rugose colonies) were measured via dilution plating on YPL plates before 24 hours of growth (Day 0), daily for the first 3 transfers, and then every other day until transfer 19. Ancestral strains, as well as 700 µl subsamples of each population following transfers 10 and 19, were preserved with 25% glycerol and stored at – 80°C. Multicellular invasion-from-rare. Since S. cerevisiae remained unicellular in cocultures during the above experiment, we conducted a separate experiment beginning with multicellular isolates to assess whether each species could invade when rare following the transition to multicellularity. Six replicate populations per treatment (5% initial frequency K. lactis and 5% initial frequency S. cerevisiae) were transferred every 24 hours with and without settling selection for 20 days. YPL plates were inoculated for each population after 24 hours of growth and after 10 and 20 transfers. Subsamples were preserved every 10 days with 25% glycerol at – 80°C. The change in the relative proportion of the two competing multicellular species was measured as the differential growth rate (s) (Dykhuizen and Hartl 1983): 42 𝑠 = ln %!!(#) !"(#) &, where x1(t) and x2(t) represent the relative frequency of S. cerevisiae and K. lactis by timepoint (t). Selection coefficients. To assess the rate that multicellular K. lactis swept through the population during settling selection with unicellular ancestors, we calculated the differential growth rate as the natural log of the frequency of multicellular K. lactis (x1) over the frequency of all other strains (x2) for each replicate population, beginning on the first day multicellular K. lactis emerged and ending when the population appeared to plateau (see above). We estimated the slope (selection coefficient) by linear regression of differential growth rate over these times and compared the mean selection coefficients across treatments. (d) Competition assays. To determine competitive dynamics across the MET, we performed a series of competition assays with unicellular and multicellular strains of each species. To assess competitive growth ability between species without settling, we inoculated 10 ml of YPD with 5 µl of ancestral K. lactis and 95 µl of ancestral S. cerevisiae (and vice versa), establishing six replicate populations when each species begins as rare. CFUs were measured via dilution plating before and after 24 hours of growth. To assess competitive settling ability, 10 ml of YPD were inoculated with 5 µl of one species and 95 µl of the other (all unicellular, six replicates each) and grown overnight. CFUs were then measured with dilution plating before and after one round of settling (bottom 100 µl of 1.0 ml subculture transferred following 7 minutes of benchtop settling). All of the above procedures were repeated for multicellular strains to determine competitive growth and settling abilities before and after the transition to multicellularity. 43 Forty-eight hour growth curve data was acquired for unicellular isolates using a Tecan infinite 200pro microplate reader. We transferred 5 µl of overnight cultures to 195 µl of YPD in 96-well plates, with ten replicates per treatment: monocultures of each species and cocultures of 50% each. Growth was measured by OD600 absorbance reads every 15 minutes. Population-level information like carrying capacity (K) and intrinsic growth rate (r) were determined by fitting growth curve data to the logistic equation for population size Nt at time t: 𝑁# = % &'(#$%&%& )*$'( , where N0 is the population size at the beginning of the growth curve [R package Growthcurver] (Sprouffske and Wagner 2016). To assess the possibility that each species could utilize a byproduct of the other, we grew K. lactis and S. cerevisiae on their own and each other’s spent media. To do this, we inoculated 10 ml of YPD with 100 µl of a grown culture of each species, separately, and grew the monocultures for 24 hours (15 tubes per species). Following overnight growth, we removed the cells using a sterile 0.22 µl filter, combining the spent media from each replicate into sterile glass bottles, and then distributing 10 ml into fresh tubes. Each tube was inoculated with 100 µl of overnight cultures, with 6 replicates per treatment (four treatments: each species grown on its own and each other’s spent media). After 24 hours of growth on spent media, density of cells was measured via dilution plating and counting CFUs. III. Results (a) K. lactis becomes multicellular first in monocultures 44 Monocultures of each unicellular species were propagated with and without settling selection for 19 days. When transferred with settling, multicellularity in K. lactis emerged first, with all six replicate populations containing multicellular individuals after 7 days of selection (Fig. 2a). In the S. cerevisiae monocultures, however, multicellularity did not appear until day 19 of the experiment (Fig. 2c) and was never observed without settling (Fig. 2d). The occasional appearance of multicellular colonies even in the K. lactis control populations (Fig. 2b) suggests that the mutation causing multicellular clusters either occurs more frequently than in populations of S. cerevisiae, or its effects are masked in S. cerevisiae heterozygotes (Baselga-Cervera et al. 2022). Figure 2. Relative frequencies of unicellular and multicellular K. lactis and S. cerevisiae in monocultures during 19 days with and without settling selection. Monocultures began as isogenic unicellular isolates with an initial frequency of 100% K. lactis or S. cerevisiae (6 replicate populations per treatment). Over the course of the experiment, multicellular K. lactis emerged and increased in frequency when propagated with settling (a) but not without 0 5 10 15 20 0 25 50 75 100 Time (Days) Fr eq ue nc y (% ) 100% initial frequency of K. lactis with settling selection K. lactis multi K. lactis uni 0 5 10 15 20 0 25 50 75 100 Time (Days) Fr eq ue nc y (% ) 100% initial frequency of S. cerevisiae with settling selection S. cerevisiae uni S. cerevisiae multi 0 5 10 15 20 0 25 50 75 100 Time (Days) Fr eq ue nc y (% ) 100% initial frequency of K. lactis without settling selection K. lactis uni K. lactis multi 0 5 10 15 20 0 25 50 75 100 Time (Days) Fr eq ue nc y (% ) 100% initial frequency of S. cerevisiae without settling selection S. cerevisiae uni S. cerevisiae multi a) b) c) d) 45 settling selection (b). Monocultures of S. cerevisiae remained unicellular until day 19 with settling selection (c) and never evolved multicellularity in control replicates (d). (b) Species coexist in the absence of settling Without settling selection, species remained unicellular and coexisted throughout the 19-day experiment in cocultures (Fig. 3b; Supplemental Materials, Fig. S1). To understand why they could coexist, we grew each species on the other’s spent media (see: Methods) and found that both species reached a greater density when grown on the spent media of the other species, compared to their own (Fig. 4) (K. lactis: p<0.001, S. cerevisiae: p=0.0041, one-way ANOVA, Supp. Table 1 and 2). This indicates that both species may be able to grow on a byproduct of the other, or otherwise preferentially utilize different resources in the media. The species therefore occupy sufficiently different niches when grown together in YPD media, enabling their coexistence. Figure 3. Relative frequencies of K. lactis and S. cerevisiae in cocultures during 19 days with and without settling selection. Cocultures began as isogenic unicellular isolates with an initial frequency of ~ 50% K. lactis (volumetric). Unicellular S. cerevisiae remains at a higher frequency than unicellular K. lactis until multicellular K. lactis emerges during settling selection (a). S. cerevisiae appears to be excluded in 3 out of the 6 experimental replicates (a), whereas species coexist in all control populations (b) (see Supplemental Materials for cocultures with initial frequencies of 25% and 75% K. lactis, Fig. S1 & S2). 0 5 10 15 20 0 25 50 75 100 Time (Days) Fr eq ue nc y (% ) 50% initial frequency by volume of K. lactis without settling selection K. lactis uni K. lactis multi S. cerevisiae uni S. cerevisiae multi 0 5 10 15 20 0 25 50 75 100 Time (Days) Fr eq ue nc y (% ) 50% initial frequency by volume of K. lactis with settling selectiona) b) 46 Figure 4. Density of cells after 24 hours of growth on spent media. K. lactis reached a greater density when grown on spent media from S. cerevisiae, compared to its own spent media (p<0.0001, one-way ANOVA, Suppl. Table 2). Likewise, there was a greater density of S. cerevisiae following growth on K. lactis spent media, compared to its own (p=0.0041, one-way ANOVA, Suppl. Table 3). Results indicate that the two species occupy different niches in YPD media, possibly due to their ability to grow on a byproduct of the other. (c) Multispecies context slows the emergence of multicellularity As expected from the above results in monocultures, multicellular K. lactis appeared first in cocultures with S. cerevisiae and rapidly rose in frequency during 19 days of settling selection (Fig. 3a, see Supplemental Materials for cocultures with initial frequencies of 25% and 75% K. lactis, Fig. S2). In some cases, multicellular K. lactis ultimately appeared to exclude S. cerevisiae; there were no S. cerevisiae colonies present per ~100 CFUs after 19 days in 3 out of 6 replicate populations when K. lactis had an initial frequency of 25% and 50%, and S. cerevisiae was absent in 4 out of 6 populations when K. lactis had an initial frequency of 75%. This suggests that a different tempo in the transition to multicellularity may lead to competitive exclusion. Unicellular K. lactis, however, was retained in the populations and sometimes increased in frequency at the end of the experiment. This is likely due to the adaptation of unicellular free-riders, which have K. lactis S. cerevisiae 0 1×106 2×106 3×106 4×106 5×106 Spent Media C FU /m l Growth on spent media K. lactis S. cerevisiae 47 been found to evolve a greater competitive ability, compared to unicellular ancestors, in the K. lactis species (Driscoll and Travisano 2017). The rate that multicellular K. lactis swept through the populations differed between treatments (Fig. 5). We calculated the selection coefficients during the sweep of multicellular K. lactis (see: Methods) and found that populations without S. cerevisiae (100% K. lactis) had a greater mean selection coefficient, compared to cocultures with initial frequencies of 75% (p=0.0086, Tukey-Kramer HSD, Supp. Table 3), 50% (p=0.0115), and 25% K. lactis (p=0.0234) (Table 1). There were no significant differences in the mean selection coefficients among cocultures, when initial frequency was treated as categorical (p>0.05, Tukey-Kramer HSD), nor when it was continuous (p>0.05, linear regression). Results indicate that the rate of the multicellular K. lactis sweep depends on the presence of S. cerevisiae; therefore, interspecies interactions shaped the response to selection. Figure 5. Frequency of K. lactis multicellular clusters during 19 days of settling selection, with initial unicellular K. lactis frequencies of 25%, 50%, 75%, and 100% (error bars ± SEM). Tempo of transition to multicellularity depends on time and presence of S. cerevisiae (r2=0.67, p<0.0001 [for both factors], linear regression). 0 5 10 15 20 0 25 50 75 100 Time (Days) M ul tic el lu la r K . l ac tis (% ) Emergence of multicellular K. lactis depends on presence of S. cerevisiae 25 50 75 100 48 Table 1. Selection coefficients during multicellular sweep of K. lactis for each initial frequency. Coefficients determined from the slope of each linear regression for the differential growth rate for each replicate (6 replicates per treatment). Populations without S. cerevisiae (100% initial frequency K. lactis) have greater selection coefficients during the increase in frequency of multicellular K. lactis, compared to cocultures with an initial frequency of 75% (p=0.0086, Tukey-Kramer HSD, Suppl. Table 1), 50% (p=0.0115), and 25% (p=0.0234). (d) Multicellular invasion is species-specific We performed a 20-day invasion-from-rare experiment beginning with multicellular isolates of both species, since S. cerevisiae did not become multicellular during 19 days of settling selection with K. lactis. We found that when both species began as multicellular, the dynamics of their invasion was species-specific. When S. cerevisiae began as rare, it initially only increased in frequency when grown without settling but later appeared to evolve greater competitive ability when settling, allowing both treatments to increase in frequency after 20 days (Fig. 6). Thus, differential growth rate between the species depends on time (adj. r2=0.824, p<0.0001, linear regression, Suppl. Table 4) and settling condition (p=0.032), with the effect of condition changing over time (p=0.0005). When K. lactis began as rare, it initially invaded when settling, but neither condition continued to increase in frequency between 10 and 20 days, demonstrating frequency- dependent selection. The differential growth rate in this case depends on time (adj. 49 r2=0.445, p=0<0.0001, linear regression, Suppl. Table 5) and settling condition (p=0.023), with an interaction between condition and time (p=0.0086). To determine whether these invasion dynamics differed by species, we compared the mean selection coefficients (i.e. the slope of the differential growth rate over time) between conditions when S. cerevisiae versus K. lactis was rare, for control and settling populations, during the first and final 10 days of the experiment. We found that the control populations were significantly different both during the first 10 days (p=0.0039, t-test, Suppl. Table 6) and final 10 days (p=0.0380, t-test, Supp. Table 6), whereas the settling populations did not differ significantly during the first 10 days (p=0.9310, t-test, Suppl. Table 6) but did between days 10 and 20 (0.0010, t-test, Suppl. Table 6), indicating that population dynamics depend on which species begins as rare. Thus, when the tempo of the transition to multicellularity is similar such that both species are multicellular, the species are able to coexist, but the nature of their invasion is species-specific. Figure 6. Differential growth rate of multicellular K. lactis and S. cerevisiae after 1, 10, and 20 days of selection, when (a) S. cerevisiae and (b) K. lactis begin as rare. When S. cerevisiae is rare, differential growth rate depends on time (adj. r2=0.0812, p<0.0001, linear regression, Suppl. Table 4) and settling condition (p=0.032), with the effect of condition changing over time (p=0.0005). When K. lactis is rare, differential growth rate depends on time (adj. r2=0.445, p<0.0001, linear regression, Suppl. Table 5) and settling condition (p=0.0233), with an interaction between condition and time (p=0.0086). Bold lines indicate 1 10 20 -4 -2 0 2 Time (Days) ln (x 1/ x 2) Multicellular S. cerevisiae invades when rare x1 = frequency of S. cerevisiae x2 = frequency of K. lactis a) b) 1 10 20 -4 -2 0 2 Time (Days) ln (x 1/ x 2) K. lactis dynamics are frequency-dependent x1 = frequency of K. lactis x2 = frequency of S. cerevisiae Settling Control 50 the mean for each treatment (error bars ± SEM), with fainter lines indicating individual replicates. (e) Competitive dynamics shift across the transition to multicellularity We obtained 48-hour growth curves to assess the competitive dynamics between species before the transition to multicellularity and found that monocultures of S. cerevisiae had the highest population growth rates (r) (r=0.808 ± 0.009 SE), followed by the cocultures (r=0.749 ± 0.004 SE), and monocultures of K. lactis (r=0.547 ± 0.003 SE), with a significant difference among the three treatments (p<0.0001, one-way ANOVA, Suppl. Table 7) (Supplemental Materials, Fig. S3). This finding aligns well with results from competition assays (Fig. 7), where the relative frequency of S. cerevisiae increased from rare after 24 hours of growth when unicellular (p=0.0006, linear regression, Suppl. Table 8), but the frequency of K. lactis when rare did not change (p=0.51, linear regression, Suppl. Table 9). Following 7 minutes of settling, however, there was no significant change in frequency from rare for either species (S. cerevisiae rare: p=0.45, linear regression, Suppl. Table 10; K. lactis rare: p=0.64, linear regression, Suppl. Table 11). Results indicate that S. cerevisiae has a competitive advantage when grown together (Fig. 7a), but neither species can invade as a result of settling alone (Fig. 7b), when unicellular. 51 Figure 7. Invasion dynamics of S. cerevisiae and K. lactis. Frequency of each species, grown together without settling when unicellular (a) and multicellular (c), and before and after settling when unicellular (b) and multicellular (d), when either K. lactis or S. cerevisiae begin as rare. Relative frequency of S. cerevisiae increases from rare after 24 hours of growth when unicellular (p=0.0006, linear regression) but only after 48 hours of growth when multicellular (p=0.0013, linear regression), whereas frequency of K. lactis when rare does not change in both conditions (unicellular: p=0.51, linear regression; multicellular: p=0.37, linear regression). Relative frequencies of unicellular K. lactis and S. cerevisiae do not change after 7 minutes of settling, when either species begins as rare (S. cerevisiae rare: p=0.45, linear regression; K. lactis rare: p=0.64, linear regression). When multicellular, relative frequency of K. lactis increases after 7 minutes of settling (p=0.0047, linear regression), whereas frequency of S. cerevisiae does not change significantly when rare (p=0.68, linear regression). Multicellular K. lactis has a competitive advantage when settling – an advantage that wasn’t present when both species were unicellular. Initial frequency of 0 24 0 5 10 15 Time (hours) Fr eq ue nc y (% ) 0 24 48 0 5 10 15 Time (hours) Fr eq ue nc y (% ) Before Settling After Settling 0 5 10 15 20 Fr eq ue nc y (% ) Frequency of K. lactis when rare Frequency of S. cerevisiae when rare Before Settling After Settling 0 5 10 15 20 Fr eq ue nc y (% ) Growth Settling U ni ce llu la r M ul tic el lu la r a) b) c) d) 52 K. lactis is greater due to smaller cell size and therefore more cells per milliliter (5 µl inoculation when rare). These competitive dynamics shift once the strains are multicellular. In the absence of settling, the increase in frequency of S. cerevisiae when grown together was no longer significant after 24 hours (p=0.19, linear regression, Suppl. Table 12), although it was after 48 hours (p=0.0013, linear regression, Suppl. Table 13), when multicellular. Similar to when unicellular, the frequency of K. lactis did not change significantly after 24 hours (p=0.37, linear regression, Suppl. Table 14) or 48 hours (p=0.96, linear regression, Suppl. Table 15) of growth (Fig. 7c). When settling, however, multicellular K. lactis increased significantly from rare (p=0.0047, linear regression, Suppl. Table 16), while the frequency of S. cerevisiae remained relatively unchanged (p=0.68, linear regression, Suppl. Table 17) (Fig. 7d). Thus, S. cerevisiae still appears to have a growth advantage when multicellular, but multicellularity provides a competitive settling advantage for K. lactis that was not present when unicellular. III. Discussion This work documents how ecological context shapes the evolution of multicellularity, which in turn shapes the ecological context of two species. We expand on previous findings in a single-species system, where sociality emerged and strengthened the benefits of multicellularity in K. lactis, underscoring the importance of ecological context across the transition. By incorporating both K. lactis and S. cerevisiae, we demonstrate not only the significance of cooperative and competitive dynamics across several scales but also how transitions in complexity can themselves be understood as eco-evolutionary feedbacks. Ecological context shapes evolution of multicellularity 53 When unicellular, we found that S. cerevisiae had a numerical advantage when grown with K. lactis. This was demonstrated both by their difference in growth rates and competition assays, where only S. cerevisiae could invade when rare. Nevertheless, the two species could coexist over evolutionary time, as observed in our control populations. Once multicellularity emerged, however, the rate at which multicellular K. lactis swept through the populations slowed when both species were initially present. This delay in the emergence of multicellularity may be owed to the greater competitive ability of S. cerevisiae when grown together, despite its disadvantage when settling. However, since interactions among species can include both negative (e.g. competitive) and positive components (e.g. as shown via faster growth on the other species’ spent media), we can only infer that the sum of these interactions with S. cerevisiae slowed the transition in K. lactis. In addition to shaping the selection coefficients of multicellular K. lactis, the presence of S. cerevisiae also appeared to increase the rate at which unicellular competitive ability evolved, as more unicellular K. lactis are present in cocultures at the end of the experiment, compared to K. lactis alone. Hence, the emergence of multicellularity and sociality in one species may depend on interactions with the other. Evolution of multicellularity shapes ecological context As a consequence of multicellularity arising in K. lactis, S. cerevisiae either became rare or appeared to be excluded from the populations. In the cases of competitive exclusion, this evolutionary outcome appears to mark a dramatic shift in the ecological landscape, where the species had previously coexisted. Exclusion also precludes the emergence of multicellularity in S. cerevisiae. In the instances where both species persist, S. cerevisiae becomes rare as a result of competition with multicellular K. lactis, which presumably would allow multicellularity to arise concurrently in both species. Even so, the tempo of the transition to multicellularity would likely be prolonged in S. cerevisiae, as the population would be smaller. Since multicellular invasion is species-specific, we 54 demonstrate how the tempo of this transition shapes community composition, an example of priority effects during the emergence of multicellularity. The evolution of multicellularity also shifts the competitive dynamics between species. Where neither species demonstrated a settling advantage when unicellular, only K. lactis can invade from rare when both are multicellular. This ecological shift agrees with previous findings that floccing among multicellular clusters provides a greater settling ability for K. lactis (Driscoll and Travisano 2017). Here, we document the evolutionary consequences of this intraspecific cooperation. In this experiment, the unicellular species compete for resources and in some cases cooperate perhaps through byproduct utilization. Our selection regime then introduces a second level of competition, which shifts the competitive environment: competition for settling ability. Since K. lactis is able to respond to this selection pressure faster, settling selection provides a benefit to K. lactis and alters the interaction landscape. Dynamics of adaptation and diversification Disentangling how life adapts and diversifies is central to understanding both the history of life on Earth and predicting its future. Experimental evolution has been an important tool for exploring processes of adaptation and diversification, in particular among microbial taxa (Kawecki et al. 2012). The Long-Term Evolution Experiment (LTEE) began in large part to explore these dynamics, and over thousands of generations, has challenged previous ideas that sustained adaptation usually only occurs in response to environmental change. Instead, it appears that while the rate of fitness gain declines over time, both adaptation and divergence can continue unbounded, even in a constant environment (Lenski et al. 2015). Adaptation even led to the emergence of a significant metabolic innovation in one out of the 12 LTEE lines, where E. coli gained the ability to use citrate (Blount et al. 2008). Interest in the lack of an “upper limit” to adaptation has also led to an increasing recognition for the role that niche construction plays during the diversification of species (Bajic et al. 2021). As organisms modify niches, they shift 55 selective pressures as well, leading to the convergence of ecological and evolutionary timescales and eco-evolutionary feedbacks (De Meester et al. 2019). These feedbacks have been well-studied over short timescales in single-species systems, where the effects of constructed environments depends on the abundance of organisms (e.g. frequency and density-dependent selection) (Clark 1972, Travis et al. 2013). Even rapid diversification events like adaptive radiations have been observed in single-species experiments, such as the emergence of niche specialists in Pseudomonas fluorescens, where negative frequency- dependent selection maintains distinct morphologies (Rainey and Travisano 1998). More recently, the importance of interspecific interactions in shaping eco- evolutionary dynamics has led to a large number of multispecies experimental studies (De Meester et al. 2019, Evans et al. 2020, Castledine et al. 2020, Brokhurst and Koskella 2013, Kassen et al. 2000, Gómez et al. 2022). By involving several species, these studies have demonstrated a positive feedback between microbial diversity, the construction of new ecological niches, and further diversification (Estrela et al. 2022, Evans et al. 2020), as well as the coexistence of several species via intransitive interactions (Kerr et al. 2002). In the work presented here, we observe species coexistence both prior to the emergence of multicellularity and when both species begin as multicellular. The dynamics in these two scenarios meet expectations where species coexist either by occupying different niches, or benefitting from niche construction of the other, as shown by growing each species on spent media. It is the emergence of multicellularity, however, that shifts these dynamics. Our findings therefore contribute to a less-explored area of the literature: the role of eco- evolutionary feedbacks during the emergence of biological innovation (Bajic et al. 2021, Aktipis and Maley 2017, Kikvidze and Callaway 2009). In our system, the appearance of multicellularity in K. lactis dramatically shifts interspecific interactions such that in several cases, S. cerevisiae is excluded. As a consequence, we observe divergent outcomes, whereby concurrent emergence of multicellularity in both species depends on historical contingencies. 56 Additionally, the dynamics in our system don’t appear to stabilize over time. As multicellularity and sociality emerge in K. lactis, new evolutionary strategies emerge as well, such as increased competitive ability in unicellular K. lactis, and the evolution of greater settling ability of multicellular S. cerevisiae when in competition with multicellular K. lactis. As a result, eco-evolutionary feedbacks seem to continue following this MET, possibly promoting further adaptation, divergence, and innovation. Previous literature on METs has been careful to caution that the adaptive benefits provided by these transitions cannot be used as an explanation for their origin (Száthmary and Maynard Smith 1995). Nevertheless, it can be easy for retrospective explanations to strip away the ecological complexity that must have been present. In this paper, we show that just one step in ecological complexity (i.e. two species instead of one) has profound consequences for how multicellularity emerges. Conclusion When the concept of Major Evolutionary Transitions was first proposed, a primary concern was the potential for within-group conflict to inhibit cooperation (Szathmáry and Smith 1995, Queller and Strassman 2013). In our system, this risk is mitigated since multicellularity arises via ‘staying together’ of cells during asexual reproduction, a possibility previously recognized when a multicellular individual is clonal (Queller 2000, Queller and Strassmann 2010, Maynard Smith 1989). Instead, our work highlights the potential for cooperation both within and between groups, as well as competition not only among unicellular ancestors but with other species in the community. Elements of determinism and stochasticity found here contribute to findings from other parallel replay experiments (Blount et al. 2018), as well the evolution of complexity more generally. While METs have long been recognized for their ability to transform adaptive landscapes, how the landscape transforms the transition itself – and its corresponding feedback – is key to understanding the evolution of complexity. 57 References 1. Pfennig, D. W., & Pfennig, K. S. (2012). Evolution's wedge: competition and the origins of diversity (No. 12). Univ of California Press. 2. Baldauf, S. A., Engqvist, L., & Weissing, F. J. (2014). Diversifying evolution of competitiveness. Nature Communications, 5(1), 5233. 3. Shinen, J. S., & Morgan, S. G. (2009). Mechanisms of invasion resistance: competition among intertidal mussels promotes establishment of invasive species and displacement of native species. Marine Ecology Progress Series, 383, 187-197. 4. Grant, P. R., & Grant, B. R. (2002). Adaptive radiation of Darwin's finches: Recent data help explain how this famous group of Galapagos birds evolved, although gaps in our understanding remain. American scientist, 90(2), 130-139. 5. Blount, Z. D., Lenski, R. E., & Losos, J. B. (2018). Contingency and determinism in evolution: Replaying life’s tape. Science, 362(6415), eaam5979. 6. Travisano, M., Mongold, J. A., Bennett, A. F., & Lenski, R. E. (1995). Experimental tests of the roles of adaptation, chance, and history in evolution. Science, 267(5194), 87-90. 7. Rainey, P. B., & Rainey, K. (2003). Evolution of cooperation and conflict in experimental bacterial populations. Nature, 425(6953), 72-74. 8. Griffin, A. S., West, S. A., & Buckling, A. (2004). Cooperation and competition in pathogenic bacteria. Nature, 430(7003), 1024-1027. 9. Nadell, C. D., Drescher, K., & Foster, K. R. (2016). Spatial structure, cooperation and competition in biofilms. Nature Reviews Microbiology, 14(9), 589-600. 10. Rainey, P. B., & Travisano, M. (1998). Adaptive radiation in a heterogeneous environment. Nature, 394(6688), 69-72. 11. Friesen, M. L., Saxer, G., Travisano, M., & Doebeli, M. (2004). Experimental evidence for sympatric ecological diversification due to frequency‐dependent competition in Escherichia coli. Evolution, 58(2), 245-260. 58 12. Ghoul, M., Griffin, A. S., & West, S. A. (2014). Toward an evolutionary definition of cheating. Evolution, 68(2), 318-331. 13. Nowak, M. A. (2006). Five rules for the evolution of cooperation. science, 314(5805), 1560-1563. 14. Aktipis, A., & Maley, C. C. (2017). Cooperation and cheating as innovation: insights from cellular societies. Philosophical Transactions of the Royal Society B: Biological Sciences, 372(1735), 20160421. 15. Rius, M., & McQuaid, C. D. (2009). Facilitation and competition between invasive and indigenous mussels over a gradient of physical stress. Basic and Applied Ecology, 10(7), 607-613. 16. Crombie, A. C. (1947). Interspecific competition. The Journal of Animal Ecology, 44- 73. 17. Sachs, J. L., Mueller, U. G., Wilcox, T. P., & Bull, J. J. (2004). The evolution of cooperation. The Quarterly review of biology, 79(2), 135-160. 18. Nowak, M. A. (2012). Evolving cooperation. Journal of theoretical biology, 299(0), 1- 8. 19. Harcombe, W. R., Chacón, J. M., Adamowicz, E. M., Chubiz, L. M., & Marx, C. J. (2018). Evolution of bidirectional costly mutualism from byproduct consumption. Proceedings of the National Academy of Sciences, 115(47), 12000-12004. 20. Nahum, Joshua R., Brittany N. Harding, and Benjamin Kerr. "Evolution of restraint in a structured rock–paper–scissors community." Proceedings of the National Academy of Sciences 108, no. supplement_2 (2011): 10831-10838. 21. Majeed, H., Gillor, O., Kerr, B., & Riley, M. A. (2011). Competitive interactions in Escherichia coli populations: the role of bacteriocins. The ISME journal, 5(1), 71-81. 22. Szathmáry, E., & Smith, J. M. (1995). The major evolutionary transitions. Nature, 374(6519), 227-232. 59 23. West, S. A., Fisher, R. M., Gardner, A., & Kiers, E. T. (2015). Major evolutionary transitions in individuality. Proceedings of the National Academy of Sciences, 112(33), 10112-10119. 24. Michod, R. E., & Roze, D. (1997). Transitions in individuality. Proceedings of the Royal Society of London. Series B: Biological Sciences, 264(1383), 853-857. 25. Michod, R. E. (2011). Evolutionary transitions in individuality: multicellularity and sex (Vol. 2348, pp. 167-97). Cambridge: MIT Press. 26. Queller, D. C., & Strassmann, J. E. (2013). Experimental evolution of multicellularity using microbial pseudo-organisms. Biology letters, 9(1), 20120636. 27. Buss, L. W. (2014). The evolution of individuality (Vol. 796). Princeton University Press. 28. Michod, R. E. (1996). Cooperation and conflict in the evolution of individuality. II. Conflict mediation. Proceedings of the Royal Society of London. Series B: Biological Sciences, 263(1372), 813-822. 29. Driscoll, W. W., & Travisano, M. (2017). Synergistic cooperation promotes multicellular performance and unicellular free-rider persistence. Nature communications, 8(1), 15707. 30. Queller, D. C., & Strassmann, J. E. (2009). Beyond society: the evolution of organismality. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1533), 3143-3155. 31. Bastiaans, E., Debets, A. J., & Aanen, D. K. (2016). Experimental evolution reveals that high relatedness protects multicellular cooperation from cheaters. Nature communications, 7(1), 11435. 32. Ratcliff, W. C., Denison, R. F., Borrello, M., & Travisano, M. (2012). Experimental evolution of multicellularity. Proceedings of the National Academy of Sciences, 109(5), 1595-1600. 60 33. Ratcliff, W. C., Herron, M. D., Howell, K., Pentz, J. T., Rosenzweig, F., & Travisano, M. (2013). Experimental evolution of an alternating uni-and multicellular life cycle in Chlamydomonas reinhardtii. Nature communications, 4(1), 2742. 34. Hammerschmidt, K., Rose, C. J., Kerr, B., & Rainey, P. B. (2014). Life cycles, fitness decoupling and the evolution of multicellularity. Nature, 515(7525), 75-79. 35. Koschwanez, J. H., Foster, K. R., & Murray, A. W. (2013). Improved use of a public good selects for the evolution of undifferentiated multicellularity. Elife, 2, e00367. 36. Baselga-Cervera, B., Gettle, N., & Travisano, M. (2022). Loss-of-heterozygosity facilitates a fitness valley crossing in experimentally evolved multicellular yeast. Proceedings of the Royal Society B, 289(1976), 20212722. 37. Ratcliff, W. C., & Travisano, M. (2014). Experimental evolution of multicellular complexity in Saccharomyces cerevisiae. Bioscience, 64(5), 383-393. 38. Tarnita, C. E., Taubes, C. H., & Nowak, M. A. (2013). Evolutionary construction by staying together and coming together. Journal of theoretical biology, 320, 10-22. 39. Clarke, B. (1972). Density-dependent selection. The American Naturalist, 106(947), 1- 13. 40. Travis, J., Leips, J., & Rodd, F. H. (2013). Evolution in population parameters: density- dependent selection or density-dependent fitness?. The American Naturalist, 181(S1), S9-S20. 41. Ratcliff, W. C., Pentz, J. T., & Travisano, M. (2013). Tempo and mode of multicellular adaptation in experimentally evolved Saccharomyces cerevisiae. Evolution, 67(6), 1573-1581. 42. Travisano, M., & Lenski, R. E. (1996). Long-term experimental evolution in Escherichia coli. IV. Targets of selection and the specificity of adaptation. Genetics, 143(1), 15-26. 43. Dykhuizen, D. E., & Hartl, D. L. (1983). Selection in chemostats. Microbiological reviews, 47(2), 150-168. 61 44. Sprouffske, K., & Wagner, A. (2016). Growthcurver: an R package for obtaining interpretable metrics from microbial growth curves. BMC bioinformatics, 17(1), 1-4. 45. Kawecki, T. J., Lenski, R. E., Ebert, D., Hollis, B., Olivieri, I., & Whitlock, M. C. (2012). Experimental evolution. Trends in ecology & evolution, 27(10), 547-560. 46. Lenski, R. E., Wiser, M. J., Ribeck, N., Blount, Z. D., Nahum, J. R., Morris, J. J., ... & Hajela, N. (2015). Sustained fitness gains and variability in fitness trajectories in the long-term evolution experiment with Escherichia coli. Proceedings of the Royal Society B: Biological Sciences, 282(1821), 20152292. 47. Blount, Z. D., Borland, C. Z., & Lenski, R. E. (2008). Historical contingency and the evolution of a key innovation in an experimental population of Escherichia coli. Proceedings of the National Academy of Sciences, 105(23), 7899-7906. 48. Bajić, D., Rebolleda-Gómez, M., Muñoz, M. M., & Sánchez, Á. (2021). The macroevolutionary consequences of niche construction in microbial metabolism. Frontiers in Microbiology, 12, 718082. 49. De Meester, L., Brans, K. I., Govaert, L., Souffreau, C., Mukherjee, S., Vanvelk, H., ... & Urban, M. C. (2019). Analysing eco‐evolutionary dynamics—The challenging complexity of the real world. Functional Ecology, 33(1), 43-59. 50. Evans, R., Beckerman, A. P., Wright, R. C., McQueen-Mason, S., Bruce, N. C., & Brockhurst, M. A. (2020). Eco-evolutionary dynamics set the tempo and trajectory of metabolic evolution in multispecies communities. Current biology, 30(24), 4984-4988. 51. Castledine, M., Padfield, D., & Buckling, A. (2020). Experimental (co) evolution in a multi‐species microbial community results in local maladaptation. Ecology Letters, 23(11), 1673-1681. 52. Brockhurst, M. A., & Koskella, B. (2013). Experimental coevolution of species interactions. Trends in ecology & evolution, 28(6), 367-375. 53. Kassen, R., Buckling, A., Bell, G., & Rainey, P. B. (2000). Diversity peaks at intermediate productivity in a laboratory microcosm. Nature, 406(6795), 508-512. 62 54. Gómez, P., Hall, A. R., Paterson, S., & Buckling, A. (2022). Rapid decline of adaptation of Pseudomonas fluorescens to soil biotic environment. Biology Letters, 18(3), 20210593. 55. Estrela, S., Diaz-Colunga, J., Vila, J. C., Sanchez-Gorostiaga, A., & Sanchez, A. (2022). Diversity begets diversity under microbial niche construction. BioRxiv, 2022- 02. 56. Kerr, B., Riley, M. A., Feldman, M. W., & Bohannan, B. J. (2002). Local dispersal promotes biodiversity in a real-life game of rock–paper–scissors. Nature, 418(6894), 171-174. 57. Kikvidze, Z., & Callaway, R. M. (2009). Ecological facilitation may drive major evolutionary transitions. BioScience, 59(5), 399-404. 58. Queller, D. C. (2000). Relatedness and the fraternal major transitions. Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences, 355(1403), 1647-1655. 59. Queller, D. C., & Strassmann, J. E. (2010). Complex societies: eusociality and multicellularity. Evolutionary behavioral ecology, 327-340. 60. Maynard Smith, J. (1988). Evolutionary progress and levels of selection. Evolutionary progress, 219-230. 61. Kalambokidis, M., & Travisano, M. (2023). Data from: Multispecies competition shapes the transition to multicellularity (Version 1) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.7834316 63 Supplemental Materials Figure S1. Control populations propagated for 19 days without settling. Cocultures began as isogenic unicellular isolates with an initial frequency of 25% (a) and 75% (b) K. lactis. Over the course of the experiment, populations mostly remained unicellular, with multicellular K. lactis occasionally appearing at low frequencies. Figure S2. Relative frequencies of K. lactis and S. cerevisiae during 19 days of settling selection. Cocultures began as isogenic unicellular isolates with an initial frequency of 25% (a) and 75% (b) K. lactis. Over the course of the experiment, multicellular K. lactis emerged and increased in frequency. Frequency of unicellularity depends on time and the presence of multicellular clusters. 0 5 10 15 20 0 25 50 75 100 Day Fr eq ue nc y (% ) 25% initial frequency of K. lactis with settling selection K. lactis uni K. lactis multi S. cerevisiae uni S. cerevisiae multi 0 5 10 15 20 0 25 50 75 100 Day Fr eq ue nc y (% ) 25% initial frequency of K. lactis without settling selection 0 5 10 15 20 0 25 50 75 100 Day Fr eq ue nc y (% ) 75% initial frequency of K. lactis without settling selection a) b) a) b) 0 5 10 15 20 0 25 50 75 100 Day Fr eq ue nc y (% ) 25% initial frequency of K. lactis with settling selection 0 5 10 15 20 0 25 50 75 100 Day Fr eq ue nc y (% ) 75% initial frequency of K. lactis with settling selection 0 5 10 15 20 0 25 50 75 100 Day Fr eq ue nc y (% ) 25% initial frequency of K. lactis with settling selection K. lactis uni K. lactis multi S. cerevisiae uni S. cerevisiae multi 64 Figure S3. Growth curves for cultures with a single species (S. cerevisiae; K. lactis) and both species (50% each), grown over 48 hours. Optical density (OD600) measured every 15 minutes using a Tecan infinite 200pro. The three treatments (10 replicates each) had significantly different carrying capacities (K) and intrinsic population growth rates (r) (p<0.0001 [for both factors], one-way ANOVA). Monocultures of S. cerevisiae had the highest carrying capacity (K=1.220 ± 0.006 SE) and population growth rate (r=0.808 ± 0.009 SE), followed by the cocultures (K=1.106 ± 0.006 SE; r=0.749 ± 0.004 SE), and monocultures of K. lactis (K=0.980 ± 0.007 SE; r=0.547 ± 0.003 SE). Suppl. Table 1. Analysis of Variance of K. lactis cell abundance (CFUs) on K. lactis and S. cerevisiae spent media. Source DF Sum of Squares Mean Square F Ratio Prob > F Spent Media 1 76002.083 76002.1 135.4077 <0.0001 * Error 10 5612.833 561.3 C. Total 11 81614.917 0 10 20 30 40 50 0.0 0.5 1.0 1.5 Time (hours) O pt ic al D en si ty (O D 60 0) Growth Curve K. lactis S. cerevisiae Coculture (50%) 65 Suppl. Table 2. Analysis of Variance of S. cerevisiae cell abundance (CFUs) on K. lactis and S. cerevisiae spent media. Source DF Sum of Squares Mean Square F Ratio Prob > F Spent Media 1 28616.333 28616.3 13.6683 0.0041 * Error 10 20936.333 2093.6 C. Total 11 49552.667 Suppl. Table 3. Analysis of Variance of the selection coefficients and Tukey-Kramer HSD connecting letters for cocultures and K. lactis monoculture during multicellular sweep. Selection coefficients calculated as the slope of the linear regression model of differential growth rate over time for each population. Connecting Levels Report Treatment Grouping Mean 100 A 1.0299991 25.75 B 0.6800059 50.50 B 0.6442698 75.25 B 0.6296753 Ordered Differences Report Level -Level Difference p-value 100 75.25 0.4003239 0.0086 * 100 50.50 0.3857293 0.0115 * 100 25.75 0.3499932 0.0234 * 25.75 75.25 0.0503305 0.9680 25.75 50.50 0.0357360 0.9880 66 50.50 75.25 0.0145945 0.9992 * Levels not connected by the same letter are significantly different Suppl. Table 4. Linear regression analysis of differential growth rate when S. cerevisiae begins as rare. Differential growth rate was calculated as the ln(x1/x2), where x1 is frequency of multicellular K. lactis (%) and x2 is frequency of multicellular S. cerevisiae (%). Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 2.9654901 0.11185 26.51 <0.0001 * Time -0.005689 0.000416 -13.66 <0.0001 * Condition -0.178878 0.080692 -2.22 0.0319 * Condition*Time -0.00156 0.000416 -3.75 0.0005 * Adj. R2: 0.811703, df = 44 Suppl. Table 5. Linear regression analysis of differential growth rate when K. lactis begins as rare. Differential growth rate was calculated as the ln(x1/x2), where x1 is frequency of multicellular K. lactis (%) and x2 is frequency of multicellular S. cerevisiae (%). Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) -1.128777 0.223764 -5.04 <0.0001 * Time 0.0044652 0.000927 4.82 <0.0001 * Condition -0.614994 0.261288 -2.35 0.0233 * 67 Condition*Time -0.002555 0.000927 -2.76 0.0086 * Time*Time -2.223e-5 5.883e-6 -3.78 0.0005 * Time*Time*Condition 1.1385e-5 5.883e-6 1.94 0.0597 Adj. R2: 0.378479, df = 42 Suppl. Table 6. T-test comparisons of mean selection coefficients when S. cerevisiae is rare versus when K. lactis is rare, for control and settling populations, during the first and last 10 days of the multicellular invasion experiment. Condition Time DF T Ratio Prob > t Control Day 1 - 10 6.903722 3.702533 0.0039 * Control Day 10 - 20 8.725115 2.01205 0.0380 * Settling Day 1 - 10 9.996444 -1.61219 0.9310 Settling Day 10 - 20 8.512716 4.37352 0.0010 * Suppl. Table 7. Analysis of Variance of growth rate for monocultures and coculture. Growth rates (r) were determined by fitting growth curve data to a logistic equation for population size [R package Growthcurver]. Source DF Sum of Squares Mean Square F Ratio Prob > F Treatment 2 0.37463322 0.187317 497.8785 <0.0001 * Error 27 0.01015820 0.000376 C. Total 29 0.38479142 68 Suppl. Table 8. Linear regression analysis of unicellular species frequency over 24 hours of growth, when S. cerevisiae begins as rare. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 2.3343848 0.554144 4.21 0.0018 * Time 3.8326944 0.783678 4.89 0.0006 * Adj. R2: 0.675693, df = 10 Suppl. Table 9. Linear regression analysis of unicellular species frequency over 24 hours of growth, when K. lactis begins as rare. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 10.698364 1.744067 6.13 0.0001* Time -1.675641 2.466484 -0.68 0.5123 Adj. R2: -0.05147, df = 10 Suppl. Table 10. Linear regression analysis of unicellular species frequency following 7 minutes of settling, when S. cerevisiae begins as rare. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 3.1014948 0.660274 4.70 0.0008 * Time 0.7262962 0.933768 0.78 0.4547 Adj. R2: -0.03725, df = 10 69 Suppl. Table 11. Linear regression analysis of unicellular species frequency following 7 minutes of settling, when K. lactis begins as rare. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 10.648933 1.657351 6.43 <0.0001 * Time -1.11852 2.343848 -0.48 0.6435 Adj. R2: -0.07551, df = 10 Suppl. Table 12. Linear regression analysis of multicellular species frequency over 24 hours of growth, when S. cerevisiae begins as rare. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 3.3766688 1.080771 3.12 0.0108 * Time 2.1485787 1.528441 1.41 0.1901 Adj. R2: 0.081503, df = 10 Suppl. Table 13. Linear regression analysis of multicellular species frequency over 48 hours of growth, when S. cerevisiae begins as rare. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 3.3766688 1.515321 2.23 0.0500 * 70 Time 4.7474481 1.071494 4.43 0.0013 * Adj. R2: 0.628767, df = 10 Suppl. Table 14. Linear regression analysis of multicellular species frequency over 24 hours of growth, when K. lactis begins as rare. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 6.8477662 2.740547 2.50 0.0315 * Time -3.610595 3.875718 -0.93 0.3735 Adj. R2: -0.01216, df = 10 Suppl. Table 15. Linear regression analysis of multicellular species frequency over 48 hours of growth, when K. lactis begins as rare. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 6.8477662 2.831313 2.42 0.0361 * Time -0.1076 2.00204 -0.05 0.9582 Adj. R2: -0.09968, df = 10 71 Suppl. Table 17. Linear regression analysis of multicellular species frequency following 7 minutes of settling, when S. cerevisiae begins as rare. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 3.376704 0.69064 4.89 0.0006 * Time 0.4137366 0.976712 0.42 0.6808 Adj. R2: -0.08061, df = 10 Suppl. Table 16. Linear regression analysis of multicellular species frequency following 7 minutes of settling, when K. lactis begins as rare. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 10.285159 1.378758 7.46 <0.0001 * Time 7.0667137 1.949859 3.62 0.0047 * Adj. R2: 0.524529, df = 10 72 Chapter 2: The maintenance and emergence of diversity promotes open- ended evolution in a pre-cellular system Abstract The evolution of novelty is central to understanding the emergence of life. We explored how selective contexts promote innovative outcomes, using an empirical model for pre- cellular evolution: in vitro selection of single-stranded DNA with two binding substrates, magnetic beads and yeast cells. Evolution was allowed to proceed in response to selective pressures both for binding ability and replication efficiency, and variable regimes were incorporated to investigate responses across eco-evolutionary contexts. We found that sequence diversity decreases over eight rounds of selection with beads, whereas diversity is not only maintained but generated during selection with cells. Particular sequences predominated in populations selected with beads, while diversification during selection with cells produced long, complex sequences. This open-ended evolutionary response with cells introduced new phenotypes through the recombination of separate genotypes. Conditions with low selection stringency facilitated the emergence of these recombinant sequences by allowing diversity to persist, particularly during variable selection with two substrates. These results demonstrate how the maintenance of diversity provides opportunities for ecological interactions which promote the emergence of innovation, transforming a pre-cellular landscape. 73 I. Introduction The complexity of life today attests to the evolution of innovation at its origin. How novel traits and increased complexity arise is therefore key to understanding the emergence of life (1-3). While evolution is often unpredictable, a vast amount of work has aimed to determine the appearance of biological precursors in a series of prebiotic systems, such as during the origins of inheritance (4-6), encapsulation (7-8), or metabolism (9-10). Among these approaches is the use of in vitro selection to isolate functional molecules from large, random pools, i.e. “systematic evolution of ligands by exponential enrichment” (SELEX) (11-14). This work has led to several important insights, particularly in characterizing the catalytic and binding capabilities of RNA molecules (12, 15-18). However, trajectories during laboratory selection tend to follow a narrow evolutionary path, where functional motifs are defined and then refined during subsequent rounds of selection (19). For instance, in vitro selection often relies on predetermining a goal (e.g. a particular catalytic function) and implementing selective conditions that ensure it is achieved (e.g. increased selection stringency) – an approach not only limited by the researcher’s imagination but also the particular hypothesis that presupposed the molecule’s existence. Instead, we draw from an existing body of literature in ecology and evolutionary biology, which emphasizes the importance of ecological interactions, genetic drift, and environmental variability in promoting evolutionary innovation (20-24). In contrast with most SELEX studies where selection stringency is increased in order to isolate highly functional sequences, studies of eco-evolutionary processes today have shown that more relaxed selective conditions can provide contexts favorable for innovative solutions (25-27). Genetic drift and natural selection can act together during adaptive evolution, either by maintaining diversity that is subsequently selected on (25) or crossing fitness valleys before climbing new adaptive peaks (28-29). Experimental studies that allow multiple possible solutions to given selective pressures – a cornerstone of natural selection – have been successful in driving the evolution of novel traits such as 74 multicellularity (30) and collective action (31). Similarly, the “deformation” of fitness landscapes via niche construction and eco-evolutionary feedbacks can play a key role in circumventing evolutionary plateaus (32-33). This emphasis on organism-environment interactions further signifies the importance of genetic diversity both for persistence within and as an outcome of evolution in variable environments (34-36), where spatial and temporal heterogeneity can promote diversification (34-39). Moreover, evolving populations are almost always subject to competing selective pressures, which can lead to the coexistence of multiple phenotypes and opportunities for complex interactions (40). This work illustrates that maintaining diversity – either through reduced selective stringency, environmental variability, or niche partitioning – may provide contexts favorable for novel interactions that alter adaptive possibilities. Here, we incorporate these insights into an in vitro selection experiment, using single-stranded DNA (ssDNA) as a model for pre-cellular evolution, given its tractability and evolvability (41-42), as well as its potential relevance during abiogenesis, since DNA may have appeared concurrently with RNA, rather than being its later descendant (5). DNA can fold into unique 3D structures that bind to other molecules, so the SELEX process selects on DNA’s phenotype (i.e. its 3D conformation), which is dictated by its genotype (i.e. nucleotide sequence), allowing for the enrichment of sequences with high- affinity binding. We completed eight rounds of selection with a single binding target (magnetic beads or yeast cells) and four rounds with one target before switching to the other for the remaining four rounds (i.e. four rounds with beads before switching to cells, and vice versa). Hence, our selection process incorporated temporal variability via switching targets and spatial heterogeneity as the complexity of the cells’ surfaces (e.g. surface-bound proteins, chitin of the cell wall etc.). We asked: How does selection in a variable environment shape sequence diversity? And conversely: How does diversity shape adaptation in a variable environment? By applying eco-evolutionary contexts which promote novelty in extant life, our aim was to explore how innovation can emerge in a pre- cellular system. 75 This novel approach builds on previous work using the SELEX procedure, but critically, our goals are distinct from most prior experiments, which have aimed to identify aptamers with high binding affinity and specificity. In this study, we permit evolution to proceed without systematic interference to optimize selection’s ability to promote better- binding aptamers. Instead, we explicitly relax selection pressures to allow for both neutral and adaptive processes. We therefore deviate from standard practices in several key ways. First, our protocol remained constant through time, rather than increasing in stringency, providing more possibilities to persist through rounds of selection. Second, we incorporate temporally variable selection, which would of course be detrimental if our goal was to isolate high-specificity aptamers. Third, we do not eliminate the reverse strand following amplification, allowing both strands to persist and therefore retain a diversity of non- selected sequences each round. In doing so, we allow for the effects of multiple selection pressures in this system – selection for binding to the substrate as well as amplification efficiency during PCR – akin to many microbial evolution experiments, where there may be selection both to perform a particular function and for faster growth rate. Derived phenotypes will therefore depend on tradeoffs between these pressures, along with any neutral processes like drift or de novo mutations. Some studies of in vitro selection with RNA ribozymes illustrate the importance of these ecological and evolutionary processes. For example, previous work has demonstrated the link between high initial diversity and the rate of adaptation (43), the ability for ribozymes to diversify and occupy different niches (44), the limits of neutral drift to introduce variation when compartmentalized (45), and the potential for low selection pressures to facilitate the emergence of cooperative mutations (46). Here, we show that open-ended evolution emerges from including even more biological complexity – environmental variability and heterogeneity, in addition to multiple sources of relaxed selection pressures. 76 We found that diversity was not simply preserved, but also promoted during selection with cells, compared to a decrease in diversity with beads, and the diversity itself likely enabled the emergence of novel structures. Selection in static and variable environments led to unique forms of adaptation: one where better-binding sequences rose Box. 1 A note on prebiotic relevance The inability to recreate primordial conditions has long been a dilemma in research on the origins of life. Even if we had perfect knowledge of environments ~4 billion years ago, the unpredictability of evolution renders it impossible to recreate each step during the emergence of life (24). As a result, there are no natural systems or observational studies; the field relies solely on model systems to explore prebiotic evolution. In this study, we use a non-cellular system as an empirical model to investigate prebiotic evolutionary processes. It is important to note that we are not trying to recapitulate events during abiogenesis or generate chemical entities we thought were present, since Earth during the early Archean did not include thermocyclers, eukaryotic yeast cells, or possibly even ssDNA. Our approach, instead, is to explore life-like processes that were likely present, rather than particular structures or geochemical conditions. While the materials we use do not reflect a prebiotic context, we consider the selective conditions to be more prebiotically plausible than a typical aptamer selection regime. An analogy is the Long-Term Evolution Experiment (LTEE) with E. coli, which provides general insights into evolutionary processes by leveraging a microbial model system, despite reliance on a simple, artificial environment (47). Since most aptamer selection studies impart strong selection or attempt to isolate high-function individuals, their outcomes may be evolutionarily fragile; optimization to specific selection pressures, as well as reduced variation, likely limits their ability to persist in a dynamic prebiotic landscape. Thus, we reduced selection stringency and introduced variable environments, exploring how the maintenance of diversity enables ecological interactions. This approach aligns with evolutionary theory, which indicates that drift in combination with selection may play an important role during complexification. Localization on a surface has been proposed to be important during the emergence of the first evolvable systems (48-51). Spatial structure is also widely found to be important for the evolution of cooperative interactions in life today (52-54). Hence, we co-opt a simple, evolvable system (ssDNA) and incorporate selection pressures for growth and surface attachment. We compare outcomes across conditions to identify eco-evolutionary contexts which promote innovative solutions. 77 in frequency and another where the process of diversification opened niches for long sequences. This diversification process generated new phenotypes of greater structural complexity whose genotypes provided avenues for further adaptation, indicative of open- ended evolution. By employing an “eco-evolutionary” approach to the origins of life (24), we demonstrate how innovation can evolve and transform a pre-cellular landscape. II. Results (a) High-Throughput Sequencing The sequences in the starting DNA library consisted of 40 random nucleotides, with both the 5’- and 3’- ends flanked by 18-base pair defined priming sites for PCR amplification (XELEX DNA Core Kit). Since these sequences contained a 40-nt region with two 18-nt primer binding sites, the length of sequences in the starting library was around 76-nt. All DNA pools from rounds 1, 4, 7, and 8 of selection (3 replicates per treatment), as well as the random starting library and control library of a single sequence, were submitted for next-generation sequencing (Illumina NovaSeq S4 2x150-bp lane). Sequencing from 62 samples generated 1,843,958,209 reads in total, with a mean of 29,741,261 pair-end reads per sample (38 samples used in this study). Mean Phred scores were ≥ Q30, meaning 99.9% of correct likelihood call. We assessed each populations’ response to selection with either beads or cells as the binding substrate (Fig. 1) or during variable selection where the binding substrate switched from beads to cells or vice versa (Fig. 2). All derived populations were then characterized (Fig. 3). 78 (b) Selection with beads leads to sequence enrichment As expected, iterative cycles of selection with streptavidin-coated magnetic beads led to the enrichment of particular sequences (Fig. 1, A to C). Beginning with an extremely diverse ssDNA library of random 40-nt sequences flanked by primer binding sites (~1013 to 1015 different molecules), all sequences were initially rare. The most common sequences in the final selected pools increased from rare over 8 selection cycles, indicating the proliferation of sequences with greater-binding affinity, with the 20 highest-frequency sequences collectively comprising ~50% of the final population (Fig. 1, A to C). This enrichment correlates with a decrease in sequence diversity over 8 rounds of selection with beads for all replicate populations (p-value = 0.0005, linear regression, table S1) (Fig. 3B). The majority of the sequences present in the pools were of the expected length throughout the selection process (~76nt) (Fig. 1def), with a slight increase in the diversity of sequence lengths in round 8 (p-value = 0.0091, linear regression, table S2) (Fig. 3C). Upon searching the entire sequenced pools (~30 million reads each) for the presence of the final 20 most common sequences, we confirmed that each sequence was initially present (frequency < 0.0001% in round 1) and persisted throughout the selection cycles, suggesting that these sequences rose in frequency from rare rather than were generated during the amplification process. Consistent increase in frequency also indicates the effects of selection, as opposed to drift, which was further verified through the positive selection coefficients (s) of the final common sequences (s = 1.836 ± 0.1252). As these sequences rose in frequency, GC- content likewise increased across replicates (p-value = 0.0128, linear regression, table S3), which has been shown to generate complex structures important for binding, particularly binding to streptavidin (55). In addition to adaptation observed within populations of individual sequences, motifs between 6 and 40 nt rose in frequency during selection (Fig. 3E; Fig. S1). These motifs include multiple regions with common sequences for streptavidin binders (ACGC or CGCA) (56-57), providing further evidence for the evolution of greater binding ability. 79 Altogether, these results suggest that selection with magnetic beads as binding targets yielded a standard adaptive response: the differential propagation of better-binding sequences. Fig. 1. Adaptation during static selection with beads and cells. The frequency of the 20 most common sequences in round 8 of selection are shown following selection rounds 1, 4, 7, and 8 with magnetic beads (A, B, C) or yeast cells (G, H, I) (3 replicates each; see Fig. 80 S2 for the legend with each sequence). Populations become enriched with these sequences during selection with beads but not cells (small boxes indicate the frequencies of the final most common sequences that are otherwise unidentifiable on the larger plot). Length diversity is depicted during selection with beads (D, E, F) and cells (J, K, L), where the height of each band indicates the frequency of sequences whose lengths fall within a 5-nt range. Populations selected with cells exhibit rapid diversification of sequence length, with sequences of various lengths longer than the expected range (> 76nt) increasing in frequency over time. Adaptive outcomes depend on the selection substrate. (c) Selection with cells promotes sequence diversification Unlike in vitro selection with beads, selection with cells did not lead to the enrichment of particular sequences (Fig. 1, G to I). The frequency of individual sequences did not increase significantly over rounds of selection (Fig. 1, G to I), nor did sequence diversity decrease with time (p-value = 0.3369, linear regression, table S4) (Fig. 3B). Instead, iterative cycles of selection led to a significant increase in the diversity of sequence lengths (p-value = 0.0001, linear regression, table S5) (Fig. 3C), particularly for sequences longer than the starting library (> 76 nt) (Fig. 1, J to L). As a result, the majority of the pools in round 8 of selection was comprised of sequences outside this expected range (71 – 80 nt) (Fig. 1, J to L). The selection coefficients for these non-target length sequences were positive across replicates (s = 0.520 ± 0.111), suggesting that longer length confers higher fitness in this selection regime. However, increased sequence length diversity could either result from the proliferation of preexisting sequences of varying lengths, or from diversification and therefore generation of new sequences. To resolve this, we found that the most common long sequences in the final population were not present in previous rounds of selection (frequency of 0 out of ~30 million sequences). Moreover, the vast majority of the long sequences present in round 8 of selection were found at extremely low final frequencies (1 or 2 out of the 100,000 sequences analyzed). In other words, the 81 majority of the final pool was made up of long, rare sequences. This suggests that selection with cells promoted diversification, particularly a diversification process yielding longer sequences. Assessment of motif data likewise diverged from selection with beads; almost all of the identified motifs in round 8 of selection consisted of primer binding regions (Fig. 3E), as opposed to motifs from the random 40-mer. This indicates the repetition of primer binding sites within the generated long sequences and suggests a mode of replication previously observed during in vitro selection (58-60), which we discuss below. Prior to the proliferation of these long sequences, the populations appear to adapt as expected, with motifs from the random region increasing in frequency between rounds 4 and 7 of selection (Fig. S1). Their decreased frequency in round 8 indicates a lower competitive ability compared to the generated long sequences. Overall, the nature of the response to selection with cells differed qualitatively from selection with beads; rather than an increasing frequency of particular sequences, the process of diversification favored the increased frequency of novel, long sequences. 82 Fig. 2. Adaptation during variable selection with beads and cells. The frequency of the 20 most common sequences in round 8 of selection are shown following 4 rounds of selection with beads before switching to cells (A, B, C) and vice versa for populations selected with cells before beads (G, H, I) (3 replicates each; see Fig. S3 for the legend with each sequence). All sequences remain rare throughout selection cycles. Diversity of sequence lengths diverges based on variable selection regime, where populations selected with beads first vary across replicates in their degree of diversification (D, E, F), and populations selected with cells first have high levels of diversification across replicates (J, 83 K, L). This process generates a variety of long sequences whose frequencies are depicted as the height of each band covering a 5-nt range. Results indicate that variable selection maintains and generates sequence diversity, and selection history constrains how stochasticity drives adaptive outcomes. (d) Selection history constrains adaptation in variable environments To explore adaptation in variable environments, samples from round 4 of selection with beads or cells were split; aliquots from each either continued selection with the same target or initiated lineages selected with the other target for the remaining four rounds (experimental design; Fig. 3A). We found that in all samples with alternating selection, populations did not become enriched with particular sequences, nor did sequence diversity change significantly over time (beads to cells: p-value = 0.5197; cells to beads: p-value = 0.1576, linear regression, table S6 and S7), akin to selection with cells alone. As seen with static selection, sequence length diversity also increased with rounds of selection (beads to cells: p-value = 0.0278; cells to beads: p-value = 0.0001, linear regression, table S8 and S9) (Fig. 3C). However, for populations selected with beads before switching to cells, evolutionary outcomes for length diversity diverged across replicates. One replicate population experienced high amounts of length diversification, where only 13.45% of the final population was in the expected length range (71 – 80nt) (Fig. 2F), whereas another replicate exhibited very little change in length diversity – the majority of the population (86.69%) retaining original lengths (Fig. 2E). By contrast, samples selected with cells before beads exhibited a high level of parallelism across replicates. All populations in this regime experienced a significant increase in length diversity over time (Fig. 2, J to L), to an even greater extent than what was observed with cells alone (> 82.99% of sequences longer than 80nt in the final populations). Comparing across all treatments (static and alternating), length diversity changed over time (p-value < 0.0001) and differed by 84 selection treatment (p-value = 0.0026), with the effects of treatment changing over time (p- value =0.0056, ANCOVA, table S10). Motif discovery for both alternating regimes yielded different results than the static treatments. Indeed, the motifs that rose in frequency during variable selection included a mixture of the two types observed with beads or cells alone: motifs from either the random 40-mer or primer binding sites (Fig. 3E; Fig. S1). These results indicate that while populations appear relatively unchanged between rounds 1 and 4 of selection when considering several other metrics (sequence enrichment, diversity, or length), selection during these rounds clearly constrains evolutionary outcomes, and the resulting motifs retain a signature of past selection. This constraint via selection history is specific to the target substrate; beginning with beads before switching to cells affects motif and length diversity differently than the converse. Results from variable selection also indicate that switching substrates may constitute a relaxed selection pressure, whereby the initially most common sequences persist as comparatively high sequences by round 8 (Fig. 2, A and B). In these cases, the sequence with the highest frequency in round 1 is one particular sequence, along with its reverse complement (TGACACCGTACCTGCTCTACGTGCTAGCGTGATCTGAGTGTATGACGCATCAG CGTCTAAGCACGCCAAGGGACTAT). Remarkably, this sequence is consistently among the highest-frequency sequences in rounds 1 and 4 for all populations in all treatments (including beads and cells alone), despite being below detection in the starting library. This “generalist” sequence rises in frequency during initial rounds and then declines as the populations either become enriched or diversify (Fig. 3D). 85 Fig. 3. Characterization of derived populations. Experimental design for static and alternating selection is shown (A) alongside Shannon Diversity Indices (H’) for sequence diversity (B) and length diversity (C) over 8 rounds of selection for all treatments, including the random starting library and a “control” population comprised of a single sequence prior to PCR amplification. Sequence diversity decreases significantly over time during selection with beads (p-value = 0.0005, linear regression), whereas length diversity increases for all populations (p-value < 0.0001), differing by treatment (p-value < 0.0001, ANCOVA). During these adaptive responses, the frequency of a “generalist” sequence (named for its uniquely high initial frequency across treatments) rises and then falls in frequency (D). The frequency of sequence motifs (6-40nt) in the final round of selection are shown for representative populations selected with beads, cells, or both (E). Selection with beads led to a rise in frequency of motifs from the random 40nt region (light), whereas motifs following selection with cells are primarily from the primer binding sites (dark). Alternating selection regimes yield motifs from both regions of the sequence, retaining the effects of past selection (see Fig. S1 for each round of selection). Overall, results demonstrate that the selective context shaped innovative outcomes. 86 (e) Diversification generates novel phenotypes As discussed above, diversification during selection with cells and varying substrates led to the generation of long sequences. Since function is linked to secondary structure formation (12), we explored how the generation of these sequences facilitated the emergence of novel phenotypic features. For example, the higher diversity of sequence lengths in round 8 of selection with cells corresponds with a greater diversity of melting temperatures, compared to selection with beads (Fig. S4). Selection with cells also leads to a greater number of loops and multi-branch loops in the secondary structures of sequences between initial and final rounds of selection (total loops: p-value < 0.0001; multi-branch loops: p-value < 0.0001, linear regression, table S11 and S12), while there is no significant difference in secondary features following selection with beads (total loops: p-value = 0.6188; multi-branch loops: p-value = 0.2271, linear regression, table S13 and S14). This is unsurprising given that more loops are likely a byproduct of longer sequence length. However, when sequences of the same length and GC content were randomly generated and compared to long sequences that emerged during selection with cells, we found that there were a greater number of loops in the secondary structures of derived sequences, compared to those randomly generated (p-value = 0.0003, one-way ANOVA, table S15) (Fig. S5), indicating that the types of structures formed are shaped by their selection history. Combined with the motif data which suggested that these new sequences contain repetitions of primer binding sites, these results demonstrate that the diversification process during selection with cells led to the generation of novel phenotypes (Fig. 4). 87 Fig. 4. Innovation emerges during selection with cells. Three sequences which are representative of the majority of the derived sequences are shown, following static selection with beads or cells. The structure of the starting library – a random 40-mer flanked by primer binding sites (highlighted) – is largely preserved during selection with beads. The tertiary structure of each sequence is shown to its right. Derived sequences selected with cells, however, are remarkably different from the ancestral library. Most sequences are longer, with repetitions of both primer binding sites, interspersed with random regions. Tertiary structures are consequently transformed, generating a diversity of complex conformations. This shift in complexity fundamentally alters adaptive possibilities. 88 III. Discussion A unifying requisite among all hypotheses for the origins of life is the evolution of novelty. The leap in complexity from a prebiotic environment to cells ensures this agreement, irrespective of contention over which geochemical conditions favored the transition. Here, we explored the selective contexts that may favor the emergence of innovation, using an empirical model for pre-cellular evolution: in vitro selection with ssDNA. We demonstrate how a selection regime with reduced stringency (i.e. cells as binding substrates) and variable selection (i.e. switching substrates) allow populations to circumnavigate evolutionary plateaus. Maintenance of genetic diversity facilitates the emergence of novel genotypes, transforming possibilities for further adaptation. These results offer broad insights into several possible prebiotic scenarios, such as during the emergence of cooperation among replicating polymers (61-62), the attachment of these polymers to nearby surfaces or vesicles (48-49, 63), or variable environments with dual selection pressures for replication (which often favors shorter sequence length) and other functions (which may favor longer length, e.g. catalysis) (50, 64-65). Our results provide implications for how selection and diversity can generate open-ended outcomes in these possible scenarios and suggest avenues for further investigation. In vitro mechanism of diversification Previous research using SELEX typically culminates in a standard adaptive response, observed here during selection with streptavidin beads: complex DNA libraries become enriched with high-affinity sequences (55). During previous work, researchers discovered another tendency for pools to rapidly diversify via by-product formation, leading to a smear visible on an agarose gel above the product band (58). This process occurs during the amplification of large libraries of random DNA sequences, where the bases in a random region can be complementary to another single-stranded sequence and serve as a primer that is extended by the polymerase to produce longer dsDNA products 89 (59). Hence, while PCR amplification of homogenous DNA templates typically risks primer-primer hybridization and the formation of short by-products, amplification of complex DNA libraries provides the potential for product-product hybridization and a suite of long by-products (60). This outcome reflects the multiplicity of selection pressures in the system, including selection not only for binding ability but also the capacity to reproduce during amplification. Since the goal of most in vitro selection studies has been to identify aptamers with high binding affinity and specificity (rather than reproductive efficiency), by-products were viewed as “parasites” and detrimental to the selection process (59,66). To prevent by-product accumulation, researchers can increase the stringency of selection during subsequent cycles – a balance between avoiding loss of rare, functional sequences and risking by-product formation via maintained diversity (67). Another strategy has been the use of emulsion PCR, where amplification is conducted simultaneously in oil droplets containing a small fraction of sequences to reduce the likelihood of interactions among partially complementary sequences (58, 67). In this study, the process of by-product formation constitutes an evolutionary response of rapid complexification. We find evidence for both types of by-products previously described (ladder and non-ladder like) during selection with cells, particularly variable selection that begins with cells (67). Populations selected with beads, by contrast, function as our controls for standard adaptation. Interestingly, by-product formation shares many similarities with recombination during sexual reproduction, where novel genetic combinations are generated from different genotypes. Since adaptation during SELEX typically proceeds by enriching sequences that are already present, SELEX without recombination can be limited in its ability to explore the sequence space of a given nucleic acid, even with error-prone PCR introducing point mutations at a faster rate (68). As a result, recombination during in vitro selection can be important for generating novel functions and removing detrimental genotypes and has been demonstrated to aid the discovery of global optima (69). In our study, a recombination-like process was not imposed at the outset but rather emerged as an evolutionary mechanism. We find that 90 certain contexts are more favorable for this process to occur, with the maintenance of sequence diversity likely being a key factor in facilitating the generation of long sequences. The resulting complexification process found here includes features that are hallmarks of major transitions in complexity (1), a) cooperation among individuals which results in, b) a modified replication process. Novel and unexpected functions have emerged previously during in vitro selection (19, 60, 66-67, 70-71), but the resulting structures are usually either, a) considered undesired and therefore suppressed, b) acknowledged but not studied in detail, or c) synthesized and analyzed in isolation. The first two points illustrate the nature of most SELEX studies to shuffle populations towards a desired goal, while the third point, the analysis of isolated sequences, has been informative for determining individual function but is limited by the same challenges faced in community ecology and evolution: the function of individual species depends on its interactions within the community (72-74). Therefore, functional analysis of isolated species can misrepresent their ecological role, and the significance of community diversity – and the interactions therein – may be missed. Diversity begets complexity The complexification process described above results from the diversity of the DNA library. The longer this initial diversity is maintained – where every sequence is rare and unique – the greater the likelihood that partially complementary sequences will interact during amplification (67). Even during emulsion PCR, the higher the number of different templates present in a compartment, the greater the formation of by-products (58). Here, we see clear divergence in evolutionary outcomes based on selection substrate, consistent with previous findings that the target plays an important role in by-product accumulation (59). Since cellular surfaces are far more heterogeneous than magnetic beads, we hypothesized that cells provide more opportunities for adaptation and the persistence of a 91 variety of sequences. As a result, selection with cells is likely less stringent, allowing the maintenance of more diversity during early rounds of selection. This initial effect on population diversity is not detectable during rounds 1 and 4 of selection, given inherent constraints in sequencing resolution. Since the nature of the random library maximizes sequence diversity, the total number of different sequences in the entire pool necessarily decreases as unbound sequences are discarded with each round of selection, regardless of binding target, but this change does not register in the diversity metrics of only a subsample. However, we see evidence for the impact of early selection with cells versus beads when we examine the populations selected in variable environments. When populations begin with cells before beads, rapid diversification occurs in final rounds of selection, the degree of which is highly parallel across replicates (Fig. 2, J to L). The proliferation of these generated sequences is even greater than selection with cells alone, which is likely owed to the adaptation of cell-specific sequences during 8 rounds of selection with cells. We see evidence for the rise in frequency of particular motifs prior to diversification during selection with cells (Fig. S1), whose competitive ability may have slowed the takeover of by-products, compared to populations that switched to beads. By contrast, populations selected with beads before cells diverge across replicates in their degree of diversification. This may be due to the early loss of diversity during the first 4 rounds with beads, which decreased the probability for sequence-sequence hybridization upon switching to cells, allowing stochasticity to drive outcomes. The proliferation of this new process of replication opens a niche for long sequences. While previously there was a tradeoff between sequence length and amplification efficiency, long by-products are rapidly able to outcompete original sequences due to their repetition of primer-binding sites (67). An excellent indicator of these shifting competitive dynamics is the rise and fall in frequency of the “generalist” sequence, which decreases in frequency following either diversification or enrichment of 92 the populations (Fig. 3D). The opening of this niche for a variety of long, rare sequences ultimately transforms the sequence pool, providing greater phenotypic complexity for further selection. We believe there are three possible explanations for why by-product formation is more prevalent during selection with cells versus beads, each of which is an exciting possibility that may be explored in future work. The first explanation is that the ability to bind to the surface of cells is shared among almost all of the sequences; selection for binding ability is therefore weak, and outcomes depend on drift during the binding phase and selection during replication. While possible, we do not find this explanation likely, given the abundance of past work demonstrating differential binding ability during in vitro selection with cells (75), as well as our own evidence indicating a response to selection for binding prior to diversification. Second, there may be selection for binding to cells, but their heterogeneous surface provides many ecological niches where the sequences can bind. In this case, a rugged adaptive landscape may favor diversification, where negative frequency-dependent selection maintains diversity. Our results are consistent with this explanation, yet we cannot distinguish between this possibility and the third, which is that selection during the binding phase with cells not only helps retain diversity but also favors long, complex sequences. If this is the case, these long sequences are generated during amplification and then maintained both by this continued process and differential binding. While we do not provide evidence that these long sequences are indeed better at binding than those of original length, it is evident that they still can bind, given this is necessary to persist through multiple wash steps. In either case, it is clear that in this selective context (including a binding and replication phase), these long sequences have a strong advantage. Reiterations of primer binding sites promote a novel replication process, where individual sequences forgo vertical transmission of their genotype in favor of recombination, enabling these “species” to overtake the population while retaining binding function. 93 Open-ended evolution at the origins of life The ability to continuously evolve new forms has long been recognized as vital to the origins of life (76-77). This concept of open-endedness often describes the emergence of novel functions, avoiding evolutionary plateaus at some ‘optimal’ or stable position (78). While many studies have explored prebiotic evolvability, investigating a system capable of open-ended evolution has been a greater challenge (76). In this study, we find evidence for both stable and open-ended evolution. During selection with beads, populations appear to evolve toward a plateau, where sequences of similar length and structural complexity sweep through the populations. Eventually, this process might culminate in the dominance of only a few, high-fitness sequences. Alternatively, populations can circumvent this plateau via the generation of novelty, as seen during selection with cells. In these cases, new genotypes and phenotypes emerge. Sequences become much longer with unique iterations of primer binding sites, their secondary structures significantly more complex. Since their internal sequence impacts further hybridization – and their three-dimensional phenotypes determine binding ability – the evolutionary potential of these pools has been transformed. This outcome is indicative of open-ended evolution, whether defined as an increase in internal complexity (79) or the occupation of ever more diverse niches (80). We envision this adaptive pathway as a three-step process (Fig. 5): 𝑋 + ⇉ -.- where there is preexisting diversity, X, on which natural selection, Y, has the opportunity to act, resulting in an adaptive response to selection, denoted as Z or Z’, where Z’ is open- ended and Z is not. The key distinction among these outcomes is that Z results in a period of evolutionary stasis (i.e. the population reaches a local optima) until some external phenomena introduces change, such as an environmental perturbance or the influx of variation via dispersal or mutation. An open-ended response, Z’, however, enables a 94 feedback to reinstate the entire process, introducing novelty on which selection has new opportunities to act. We find that whether evolution is open-ended depends on the selective context. When the stringency of selection is low (also referred to as “soft” competition, i.e. there are more possibilities to persist each cycle), there is a lower temporal constraint on how quickly novelty needs to emerge. This also provides more opportunities for neutral processes to influence outcomes, allowing the exploration of phenotypic space (28). As diversity is maintained, there are more possibilities for interactions among individuals which may deform the context itself. In this study, the maintenance of diversity over time increased the likelihood of chance encounters among individuals, allowing them to interact to generate new forms. Interactions in this case were cooperative, akin to sexual reproduction, where different sequences hybridized to create new genotypes. We provided relaxed selective conditions in three ways: 1) Incubation time remained constant over selection cycles, to allow the persistence of less-fit sequences, 2) Cells as targets provided opportunities for a diverse set of sequences to bind to their heterogeneous surfaces, and 3) Variable selection with two different substrates decreased the likelihood that high-fitness specialists could dominate the pools. By comparing across treatments, we can see that each of these factors influenced the emergence of novelty. The resulting open-ended evolution created a feedback on the population’s diversity by generating unique individuals whose genotypes afforded a dramatic increase in fitness compared to ancestors. 95 Fig. 5. Alternative adaptive pathways during in vitro selection. Populations begin with extremely high sequence diversity in the random library, X, on which selection, Y, has the opportunity to act. Selection with beads results in an adaptive response, Z, which decreases sequence diversity. Selection then operates on remaining diversity, which further homogenizes the populations. Alternatively, selection with cells leads to an open-ended adaptive response, Z’, where diversity is maintained, and novel sequences are generated. These recombinant sequences provide new opportunities for adaptation, which results in the proliferation of even longer sequences whose tertiary structures are more complex and more variable among individuals, compared to the initial library. This open-ended response is conceptualized as a feedback among diversity and adaptation, where new possibilities for adaptation continue to arise via the generation of novelty within diverse populations. IV. Conclusion The evolution of prebiotic complexity is the core of the origins of life. Previous insights from ecology and evolutionary biology indicate which selective contexts might promote the emergence of novelty. Here, we bring these two lines of work together, providing an empirical demonstration of how eco-evolutionary dynamics can facilitate 96 innovation in a non-cellular system. We illustrate how certain selective contexts promote open-ended evolution, which itself is a feedback for further adaptation. Included in these contexts are a preexisting source of variation, selection which allows the persistence of diversity, and consequently the opportunity for ecological interactions to generate novel phenotypes. These findings, while shown here in a particular system, are translatable to a variety of prebiotic scenarios or hypotheses on the origins of life. By controlling selective contexts, rather than predetermining particular outcomes, experiments remain open to the possibility that evolution can produce unanticipated effects. V. Materials and Methods (a) SELEX procedure The SELEX procedure was performed using the XELEX DNA Core Kit (E3650) from EURx, based on previous work (81-82). The DNA library consists of 40 random nucleotides (universal library size: 1013 to 1015 different oligonucleotides), with both the 5’- and 3’-ends of the random 40mer flanked by 18-base pair defined priming sites for PCR amplification (5’-TGA CAC CGT ACC TGC TCT - 40nt randomized sequence - AAG CAC GCC AGG GAC TAT-3’). The forward primer sequence is 5’-TGA CAC CGT ACC TGC TCT-3’, and the reverse primer is 5’-ATA GTC CCT GGC GTG CTT-3’. The process of in vitro selection was carried out with two binding targets, separately. Streptavidin Magnetic Beads (101228-120, NEB) (1 µm) were chosen for their common use during counter-selection and therefore demonstrated ability to yield high affinity aptamers (56). For our purposes, streptavidin-coated beads served as a simpler alternative to cellular surfaces. The other selection target was the single-celled yeast species Kluyveromyces lactis (strain NRLY-1140), which during pilot testing appeared to yield higher DNA concentrations following PCR amplification, compared to other cultures 97 of yeast (Saccharomyces cerevisiae) or bacteria (Escherichia coli), monitored via agarose gel. Target cells functioned as heterogenous surfaces during selection. At the beginning of each round of selection, the selection target (0.5mg of beads or 50 μL of grown yeast culture) was washed twice with 500 μL of 0.8% saline solution and suspended in 500 μL of 1x SELEX buffer (SELEX buffer composition: 140 mM NaCl, 2 mM KCl, 5 mM MgCl2, 2 mM CaCl2, 20 mM Tris pH 7.4, 0.05% [v/v] Tween 20). This was combined with 40 μL of the ssDNA library (0.02 nmol/μL) and incubated at 25°C for 1 hour. Following incubation, beads were partitioned with a magnet and cells via centrifugation at 0.1 g for 3 min. The supernatant was removed and washed with 500 μL of 1x SELEX buffer to discard unbound sequences. Bound DNA was eluted from the beads or cells through denaturation at 94°C for 3 minutes in 50 μL of nuclease-free water and recovered for PCR amplification or storage at -20°C. The polymerase chain reactions were carried out in standard mixtures of 16 μL of DreamTaq Master Mix (FERK1071), 2 μL 5’-primer, 2 μL of 3’-primer, 10 μL DNA template, and 20 μL nuclease-free water. The PCR thermal cycle settings were: 95°C for 2 minutes followed by 30 cycles of 95°C for 30 seconds, 55°C for 30 seconds, and 72°C for 30 seconds, with the final extension at 72°C for 5 minutes. Prior to subsequent rounds of selection, amplified dsDNA pools were then combined with 500 μL of 1x SELEX buffer, denatured at 94°C for 3 minutes to allow for secondary structure formation of ssDNA, and immediately placed on ice. In order to provide opportunities for sequence enrichment, the selection cycle (denaturation of dsDNA, incubation with target, partitioning of unbound sequences, elution, and amplification) was repeated 8 times. Selection progress was monitored on a 3 % [w/v] agarose gel. (b) Experimental design To incorporate static and variable selection, 4 rounds of selection were completed with either beads or cells (3 replicates each), after which the pools were split, and aliquots of 20 μL from each amplified pool were used to either continue selection with the same 98 substrate or switch to the other (3 replicates per treatment) (see Fig. 1a). To observe evolutionary responses throughout the selection process, all replicates from the beginning (round 1), middle (round 4), and end (round 8) were submitted for sequencing, as well as round 7, to increase resolution near the end of the experiment where enrichment is found to rapidly occur (57). We also sequenced the starting 40N library and a control library of a single sequence flanked by the same primer binding sites, which provided a lower estimate for diversity, as well as verification of sequencing results. (c) DNA pool generation for Illumina sequencing and assembly Following 8 rounds of selection and amplification via PCR, all samples from rounds 1, 4, 7, and 8 of selection were purified using the purification protocol from the Micellula DNA Emulsion & Purification Kit (E3600): The amplified pools were transferred into a DNA spin column prepped with 30 µL of activation Buffer DX and spun down in a micro-centrifuge at 12,000 rpm for 1 min. The flow-through was discarded, and 500 µL of Wash DX1 buffer was added to the spin column, followed by another 1 min. at 12,000 rpm and discard of flow-through. The sample was washed again with 600 µL of Wash DX2 buffer and spun at 12,000 rpm for 1 min. After discarding flow-through, any traces of Wash DX buffer was removed following another 12,000 rpm for 2 min. The spin- column was placed into a new receiver tube, and 50 µL of Elution buffer (heated to 80°C) (elution buffer prepared using ultra-pure water with trace buffering compounds) was added to the column and incubated for 2 minutes at room temperature to elute bound DNA. Following another 1 min. at 12,000 in the micro-centrifuge, the receiver tube was capped and cooled on ice prior to sequencing submission. Independent TruSeq Nano DNA sequencing libraries were created from the 38 submitted amplicons. All libraries were combined and sequenced on a NovaSeq S4 2x150- bp lane and spiked in a target of 20% PhiX to introduce base diversity and improve base calling quality. Quality check for the raw data was carried out using FastQC v0.11.9 tool (83). Depending on the quality metrics of the data, the parameters were adjusted to remove 99 indexed adapters and dimers. After normalizing the number of reads to 100,000 pair-end reads per sample, the assignments were followed by the merging of pair ends using PEAR (84), followed by de novo assembly of each paired read separately using SPAdes v3.15.5 (85). Unless otherwise stated, analyses pertain to these pools of 100,000 pair-end reads. (d) Analysis of sequence frequency Frequencies of each sequence were measured using a Perl CountSeqs.l script (see Supplemental Materials). Individual sequences were sorted by frequency, and the 20 most frequent individuals in round 8 of selection were recorded and monitored over each selection round. Sequence length was calculated for each individual, and the frequency of individuals within each 5-nt range between 0 and 300 nt was recorded. To determine whether the common sequences in the final round of selection were preexisting versus generated during the selection process, one replicate population from each static selection regime (beads or cells only) was chosen for in-depth analysis of total reads. Contigs from the entire pool (~30 million per sample) were assembled using SPAdes v3.15.5 (85) (same methods as above), and the frequency of the final most common sequences were determined for each sequenced selection round. To determine whether adaptation occurred beyond the change in sequence or length frequency, motif discovery was performed for the DNA datasets. Contigs were listed using the EditFastas.l script (see Supplemental Materials) and submitted to MEME-suite 5.5.2 (86). Using the -classic mode on the MEME tool, we searched for 50 motifs between 6 and 40 nt in length (E-value < 0.05) (87). The 2 most frequent motifs for each sample were verified as priming sites and removed from analysis. (e) Library diversity analysis To assess changes in sequence diversity across selection cycles, the Shannon Diversity Index (H) was calculated for all populations, based on standard ecological practices (88): 100 𝐻 = −∑𝑝/ln (𝑝/), where pi represents the proportion of the population made up of sequence i. Since the frequency of individuals of different lengths also appeared to change over time, the Shannon Diversity Index was calculated for length diversity using the above equation, where pi represents the proportion of the population made up of individuals of a certain sequence length (5-nt range, from 0 to 300). Both diversity indices were calculated for all 3 replicate populations in each treatment for rounds 1, 4, 7, and 8 of selection, as well as the random starting library and control library of a single sequence. (f) Selection coefficients To estimate the rate of adaptation, we calculated selection coefficients for populations selected with beads versus cells. The most frequent sequences present after 8 rounds of selection with beads were found to be extremely rare in the first round, hence we used the pools with total contigs assembled (31,588,457 contigs in round 1) to assess the frequency of the sequences over time. Specifically, the frequencies of the 20 most common sequences in round 8 of selection with beads were measured for each round. The change in the relative proportion of a particular sequence was measured as the differential growth rate (s) (89): 𝑠 = ln %!!(#) !"(#) &, where x1(t) and x2(t) represent the relative frequency of one sequence and all other sequences, respectively, by timepoint (t). We estimated the selection coefficient for each sequence as the slope from the linear regression of differential growth rate over time. 101 For populations selected with cells, adaptation appeared to occur via diversification of sequence length, hence selection coefficients were calculated to estimate the change in fitness of sequence length over time. Since the starting library consists of a 40-nt region flanked by two 18-nt primer binding sites, the expected sequence length is 76nt. Accounting for slight deviations from this expectation, selection coefficients for the 3 replicate populations selected with cells were determined using the same methodology as above, except where x2(t) represents the sequences in this expected range (71 – 80 nt), and x1(t) is the frequency of all other lengths. (g) Assessment of derived DNA properties The function of a DNA molecule is specific to its sequence of bases. As particular sequences rise in frequency, sequence properties, such as GC-content and secondary structure formation, might indicate their higher binding affinity. To compare adaptation among populations selected with beads versus cells, we determined the GC-content for the 20 most frequent sequences in the starting library and the first and final rounds of selection with beads or cells, using a GC Content Calculator (90). Secondary structures were generated for the same populations via a Secondary Structure Predictor tool (VectorBuilder), and total number of loops and multi-branch loops were recorded. To determine whether change in number of loops was simply a byproduct of longer sequences, we randomly selected 20 sequences from round 8 of selection with cells and compared their secondary structures to 20 sequences that were randomly generated with the same length and GC content (Random DNA Sequence Generator; Maduro Lab). To better visualize DNA conformations, tertiary structures were predicted using a 3dRNA/DNA tool (91). For confirmation of a correlation between sequence length and melting temperatures (Tm), length and Tm were calculated for the 100 most frequent sequences selected with beads or cells, using a Tm calculator (modified from 92). All statistical analyses were carried out using JMP Pro 16.0.0. 102 References 1. Szathmáry, E., & Smith, J. M. (1995). The major evolutionary transitions. Nature, 374(6519), 227-232. 2. Baum, D. A., Peng, Z., Dolson, E., Smith, E., Plum, A. M., & Gagrani, P. (2023). The ecology–evolution continuum and the origin of life. Journal of the Royal Society Interface, 20(208), 20230346. 3. Sharma, A., Czégel, D., Lachmann, M., Kempes, C. P., Walker, S. I., & Cronin, L. (2023). Assembly theory explains and quantifies selection and evolution. Nature, 622(7982), 321-328. 4. Segré, D., Shenhav, B., Kafri, R., & Lancet, D. (2001). The molecular roots of compositional inheritance. Journal of Theoretical Biology, 213(3), 481-491. 5. Bhowmik, S., & Krishnamurthy, R. (2019). The role of sugar-backbone heterogeneity and chimeras in the simultaneous emergence of RNA and DNA. Nature chemistry, 11(11), 1009-1018. 6. Sokolskyi, T., Ganju, P., Montgomery-Taylor, R., & Baum, D. A. (2024). Evidence of Heritability in Prebiotically Realistic Membrane-Bound Systems. Life, 14(3), 284. 7. Adamala, K., & Szostak, J. W. (2013). Competition between model protocells driven by an encapsulated catalyst. Nature chemistry, 5(6), 495-501. 8. Mizuuchi, R., & Ichihashi, N. (2021). Primitive compartmentalization for the sustainable replication of genetic molecules. Life, 11(3), 191. 9. Lazcano, A., & Miller, S. L. (1999). On the origin of metabolic pathways. Journal of Molecular Evolution, 49, 424-431. 10. Muchowska, K. B., Varma, S. J., & Moran, J. (2019). Synthesis and breakdown of universal metabolic precursors promoted by iron. Nature, 569(7754), 104-107. 103 11. Tuerk, C., & Gold, L. (1990). Systematic evolution of ligands by exponential enrichment: RNA ligands to bacteriophage T4 DNA polymerase. science, 249(4968), 505-510. 12. Wilson, D. S., & Szostak, J. W. (1999). In vitro selection of functional nucleic acids. Annual review of biochemistry, 68(1), 611-647. 13. Robertson, D. L., & Joyce, G. F. (1990). Selection in vitro of an RNA enzyme that specifically cleaves single-stranded DNA. Nature, 344(6265), 467-468. 14. Darmostuk, M., Rimpelova, S., Gbelcova, H., & Ruml, T. (2015). Current approaches in SELEX: An update to aptamer selection technology. Biotechnology advances, 33(6), 1141-1161. 15. Burke, D. H., & Gold, L. (1997). RNA aptamers to the adenosine moiety of S-adenosyl methionine: structural inferences from variations on a theme and the reproducibility of SELEX. Nucleic acids research, 25(10), 2020-2024. 16. Walton, T., DasGupta, S., Duzdevich, D., Oh, S. S., & Szostak, J. W. (2020). In vitro selection of ribozyme ligases that use prebiotically plausible 2-aminoimidazole– activated substrates. Proceedings of the National Academy of Sciences, 117(11), 5741- 5748. 17. Akoopie, A., Arriola, J. T., Magde, D., & Müller, U. F. (2021). A GTP-synthesizing ribozyme selected by metabolic coupling to an RNA polymerase ribozyme. Science Advances, 7(41), eabj7487. 18. Horning, D. P., & Joyce, G. F. (2016). Amplification of RNA by an RNA polymerase ribozyme. Proceedings of the National Academy of Sciences, 113(35), 9786-9791. 19. Portillo, X., Huang, Y. T., Breaker, R. R., Horning, D. P., & Joyce, G. F. (2021). Witnessing the structural evolution of an RNA enzyme. Elife, 10, e71557. 104 20. Zaman, L., Meyer, J. R., Devangam, S., Bryson, D. M., Lenski, R. E., & Ofria, C. (2014). Coevolution drives the emergence of complex traits and promotes evolvability. PLoS Biology, 12(12), e1002023. 21. Lynch, M. (2007). The frailty of adaptive hypotheses for the origins of organismal complexity. Proceedings of the National Academy of Sciences, 104(suppl_1), 8597- 8604. 22. Bajić, D., Rebolleda-Gómez, M., Muñoz, M. M., & Sánchez, Á. (2021). The macroevolutionary consequences of niche construction in microbial metabolism. Frontiers in Microbiology, 12, 718082. 23. Steinberg, B., & Ostermeier, M. (2016). Environmental changes bridge evolutionary valleys. Science advances, 2(1), e1500921. 24. Kalambokidis, M., & Travisano, M. (2024). The eco-evolutionary origins of life. Evolution, qpad195. 25. Chen, P., & Kassen, R. (2020). The evolution and fate of diversity under hard and soft selection. Proceedings of the Royal Society B, 287(1934), 20201111. 26. Blount, Z. D., Borland, C. Z., & Lenski, R. E. (2008). Historical contingency and the evolution of a key innovation in an experimental population of Escherichia coli. Proceedings of the National Academy of Sciences, 105(23), 7899-7906. 27. Burnham, T. C., & Travisano, M. (2021). The landscape of innovation in bacteria, battleships, and beyond. Proceedings of the National Academy of Sciences, 118(26), e2015565118. 28. Wright, S. (1931). Evolution in Mendelian populations. Genetics, 16(2), 97. 29. Baselga-Cervera, B., Gettle, N., & Travisano, M. (2022). Loss-of-heterozygosity facilitates a fitness valley crossing in experimentally evolved multicellular yeast. Proceedings of the Royal Society B, 289(1976), 20212722. 105 30. Kalambokidis, M., & Travisano, M. (2023). Multispecies interactions shape the transition to multicellularity. Proceedings of the Royal Society B, 290(2007), 20231055. 31. Baselga-Cervera, B., Jacobsen, K. A., Denison, R. F., & Travisano, M. (2023). Experimental evolution in the cyanobacterium Trichormus variabilis: increases in size and morphological diversity. Evolution; international journal of organic evolution, qpad037-qpad037. 32. Bajić, D., Vila, J. C., Blount, Z. D., & Sánchez, A. (2018). On the deformability of an empirical fitness landscape by microbial evolution. Proceedings of the National Academy of Sciences, 115(44), 11286-11291. 33. Laland, K. N., Odling-Smee, F. J., & Feldman, M. W. (1999). Evolutionary consequences of niche construction and their implications for ecology. Proceedings of the National Academy of Sciences, 96(18), 10242-10247. 34. Lande, R., & Shannon, S. (1996). The role of genetic variation in adaptation and population persistence in a changing environment. Evolution, 434-437. 35. Bell, G. A. (1997). Experimental evolution in Chlamydomonas. I. Short-term selection in uniform and diverse environments. Heredity, 78(5), 490-497. 36. Byers, D. L. (2005). Evolution in heterogeneous environments and the potential of maintenance of genetic variation in traits of adaptive significance. Genetica, 123, 107- 124. 37. Schick, A., & Kassen, R. (2018). Rapid diversification of Pseudomonas aeruginosa in cystic fibrosis lung-like conditions. Proceedings of the National Academy of Sciences, 115(42), 10714-10719. 38. Rainey, P. B., & Travisano, M. (1998). Adaptive radiation in a heterogeneous environment. Nature, 394(6688), 69-72. 106 39. Friesen, M. L., Saxer, G., Travisano, M., & Doebeli, M. (2004). Experimental evidence for sympatric ecological diversification due to frequency‐dependent competition in Escherichia coli. Evolution, 58(2), 245-260. 40. Driscoll, W. W., & Travisano, M. (2017). Synergistic cooperation promotes multicellular performance and unicellular free-rider persistence. Nature communications, 8(1), 15707. 41. Joyce, G. F. (2004). Directed evolution of nucleic acid enzymes. Annual review of biochemistry, 73(1), 791-836. 42. Darmostuk, M., Rimpelova, S., Gbelcova, H., & Ruml, T. (2015). Current approaches in SELEX: An update to aptamer selection technology. Biotechnology advances, 33(6), 1141-1161. 43. Hayden, E. J., Ferrada, E., & Wagner, A. (2011). Cryptic genetic variation promotes rapid evolutionary adaptation in an RNA enzyme. Nature, 474(7349), 92-95. 44. Voytek, S. B., & Joyce, G. F. (2009). Niche partitioning in the coevolution of 2 distinct RNA enzymes. Proceedings of the National Academy of Sciences, 106(19), 7780-7785. 45. Petrie, K. L., & Joyce, G. F. (2014). Limits of neutral drift: lessons from the in vitro evolution of two ribozymes. Journal of molecular evolution, 79, 75-90. 46. Amini, Z. N., & Müller, U. F. (2013). Low selection pressure aids the evolution of cooperative ribozyme mutations in cells. Journal of Biological Chemistry, 288(46), 33096-33106. 47. Lenski, R. E. (2017). Experimental evolution and the dynamics of adaptation and genome evolution in microbial populations. The ISME journal, 11(10), 2181-2194. 48. Wächtershäuser, G. (1988). Before enzymes and templates: theory of surface metabolism. Microbiological reviews, 52(4), 452-484. 107 49. Cleaves II, H. J., Scott, A. M., Hill, F. C., Leszczynski, J., Sahai, N., & Hazen, R. (2012). Mineral–organic interfacial processes: potential roles in the origins of life. Chemical Society Reviews, 41(16), 5502-5525. 50. Mizuuchi, R., Blokhuis, A., Vincent, L., Nghe, P., Lehman, N., & Baum, D. (2019). Mineral surfaces select for longer RNA molecules. Chemical communications, 55(14), 2090-2093. 51. Baum, D. A. (2015). Selection and the origin of cells. Bioscience, 65(7), 678-684. 52. Hol, F. J., Galajda, P., Nagy, K., Woolthuis, R. G., Dekker, C., & Keymer, J. E. (2013). Spatial structure facilitates cooperation in a social dilemma: empirical evidence from a bacterial community. PloS one, 8(10), e77042. 53. Ciccarese, D., Micali, G., Borer, B., Ruan, C., Or, D., & Johnson, D. R. (2022). Rare and localized events stabilize microbial community composition and patterns of spatial self-organization in a fluctuating environment. The ISME Journal, 16(5), 1453-1463. 54. Nadell, C. D., Drescher, K., & Foster, K. R. (2016). Spatial structure, cooperation and competition in biofilms. Nature Reviews Microbiology, 14(9), 589-600. 55. Mencin, N., Šmuc, T., Vraničar, M., Mavri, J., Hren, M., Galeša, K., ... & Šolar, B. (2014). Optimization of SELEX: Comparison of different methods for monitoring the progress of in vitro selection of aptamers. Journal of pharmaceutical and biomedical analysis, 91, 151-159. 56. Wang, C., Yang, G., Luo, Z., & Ding, H. (2009). In vitro selection of high-affinity DNA aptamers for streptavidin. Acta Biochim Biophys Sin, 41(4), 335-340. 57. Schütze, T., Wilhelm, B., Greiner, N., Braun, H., Peter, F., Mörl, M., ... & Glökler, J. (2011). Probing the SELEX process with next-generation sequencing. PloS one, 6(12), e29604. 108 58. Shao, K., Ding, W., Wang, F., Li, H., Ma, D., & Wang, H. (2011). Emulsion PCR: a high efficient way of PCR amplification of random DNA libraries in aptamer selection. PloS one, 6(9), e24910. 59. Tolle, F., Wilke, J., Wengel, J., & Mayer, G. (2014). By-product formation in repetitive PCR amplification of DNA libraries during SELEX. PloS one, 9(12), e114693. 60. Musheev, M. U., & Krylov, S. N. (2006). Selection of aptamers by systematic evolution of ligands by exponential enrichment: addressing the polymerase chain reaction issue. Analytica chimica acta, 564(1), 91-96. 61. Eigen, M., & Schuster, P. (1977). A principle of natural self-organization. Naturwissenschaften, 64(11), 541-565. 62. Vaidya, N., Manapat, M. L., Chen, I. A., Xulvi-Brunet, R., Hayden, E. J., & Lehman, N. (2012). Spontaneous network formation among cooperative RNA replicators. Nature, 491(7422), 72-77. 63. Briones, C., Stich, M., & Manrubia, S. C. (2009). The dawn of the RNA World: toward functional complexity through ligation of random RNA oligomers. Rna, 15(5), 743- 749. 64. Mills, D. R., Peterson, R. L., & Spiegelman, S. (1967). An extracellular Darwinian experiment with a self-duplicating nucleic acid molecule. Proceedings of the National Academy of Sciences, 58(1), 217-224. 65. Hayden, E. J., & Lehman, N. (2006). Self-assembly of a group I intron from inactive oligonucleotide fragments. Chemistry & biology, 13(8), 909-918. 66. Marshall, K. A., & Ellington, A. D. (1999). Molecular parasites that evolve longer genomes. Journal of molecular evolution, 49, 656-663. 109 67. Kohlberger, M., & Gadermaier, G. (2022). SELEX: Critical factors and optimization strategies for successful aptamer selection. Biotechnology and Applied Biochemistry, 69(5), 1771-1792. 68. Bittker, J. A., Le, B. V., & Liu, D. R. (2002). Nucleic acid evolution and minimization by nonhomologous random recombination. Nature biotechnology, 20(10), 1024-1029. 69. Lehman, N., & Unrau, P. J. (2005). Recombination during in vitro evolution. Journal of molecular evolution, 61, 245-252. 70. Lam, B. J., & Joyce, G. F. (2011). An isothermal system that couples ligand-dependent catalysis to ligand-independent exponential amplification. Journal of the American Chemical Society, 133(9), 3191-3197. 71. Lau, M. W., & Unrau, P. J. (2009). A promiscuous ribozyme promotes nucleotide synthesis in addition to ribose chemistry. Chemistry & biology, 16(8), 815-825. 72. Bier, R. L., Bernhardt, E. S., Boot, C. M., Graham, E. B., Hall, E. K., Lennon, J. T., ... & Wallenstein, M. D. (2015). Linking microbial community structure and microbial processes: an empirical and conceptual overview. FEMS microbiology ecology, 91(10), fiv113. 73. Strickland, M. S., Lauber, C., Fierer, N., & Bradford, M. A. (2009). Testing the functional significance of microbial community composition. Ecology, 90(2), 441-451. 74. Diaz-Colunga, J., Skwara, A., Vila, J. C., Bajic, D., & Sanchez, A. (2024). Global epistasis and the emergence of function in microbial consortia. Cell, 187(12), 3108- 3119. 75. Sefah, K., Shangguan, D., Xiong, X., O'donoghue, M. B., & Tan, W. (2010). Development of DNA aptamers using Cell-SELEX. Nature protocols, 5(6), 1169-1185. 76. Duim, H., & Otto, S. (2017). Towards open-ended evolution in self-replicating molecular systems. Beilstein journal of organic chemistry, 13(1), 1189-1203. 110 77. Markovitch, O., Sorek, D., Lui, L. T., Lancet, D., & Krasnogor, N. (2012). Is there an optimal level of open-endedness in prebiotic evolution?. Origins of Life and Evolution of Biospheres, 42, 469-474. 78. Taylor, T. J. (1999). From artificial evolution to artificial life. 79. Ruiz-Mirazo, K., Peretó, J., & Moreno, A. (2004). A universal definition of life: autonomy and open-ended evolution. Origins of Life and Evolution of the Biosphere, 34, 323-346. 80. McShea, D. W. (1994). Mechanisms of large‐scale evolutionary trends. Evolution, 48(6), 1747-1763. 81. Villa, A., Brunialti, E., Dellavedova, J., Meda, C., Rebecchi, M., Conti, M., ... & Ciana, P. (2022). DNA aptamers masking angiotensin converting enzyme 2 as an innovative way to treat SARS-CoV-2 pandemic. Pharmacological research, 175, 105982. 82. Park, G., Lee, M., Kang, J., Park, C., Min, J., & Lee, T. (2022). Selection of DNA aptamer and its application as an electrical biosensor for Zika virus detection in human serum. Nano convergence, 9(1), 1-10. 83. Andrews, S. (2010). FastQC: a quality control tool for high throughput sequence data. 84. Zhang, J., Kobert, K., Flouri, T., & Stamatakis, A. (2014). PEAR: a fast and accurate Illumina Paired-End reAd mergeR. Bioinformatics, 30(5), 614-620. 85. Bankevich, A., Nurk, S., Antipov, D., Gurevich, A. A., Dvorkin, M., Kulikov, A. S., ... & Pevzner, P. A. (2012). SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. Journal of computational biology, 19(5), 455- 477. 86. Bailey, T. L., Johnson, J., Grant, C. E., & Noble, W. S. (2015). The MEME suite. Nucleic acids research, 43(W1), W39-W49. 111 87. Bailey, T. L., & Elkan, C. (1994). Fitting a mixture model by expectation maximization to discover motifs in bipolymers. 88. Hill, M. O. (1973). Diversity and evenness: a unifying notation and its consequences. Ecology, 54(2), 427-432. 89. Dykhuizen, D. E., & Hartl, D. L. (1983). Selection in chemostats. Microbiological reviews, 47(2), 150-168. 90. McGowan, J. (2023). GC Content Calculator. 91. Wang, J., Wang, J., Huang, Y., & Xiao, Y. (2019). 3dRNA v2. 0: an updated web server for RNA 3D structure prediction. International journal of molecular sciences, 20(17), 4116. 92. Allawi, H. T., & SantaLucia, J. (1997). Thermodynamics and NMR of internal G.T mismatches in DNA. Biochemistry, 36(34), 10581-10594. 112 Supplementary Materials Fig. S1. Change in motif frequency across selection regimes. Sequence motifs of 6-40nt length were determined for populations selected with beads or cells (middle columns) as well as populations whose binding substrate switched from beads to cells (left column) and vice versa (right column), one representative population per treatment shown. The most frequent motifs per population (E-value < 0.05) were either sequences from the random 40nt region (light) or the primer binding sites (dark) of the DNA strand. All motifs were low-frequency in selection rounds 1 and 4, whereas high-frequency motifs appear by round 8 of selection in all treatments, indicating the effects of adaptation. Selection with beads alone results in a rise in frequency of motifs from the random 40nt region, while motifs following selection with cells are primarily from the primer binding sites. Both alternating selection regimes yield final motifs from both regions of the sequence, retaining the effects 113 of past selection. Motif frequencies across selection regimes indicate that the mode of adaptation depends on the selective context. Fig. S2. The 20 most frequent sequences in round 8 of selection with beads and cells. Legends beside each sequence refer to those shown in Fig. 1, which depicts the frequency of each sequence over time (3 replicate populations per treatment). Sequences shown here following selection with beads (A, B, C) correspond to Fig. 1A, B, and C, respectively, 114 while sequences selected with cells (D, E, F) correspond with Fig. 1G, H, and I. The most abundant sequences after selection with beads retain the same length as ancestral populations (~76nt) and have high frequencies compared to the rest of the population. The most common sequences following selection with cells, on the other hand, are still extremely rare. While the majority of these populations consist of long, rare sequences (see Fig. 1, J, K, L), the most frequent ones shown here are often shorter than 76nt. 115 Fig. S3. The 20 most frequent sequences in round 8 of variable selection with beads and cells. Legends beside each sequence refer to those shown in Fig. 2, which depicts the frequency of each sequence over time (3 replicate populations per treatment). Sequences shown here following alternating selection with beads before cells (A, B, C) correspond to Fig. 2A, B, and C, respectively, while sequences selected with cells before beads (D, E, F) correspond with Fig. 2G, H, and I. Depending on the extent to which populations diversified, as observed in Fig. 2, the length of the most common sequences vary. Populations which experienced less diversification tend to retain expected sequence lengths (~76nt), whereas more diversification corresponds with sequences which diverge from this expected range. In all cases, these highest-frequency sequences are still extremely rare, collectively comprising less than 0.3% of the populations. Fig. S4. Sequence length and melting temperature in round 8 of selection with beads or cells. Melting Temperature (Tm) as a function of sequence length for the 100 most frequent sequences in round 8 of selection indicates a greater variation in sequence length and therefore Tm following selection with cells compared to beads. 116 Fig. S5. Secondary structure loops and sequence length. Number of secondary structure loops increases with sequence length, with a greater number of loops for sequences selected with cells, compared to randomly generated sequences of the same length and GC-content (p-value = 0.0003, one-way ANOVA). Examples of two tertiary structures from derived sequences are shown. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 12.856831 0.794135 16.19 < 0.0001 * Time -0.695321 0.1393 -4.99 0.0005 * Adj. R2: 0.684952, df = 2 Table S1. Linear regression analysis of sequence diversity over 8 rounds of selection with magnetic beads. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 0.2026553 0.143763 1.41 0.1890 Time 0.0813626 0.025218 3.23 0.0091 * 117 Adj. R2: 461042, df = 2 Table S2. Linear regression analysis of sequence length diversity over 8 rounds of selection with magnetic beads. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 47.317458 0.49921 94.78 < 0.0001 * Time 0.2386777 0.094583 2.52 0.0128 * Adj. R2: 0.037182, df = 2 Table S3. Linear regression analysis of GC-content over 8 rounds of selection with beads, for the 20 most abundant sequences in the starting library, round 1, and round 8 of selection. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 11.456886 0.099746 114.86 < 0.0001 * Time -0.017648 0.017497 -1.01 0.3369 Adj. R2: 0.001582, df = 2 Table S4. Linear regression analysis of sequence diversity over 8 rounds of selection with yeast cells. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) -0.361425 0.511447 -0.71 0.4959 * 118 Time 0.333302 0.089714 3.72 0.0040 * Adj. R2: 0.537863, df = 2 Table S5. Linear regression analysis of sequence length diversity over 8 rounds of selection with yeast cells. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 11.396183 0.044465 256.29 < 0.0001 * Time -0.005205 0.0078 -0.67 0.5197 Adj. R2: -0.05311, df = 2 Table S6. Linear regression analysis of sequence diversity over 8 rounds of alternating selection, beginning with 4 rounds of selection with beads before 4 rounds with cells. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 11.479103 0.12344 92.99 < 0.0001 * Time -0.033074 0.021653 -1.53 0.1576 Adj. R2: 0.108098, df = 2 Table S7. Linear regression analysis of sequence diversity over 8 rounds of alternating selection, beginning with 4 rounds of selection with cells before 4 rounds with beads. Coefficients Estimate Std. Error t Ratio Prob>|t| 119 (Intercept) -0.150036 0.515411 -0.29 0.7769 Time 0.2325773 0.090409 2.57 0.0278 * Adj. R2: 0.338057, df = 2 Table S8. Linear regression analysis of sequence length diversity over 8 rounds of alternating selection, beginning with 4 rounds of selection with beads before 4 rounds with cells. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) -0.621934 0.461285 -1.35 0.2073 Time 0.4825503 0.080915 5.96 0.0001 * Adj. R2: 0.758589, df = 2 Table S9. Linear regression analysis of sequence length diversity over 8 rounds of alternating selection, beginning with 4 rounds of selection with cells before 4 rounds with beads. Source DF Sum of Squares Mean Square F Ratio Prob > F Model 7 45.325401 6.47506 12.2954 <0.0001 * Error 40 21.064955 0.52662 C. Total 47 66.390356 Table S10. Analysis of Covariance of sequence length diversity among treatments over rounds of selection. Coefficients 120 Estimate Std. Error t Ratio Prob>|t| (Intercept) 5.2554245 0.417965 12.57 < 0.0001 * Time 0.4282233 0.07919 5.41 < 0.0001 * Adj. R2: 0.168866, df = 1 Table S11. Linear regression analysis of the total number of loops in the secondary structures of sequences from the starting library, round 1 and round 8 of selection with cells. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 0.1426887 0.074341 1.92 0.0570 Time 0.0648585 0.014085 4.60 < 0.0001 * Adj. R2: 0.126907, df = 1 Table S12. Linear regression analysis of the number of multi-branch loops in the secondary structures of sequences from the starting library, round 1 and round 8 of selection with cells. Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 5.3115566 0.183885 28.89 < 0.0001 * Time 0.0173742 0.03484 0.50 0.6188 Adj. R2: -0.00543, df = 1 Table S13. Linear regression analysis of the total number of loops in the secondary structures of sequences from the starting library, round 1 and round 8 of selection with beads. 121 Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 0.1971698 0.045146 4.37 < 0.0001 * Time -0.010377 0.008554 -1.21 0.2271 Adj. R2: 0.003383, df = 1 Table S14. Linear regression analysis of the number of multi-branch loops in the secondary structures of sequences from the starting library, round 1 and round 8 of selection with beads. Source DF Sum of Squares Mean Square F Ratio Prob > F Spent Media 1 120.0000 120.000 13.8016 0.0003 * Error 118 1025.9667 8.695 C. Total 119 1145.9667 Table S15. Analysis of Variance of the total number of loops in the secondary structures of sequences randomly chosen following 8 rounds of selection with cells, compared to randomly generated sequences of the same length and GC-content. 122 Chapter 3: Experimental evolution of halophiles: rapid divergence along a multidimensional niche Abstract Niche theory underpins key eco-evolutionary processes, yet the nature and rate at which a species’ niche breadth shifts remains largely unknown. In particular, it is unclear whether adaptation will lead to associated tradeoffs versus improvements in novel environments, as well as how these responses are correlated among niche axes. Here, we explored multidimensional niche breadth evolution among extremophiles, where tradeoffs are pervasive yet lack empirical investigation. To assess the generality of niche shifts, we propagated a halophilic archaeon (Halobacterium salinarum) and bacterium (Salinibacter ruber) separately in rich and poor media for 60 generations and measured associated changes across novel conditions. We found that niche breadth evolution is rapid and highly divergent among taxa. Multidimensional tradeoffs emerged for bacterial populations, whereas fitness improvements were observed for archaeal lineages. In both cases, the largest fitness changes were documented in novel environments, demonstrating that pleiotropic effects often do not match in magnitude the adaptive response. By comparing across taxa and niche dimensions, we found that changes in niche width were parallel, despite divergence in niche breadth shape. Ancestral niche breadth appears to underlie these instances of divergence versus parallelism, providing a predictive lens for anticipating pleiotropy. 123 I. Introduction Niche theory underpins core concepts in ecology and evolution, including convergence, divergence, and specialization. (Grinnell 1917; Elton 1927; Hutchinson 1957; MacArthur and Levins 1967; Qiao et al. 2016; Carscadden et al. 2020). A species’ niche (i.e. the range of conditions that support growth and reproduction) (Box 1) has been widely explored by measuring abundances across environments (Peterson et al. 2011), with specialists and generalists distributed among a narrower or larger range of conditions, respectively (Sexton et al. 2017; Kassem 2024). However, this method is challenged by the multidimensional nature of the niche (Hutchinson 1957; Lynch and Gabriel 1987), since the role of each environmental variable, or “niche axis,” is almost impossible to determine. Moreover, most studies have characterized a species’ current niche space, rather than document its evolution (Sexton et al. 2017; Carscadden et al. 2020). It is therefore unclear how a species’ niche breadth is correlated among niche axes, and the role of each axis in shaping niche evolution remains unknown (Sexton et al. 2017; Bennett & Lenski 2007; Carscadden et al. 2020). There are two prevailing perspectives for how selection may shape a species’ niche. The first expectation is that selection in one context results in fitness tradeoffs in a novel environment, either due to the cumulative effects of drift or as a pleiotropic consequence of beneficial mutations in the selective environment (Levins 1968; Travisano and Lenski 1996; Lang et al. 2009). Alternatively, changes in traits may systematically improve fitness in both selective and novel environments, particularly if the environments are functionally similar (Travisano and Lenski 1996; Jerison et al. 2020). So far, the vast majority of studies on niche breadth evolution have been observational or comparative (Fahimipour and Gross 2020; Maeno et al. 2021), which severely limits our ability to distinguish among these possibilities without knowledge of an ancestral state (Reznick 1985; Bennett and Lenski 2007). Experimental evolution studies provide a unique opportunity to investigate multidimensional niche evolution. Previous work with laboratory microbial populations 124 have leveraged high degrees of replication to demonstrate the stochasticity and complexity of byproduct effects (Ostrowski et al. 2005; Bennett and Lenski 2007; Lee et al. 2009; Jasmin and Zeyl 2013; Rodríguez-Verdugo et al. 2014; Leiby and Marx 2014; Jerison et al. 2020), yet most experimental studies of niche evolution have only investigated a single environmental axis, such as temperature, or a single selective environment (Carscadden et al. 2020). As a result, it is difficult to make conclusions about the generality of pleiotropic effects across niche dimensions, species, or selective regimes. It is also difficult to extend our understanding of multidimensional niche evolution beyond non-extreme conditions, since almost all experimental studies have been conducted with mesophiles (e.g. Escherichia coli, Saccharomyces cerevisiae, etc.). Extremophiles, which not only tolerate but require extreme conditions, demonstrate how fitness tradeoffs emerge as a result of constraints on energy and resource availability (Rampelotto 2013; Mauro and Ghalambor 2020). When limited resources are allocated for strategies to overcome environmental stressors, the emergence of tradeoffs is expected to be particularly important (Levins 1968), nevertheless our understanding of tradeoffs is almost exclusively based on evolution in non-extreme environments. This noticeable gap extends in particular to the Archaea domain, which is paradoxically both the main group to thrive in extreme environments and the least-studied experimentally (Rampelotto 2013; Rampelotto 2024; Medina-Chávez and Travisano 2022). Niche dimensionality highlights that neither the drivers nor the consequences of adaptation in extreme conditions are commonly univariate – even adaptation to a particular stressor is constrained by other abiotic and biotic factors (Tobler et al. 2018; Jayasingh et al. 2023). This complexity of environmental components can shape life-history traits, ecological interactions, and evolutionary tradeoffs (Snell-Rood et al. 2015; Jeyasingh and Weider 2005; Jayasingh et al. 2023), yet the multidimensional nature of niche evolution is rarely studied in extremophiles. Instead, most studies focus on niche shifts along the “extreme” environmental axis, such as high salinity, temperature, or pH (He et al. 2023; Vauclare et al. 2020), making it difficult to understand correlations among niche axes or 125 how tolerance along one axis may be dependent on another. In this study, we explored the intersection of these gaps in knowledge by experimentally evolving halophilic species of Archaea and Bacteria in two high-salinity environments and measuring corresponding shifts in their multidimensional niche breadth. Since parallel adaptive responses in one environment may cause divergent shifts in niche breadth, we chose two species which are highly convergent phenotypically yet distinct phylogenetically, allowing for the possibility that disparate outcomes may only be observed in novel environments. Both of the species used, an archaeon, Halobacterium salinarum, and a bacterium, Salinibacter ruber, are found in salt-saturated environments, representing the upper limit for adaptation at high salinity (Oren 2013). The two red- pigmented species are remarkably similar, utilizing the same “salt-in” strategy to combat osmotic stress and exhibiting convergence at both the physiological and molecular levels, while phylogenetically belonging to separate domains (Kunka et al. 2020; Oren 2013). By choosing species which are phenotypically convergent, we were able to draw conclusions about parallelism during niche evolution: Any interspecific differences observed are due to differences in organismal constraints, rather than having begun at different optima (Carsacadden et al.2020; Angert et al. 2020). If we had selected two species with different initial salt tolerances, for example, any divergence in their niche evolution would not address open questions of generality across species. Since theory predicts that selection operates differently at a species’ niche “edge” versus “core” (Angert et al. 2020), we also included two selective environments – a “poor” and “rich” media – to test whether selection is stronger when population mean fitness is low (Caruso et al. 2017). Following selection in rich and poor environments, we investigated their niche breadth evolution from three perspectives, adopting a similar conceptual framework as Bennett and Lenski (2007): (i) Directionality (within-taxa); Does selection in each environment correspond with a significant loss or gain of fitness in the non-selective environment? 126 (ii) Generality (between-taxa); Do bacterial and archaeal lineages differ in their responses to selection and corresponding shifts in niche breadth? (iii) Dimensionality; Are the consequences of selection in each environment the same across niche dimensions? We find significant divergence in niche breadth evolution among bacterial and archaeal lineages. Fitness tradeoffs rapidly emerge for bacterial populations, whereas fitness gains are observed for archaeal populations in novel environments, particularly for archaeal lineages selected in the poor environment. This effect is not only observed in the two selective and non-selective environments but also across a range of resources, where tradeoffs (for S. ruber) and improvements (for H. salinarum) are demonstrated along a gradient of resource abundances. We also find that selection history shapes fitness along a salinity axis, but this affect depends on resource abundance, indicating interactions among niche dimensions. Together, these results demonstrate that the multidimensional nature of a niche is central to understanding the drivers and consequences of adaptation in extremely high-salinity conditions. While the two species have a similar ancestral fitness in either environment (and therefore coexist in natural settings (Oren 2013)), we find that their ancestral niche breadths are indicative of specialist and generalist strategies for bacterial and archaeal populations, respectively, underscoring how knowledge of a species’ niche breadth may provide predictive power in anticipating its evolution. 127 Box 1. Quantifying the niche Definitions of the niche vary widely across the literature. Here, we consider the niche to be the range of conditions within which individuals of a genotype can grow and reproduce (Kassen 2024). In experimental systems, a genotype’s niche is determined by estimating fitness (measured using the population’s growth curve [e.g. growth rate or Area Under the Curve (Kassen 2024; Aranda-Díaz et al. 2025)]) as a function of an environmental variable. This measure of fitness is depicted as a reaction norm comparing two distinct environments (a) or a tolerance curve displaying fitness across an environmental gradient (b). These approaches to quantifying the niche can describe both the niche’s shape (i.e. variation in fitness within the hospitable conditions) and the niche’s width (i.e. the environmental limits beyond which the genotype cannot reproduce). Following selection, a common expectation is to observe an improved fitness in the selected environment, relative to the ancestor, which corresponds with a decrease in fitness in a non-selected environment (i.e. a fitness tradeoff) (a). This tradeoff may indicate an increase in specialization, as represented by the dotted line in panel (b), where the selected genotype has a steeper and narrower tolerance curve. Multidimensional niches It is important to remember that each environmental variable represents a single axis of the niche. As the niche evolves, changes in fitness can be observed across a range of environmental axes (e.g. pH, salinity, nutrient availability) (Kassen 2024). Interactions among environmental components may be critical for understanding multidimensional shifts in the niche, for example the effects of high salinity may depend on resource availability (c) (d). Experimental studies of niche evolution can explore the role of context dependence by isolating effects across environmental axes. 128 II. Materials and Methods (a) Strains and ancestral isolates Two ATCC halophilic strains were used, an archaeon Halobacterium salinarum (Harrison and Kennedy) Elazari-Volcani 19700, and a bacterium Salinibacter ruber (Anton et al. 2002) BAA-605. For initial reactivation, each strain was grown in its respective media, according to ATCC standards – ATCC 213 media for H. Salinarum and ATCC 2402 media for S. ruber. Incubation was held at 37°C for both strains, and liquid cultures were agitated at 250 rpm. Four clonal populations were established from single colonies for each strain on their respective media plates, identifiable after 10 days for H. salinarum and 14 days for S. ruber, resulting in a total of eight clonal populations (4 archaeal isolates , 4 bacterial isolates). Following isolation, each population was grown once and then partitioned; an aliquot was cryopreserved with 25% glycerol and stored at -80°C (designated as ancestral populations), and another aliquot established the selection lines, afterwards designated as derived populations. (b) Environments: media composition To explore the generality of niche breadth evolution, we used both medias (ATCC 213 and ATCC 2402) as selective environments. We designated ATCC 213 media as “rich” media, due to the high amount of yeast extract and the presence of tryptone, commonly used as a nitrogen source, media composition [per liter]: 250 g NaCl, 10 g yeast extract, 2.5 g tryptone, 10.0 g MgSO4 7H2O, 5.0 g KCl, 0.2 g CaCl2 2H2O, and 20 g bacteriological agar (for plates). We assigned “poor” media to ATCC 2402, media composition [per liter]: 195 g NaCl, 1.0 g yeast extract, 49.5 g MgSO4 7H2O, 5.0 g KCl, 1.25 g CaCl2 2H2O, 34.6 g MgCl2 6H2O, 0.625 g NaBr, 0.25 g NaHCO3, and 20 g bacteriological agar (for plates). Final pH was adjusted to 7.4 and 7.2, respectively. 129 Media components of each environment were classified into three groups for better characterization of the niche: Salt (NaCl), Resources (yeast extract and tryptone) and Other Salts (MgSO4 7H2O, KCl, CaCl2 2H2O, MgCl2 6H2O, NaBr, and NaHCO3) (Table 1). Table. 1. Abundance of salts and other resources in the rich and poor medias used during evolution experiment. Amounts given (in grams) are per one liter of water. Rich media (ATCC 213) contains higher concentrations of salt (NaCl) and resources (yeast extract and tryptone), compared to poor media (ATCC 2402), but lower concentrations of other salts (MgSO4*7H2O, KCl, CaCl2*6H2O, NaHCO3, and NaBr). Both S. ruber and H. salinarum grow significantly faster and reach higher population densities in rich media compared to poor media (see Fig. 2). For more details about the media components, see Suppl. Table 1. (c) Experimental evolution The eight clonal populations (4 archaeal and 4 bacterial) were each propagated in the two liquid environments, poor and rich media, for 10 transfers (~ 60 generations). Each transfer was conducted during stationary phase, where 100µl of well-mixed culture was transferred into 6 ml of fresh media. Initial transfers in the rich media were conducted every 5 days, while transfers in the poor media were carried out every 20 days, in order to reach stationary phase (Fig. 1). After 4 transfers in poor media (80 days), corresponding to 130 ~ 24 generations, improved growth was observed for bacterial and archaeal populations selected in poor media, so their transfer interval was reduced to 10 days. Figure. 1. Experimental design for selection and niche breadth assays. Four clonal populations were each established for H. salinarum and S. ruber from individual colonies (ancestral populations) and subsequently propagated in rich (ATCC 213) and poor media (ATCC 2402) for 10 transfers (derived populations). Populations were transferred every 5 days in rich media and initially every 20 days in poor media before being reduced to 10 days after the first 4 transfers (a). All ancestral and derived populations were then grown in rich and poor media (4 replicates each) (b), their growth curves compared to assess growth in the selective and novel environments. 131 (d) Growth assays To determine if the populations had responded to selection in each environment – and whether adaptation corresponded with a tradeoff in the alternative media – we conducted growth assays for ancestral and derived populations in each media (Fig. 1). The high salt concentration in the media prevents standard use of a plate reader to measure the optical densities (OD600) of the growing culture, since the salt precipitates in the media following 48 hours of incubation even on a 48-well plate. Instead, we measured growth by destructively sampling from a series of inoculated tubes, ensuring sufficient time for populations to reach stationary phase prior to salt precipitation. Measurements were carried out daily for 5 days in rich media and every 48 hours for 11 days in poor media. Samples of 1 ml from the cultures were extracted from the tubes to assess growth via optical density (O.D.) measurements at 600 nm using a spectrophotometer (Cary 300 UV-Vis). Each sample measurement was performed three times to minimize variation due to sampling error. (e) Niche breadth experiments Effects of resource abundance All ancestral and derived populations reached significantly higher densities when grown in rich media, compared to poor media. Pilot testing indicated that this growth disparity may be attributed to the disparity of resource abundances between the two medias: rich media contains 10 g of yeast extract and 2.5 g of tryptone, while the poor media contains 1 g of yeast extract and no tryptone. To confirm this observation and disentangle the relative effects of yeast extract versus tryptone on population growth, we evaluated whether poor media supplemented with the same composition of resources from rich media could retrieve population densities found in the rich environment. We grew derived archaeal populations (4 replicates) in three different poor-based medias containing an additional 1) 9 g of yeast extract (10 g total) and 2.5 g of tryptone, 2) 9 g of yeast extract (10 g total) but no tryptone, and 3) only 1 g of yeast extract but 2.5 g of tryptone. 132 Evaluating niche breadth evolution across a resource gradient Upon confirming that the significant growth disparity among populations grown in poor versus rich media can largely me attributed to the high concentrations of yeast extract present in the rich media, we created a resource gradient of rich media modified to contain 1, 3, 5, 7, 9, or 10 g of yeast extract and generated 5-day growth curves for all ancestral and derived populations along this gradient. Effects of salinity at low resources To determine how selection shaped the absolute edge of the populations’ niche breadth, we explored the effects of tryptone and salt in low-resource environments for all ancestral and derived populations. We modified the rich media to contain only 1 g of yeast extract (hence, “low-resources”) and included two conditions that contained 0 g of tryptone with either 195 g or 250 g of NaCl, as well as two conditions that contained 2.5 g of tryptone with either 195 g or 250 g of NaCl. These amounts were the equivalent abundances present in either of the two selective medias, representing all possible combinations of tryptone with sodium chloride at low resources, without the additional “other salts” present in the poor media. Growth curves were generated using our standard 5-day protocol (see below). Evaluating niche breadth evolution across a salinity gradient Since we observed the effects of selection in shaping each species’ niche breadth across a resource gradient, we investigated the possibility that salinity also contributed to their adaptation and therefore its effects could be observed at the edges of their niche breadth. To test this, we created salinity gradients at the lower and upper edges of the niche. Upon observing that sodium chloride concentrations had little effect on fitness at low resources, we decided to investigate the effects of salinity along the remaining environmental axis: concentration of “other salts.” 133 The poor media, which only has 1 g of yeast extract, no tryptone, and 195 g of NaCl (compared to 250 g in rich media), was considered near the lower edge of the niche. However, since it contains greater abundances of “other salts” than the rich media, we grew ancestral and derived populations in poor media whose composition of “other salts” was replaced with the composition present in rich media. By comparing their growth again in poor media, we were able to isolate the effects of these “other salts” based on selection history. To contrast this outcome in the upper edge of the niche, we modified rich media to contain the equivalent abundance of NaCl present in poor media (195 g) and compared populations’ growth in this environment with growth in rich media with 250 g of NaCl. In doing so, we were able to elucidate the extent to which the tradeoffs and pleiotropic responses observed can be attributed to differences in salinity among the two selective environments and whether these responses were context-specific (i.e. dependent on background resource availability). (f) Growth and fitness measurements As mentioned above, all growth curves were generated via inoculation of a series of tubes and destructively sampling 1mL from each tube daily to measure the optical density (OD600) of the population (3 measurements per tube). With the exception of the first experiment measuring growth in the two selective environments, all subsequent growth curves were generated over 5 days for all ancestral and derived populations (4 replicates per treatment). The area under the curve was used as a proxy for absolute fitness, and relative fitness was calculated as the fitness of the derived populations divided by the fitness of the ancestral populations. Since archaeal populations selected in poor media paradoxically appeared to have a lower fitness in poor media compared to ancestors, we investigated the possibility that in this case optical density measurements may not be capturing changes in competitive fitness relative to ancestors, rather than absolute fitness. We therefore proceeded to measure the 134 competitive fitness of the derived populations by generating growth curves for derived archaeal monocultures, ancestral archaeal monocultures, and 50% co-cultures in poor media. If the competitive fitness of derived populations remained unchanged (i.e. were neutral), we hypothesized that the fitness of the co-cultures would be similar to the average of the two monocultures individually. Calculations for the area under the curve were computed with GraphPad Prism (9.4.1), and all statistical analyses were carried out using JMP Pro 16.0.0. III. Results Emergence of tradeoffs is species-specific Following selection in rich and poor media, growth curves were generated for each population in their selective and novel environments. As expected, the environment had a large effect on optical densities across treatments; populations grown in rich media yielded significantly higher densities compared to poor media, regardless of selection history (p- value <0.0001, nested ANCOVA, Suppl. Table 9) (Fig. 2). For bacterial populations, selection corresponded with significant tradeoffs in the novel environments. Specifically, populations selected in poor media reached lower densities when grown in rich media (p- value <0.001, ANCOVA, Suppl. Table 10), while populations selected in rich media displayed lower densities when grown in poor media (p-value <0.0001, ANCOVA, Suppl. Table 11) (Fig. 2a, Suppl. Fig. 1), compared to ancestors and other selected populations. The rapid emergence of tradeoffs was therefore general across environments. For archaeal populations, densities also differed significantly by selection history in rich (p-value = 0.0205, ANCOVA, Suppl. Table 12) and poor (p-value <0.0001, ANCOVA, Suppl. Table 13) media, but their patterns of response diverged from S. ruber (Suppl. Fig. 1). All derived archaeal populations grew faster in rich media, compared to ancestors (Fig. 2b), yet populations selected in poor media reached lower densities 135 compared to other populations when grown in the same poor environment (Suppl. Fig. 1d). We investigated this surprising outcome further and found that while the growth (as measured via optical density) had declined for archaeal populations selected in poor media, their competitive fitness compared to ancestors appeared to have improved, indicating that their response to selection was adaptive (Suppl. Fig. 2). Overall, the repercussions of past selection were significantly different between bacterial and archaeal populations in both rich (p-value <0.0001, nested ANCOVA, Suppl. Table 14) and poor environments (p-value <0.0001, nested ANCOVA, Suppl. Table 13), indicating that the rapid emergence of trade- offs was not universal, but species-dependent. For bacterial populations, selection in one environment corresponded with lower growth in the novel environment, while archaeal populations selected in poor media exhibited pleotropic improvements in the rich environment. Figure 2. Fitness of S. ruber and H. salinarum grown in rich and poor media. Growth curves were generated for all ancestral and selected populations (Suppl. Fig. 1), and fitness was estimated as the area under the curve during five days of growth for S. ruber (a) and H. salinarum (b) (error bars ± SEM). The environment had a large effect on fitness, with populations having a significantly higher fitness when grown in rich media compared to poor media for all treatments (p-value <0001*, nested two-way ANOVA, Suppl. Table 2). For bacterial populations, selection resulted in significant tradeoffs in the non-selected 136 environment, with populations selected in poor media reaching a lower fitness when grown in rich media (p-value = 0.0013*, one-way ANOVA, Suppl. Table 3) and populations selected in rich media having a lower fitness when grown in poor media (p-value = 0.0005*, one-way ANOVA, Suppl. Table 4), compared to ancestors and other selected populations. The fitness of archaeal populations also differed significantly by treatment in rich media (p-value = 0.0179*, one-way ANOVA, Suppl. Table 5) but not poor media (p- value = 0.1741, one-way ANOVA, Suppl. Table 6) and with different relative effects: both selected populations have a higher fitness in rich media, compared to ancestors. Overall, the effects of past selection differed significantly between bacterial and archaeal populations grown in rich (p-value = 0.0002*, nested two-way ANOVA, Suppl. Table 7) and poor (p-value < 0.0001*, nested two-way ANOVA, Suppl. Table 8) environments, indicating that the emergence of tradeoffs was not uniform across species. Niche breadth evolution is multidimensional To understand how selection in each environment corresponded with shifts in niche breadth, we conducted subsequent growth assays in a variety of novel media compositions. By isolating the effects of each environmental component, we aimed to elucidate whether adaptive consequences for each species were the same across niche dimensions. Response to selection is observed across resource gradient As discussed, all lineages reached significantly lower population densities when grown in the poor versus rich environment. We hypothesized that this growth disparity is due to the lower abundance of resources present in the poor media. To determine the limiting effects of each nitrogen source (i.e. yeast extract and tryptone), we modified the poor media to contain the equivalent abundances of these resources found in the rich media. We found that fitness differed significantly among treatments (p-value <0.0001, one-way ANOVA, Suppl. Table 17), indicating that both resources contribute significantly to population growth (Suppl. Fig. 3). However, the largest growth disparities were 137 observed in populations grown in media with high levels of yeast extract (10 g), compared to the low abundances found in poor media (1 g), regardless of the presence of tryptone, suggesting that the high concentrations of yeast extract largely drove the fitness benefits observed in rich media. Given this result, we hypothesized that yeast extract abundance likely shaped the evolutionary response in the two selective environments, the effects of which may be observed across a range of resource levels. To test whether selection shaped the niche breadth of each lineage along this dimension, we measured the absolute fitness of each ancestral and derived population across a gradient of yeast extract abundances (i.e. rich media modified to contain 1, 3, 5, 7, 9, and 10 g of yeast extract) (Fig. 3). We found that fitness values differed significantly among ancestral and selected populations across this gradient for both bacteria (p-value <0.0001, nested ANOVA, Suppl. Table 18) and archaea (p-value <0.0001, nested ANOVA, Suppl. Table 19), indicating responses to selection. Interestingly, the same divergence between taxa observed previously is also evident across this range of resources. For populations selected in poor media, the relative fitness values compared to ancestors differed significantly between bacterial and archaeal populations (p- value <0.0001, one-way ANOVA, Suppl. Table 20), whereas the relative fitness of populations selected in rich media were not significantly different between taxa (p-value = 0.1150, one-way ANOVA, Suppl. Table 21). These results indicate that bacterial populations selected in poor media not only experienced fitness tradeoffs in the rich environment but also across a range of resources, compared to ancestors. Likewise, bacteria selected in the rich environment not only have a higher fitness in their selective environment, compared to ancestors, but also across a range of intermediate resource levels. In contrast, both of the selected archaeal populations have higher absolute fitness values than ancestral populations along this gradient. Together, these results suggest that each species’ response to selection corresponds with shifts in the resource dimension of their niche breadth. 138 Moreover, while the ancestral fitness of H. salinarum and S. ruber are very similar at high and low resources (i.e. the same resource levels present in the rich and poor environments), the shape of their ancestral niche breadth differs across this gradient (p- value <0.0001, nested ANOVA, Suppl. Table 22). The fitness of archaeal ancestors generally increases with higher resource levels (p-value <0.0001, linear regression, Suppl. Table 23) (Fig. 3b), whereas bacterial ancestors experience a higher fitness at intermediate abundances of yeast extract, compared to the upper and lower edges of the niche (p-value <0.0001, linear regression, Suppl. Table 24) (Fig. 3a). This preexisting difference in their niche breadth helps explain the divergent consequences of their selection, which we discuss below. Figure 3. Absolute fitness values for archaeal and bacterial populations across a gradient of yeast extract abundance. Absolute fitness was determined from growth curves measured over five days, with four populations per treatment, in rich media modified to contain 1, 3, 5, 7, 9, or 10 grams of yeast extract for S. ruber (a) and H. salinarum (b) (error bars ± SEM). Fitness values differed significantly among ancestral and selected populations across the yeast extract gradient for both bacterial (p-value <0.0001, nested ANOVA, Suppl. Table 18) and archaeal (p-value <0.0001, nested ANOVA, Suppl. Table 19) populations, indicating responses to selection. Relative fitness values compared to ancestors differed significantly between bacteria and archaea for 139 populations selected in poor media (p-value <0.0001, one-way ANOVA, Suppl. Table 20) but not for populations selected in rich media (p-value = 0.1150, one-way ANOVA, Suppl. Table 21). These results suggest that fitness tradeoffs emerged for bacterial populations selected in poor media, where these populations have lower fitness values across varying resource levels compared to ancestors, but archaeal populations selected in poor media manifest higher relative fitness values across the same gradient. Populations selected in rich media, however, exhibit parallel responses to selection across taxa, with both selected populations having higher fitness measures relative to ancestors. The nature of the niche evolution selection observed in rich and poor media (see Fig. 2) span a range of non- selected environments. Parallel synergistic effects emerge at edge of the niche Having observed shifts in niche breadth along a resource gradient, we explored whether the effects of selection extended to the edge of the species’ niche. To do this, we grew all populations in low-resource environments (1 g of yeast extract) either with or without the levels of NaCl and tryptone present in poor and rich media (195 g NaCl, 0 g tryptone; 250 g NaCl, 2.5 g tryptone, respectively), creating all possible combinations to isolate their effects. We found that the absolute fitness of populations differed significantly by the amount of salt and tryptone present (p-value <0.0001 [for both factors], two-way nested ANOVA, Suppl. Table 25), with the effects of salt changing depending on tryptone presence (p-value <0.0001, two-way nested ANOVA, Suppl. Table 25) (Fig. 4ab). Selection history, however, did not result in significant fitness differences among the two species (p-value = 0.0615, two-way nested ANOVA, Suppl. Table 25). These results indicate that selection resulted in parallel fitness effects in low-resource environments, with all populations exhibiting a non-additive increase in fitness when grown with high salt and tryptone. 140 Figure 4. Fitness of bacterial and archaeal populations grown in low and high resources with varying salinity. Fitness was estimated for all ancestral and selected populations grown in low-resource environments (1g yeast extract) (a,b) and high-resource environments (10g yeast extract) (c,d) with salt abundances altered to isolate the effects of salinity. Mean fitness among the four replicate populations per treatment are shown (error bars ± SEM), along with the effects of selection history, salt, and their interaction when resource abundance is high (two-way ANOVA) (c,d). Together, these results indicate that niche shifts along the salinity axis depend on resource availability. Selective consequences diverge along salinity axis in high-resource environment After exploring the effects of salinity at the lower edge of the niche, we examined whether this response was dependent on resource availability. In particular, we grew each population in a high-resource environment with salt abundances corresponding to each selective environment (see methods) (Fig.4cd). 141 The relative fitness of bacterial populations in the high-resource environments depends on treatment (i.e. selection history) (p-value = 0.0038, two-way ANOVA, Suppl. Table 26) and environment (p-value = 0.0242, two-way ANOVA, Suppl. Table 26), with the effects of the environment changing depending on the treatment (p-value = 0.0071, two-way ANOVA, Suppl. Table 26) (Fig. 4c). This indicates a similar Genotype x Environment interaction as observed in the poor environments, where populations selected in poor media have a higher relative fitness when salt concentrations are equivalent to their selective environment. The relative fitness of archaeal populations likewise depends on the environment (p-value = 0.0004, two-way ANOVA, Suppl. Table 27) but not selection history (p-value = 0.8375, two-way ANOVA, Suppl. Table 27), indicating that the effects of salinity on the selected populations depends on resource availability (Fig. 4d). Together, these results demonstrate that evolution along one niche axis is context- dependent; niche shifts are parallel for bacterial and archaeal species in low-resource environments but divergent at high resources. Combined with previous observations along resource gradients, these results illustrate that both bacterial and archaeal niche evolution is multidimensional, observed along both salinity and resource axes (Table 2). 142 Table 2. Media components for all possible environments based on rich and poor media. In this study, the fitness of each population was estimated in 16 out of the 19 possible environments to elucidate multidimensional niche evolution. Media components 143 are shown in grams per liter of water, and the figures illustrating comparisons among these environments are indicated on the right. A summary of the results in each environment is shown for S. ruber and H. salinarum; responses to selection are parallel among species at the lower edge of the niche but diverge when resource abundance is high. IV. Discussion Microbial life adapts to a confluence of physiochemical stressors in the landscape of extreme environments. The pleiotropic consequences of their adaptation is central to understanding patterns of diversification and resilience to environmental change, yet the nature of these tradeoffs remains largely unexplored. In particular, while niche breadth evolution over one niche axis has been investigated in a variety of systems (reviewed in Sexton et al. 2017), the correlation among different niche axes is rarely explored empirically (Carscadden et al. 2020; Bebber and Chaloner 2022), particularly among extremophiles. In this study, we used experimental evolution with a halophilic archaeon (Halobacterium salinarum) and bacterium (Salinibacter ruber) in two selective environments, allowing the investigation of niche breadth evolution directionality (within- taxa), generality (between-taxa), and along a range of environmental axes. Our study therefore informs longstanding questions about how the shape and width of a species niche breadth changes in response to selection. Directionality: Selection corresponds with significant fitness changes in non-selective environments Models of ecological specialization often expect adaptation in one environment to correspond with fitness reduction in others (Futuyma and Moreno 1988; Bono et al. 2017; Chavhan et al. 2021). Alternatively, specialization can emerge without associated trade- offs (Whitlock 1996; Bebber and Chaloner 2022) or even lead to fitness gains in new environments (Lee et al. 2009; Jerison et al. 2020). In this study, we find evidence for both 144 outcomes of selection, with tradeoffs generally emerging in bacterial populations and corresponding improvements observed for archaeal populations selected in poor media. Rapid emergence of fitness tradeoffs in bacterial populations Following 60 generations of selection in rich and poor environments, we observed a reduction in fitness for all bacterial populations grown in their non-selected environment (Fig. 2). The appearance of these tradeoffs was general among bacterial populations, emerging for populations selected in both rich and poor environments and extending across a range of resources (Fig. 3). While fitness tradeoffs have long been hypothesized as an adaptive cost, they are rarely demonstrated empirically (Bono et al. 2017; Chavhan et al. 2021) nor generally across replicates, given the influence of chance (Bennett and Lenski 2007; Jerison et al. 2020). Moreover, these tradeoffs emerged quickly, appearing after just 10 transfers (~ 60 generations) in each media, in contrast with estimates for beneficial mutations typically requiring ~ 250 generations to rise to majority (Lenski 2017; Lenski and Travisano 1994). This rapid change observed among bacterial populations may be attributed to their halophilic nature, given theoretical expectations that species with narrower niche breadths may require less time to fix beneficial mutations (Whitlock 1996). Recent experimental work with other halophilic strains reflects this prediction, where populations selected with environmental stressors experienced reduced growth rates in non-extreme conditions, although changes weren’t monitored as early as they were here (Bartha 2022). Interestingly, the fitness reduction observed in non-selected environments does not exhibit the quantitative association predicted by Levins’ principle of allocation (Levins 1968), which expects adaptation and its associated tradeoffs to match in magnitude. While the growth of poor-selected populations improved in the poor environment compared to ancestors (Suppl. Fig. 1), a clear trade-off emerged for rich-selected populations in poor media regardless of improvement in the rich environment. A lack of correlation between tradeoffs and the degree of adaptation aligns well with previous experimental results with 145 non-extremophiles in which such associations appear to be random (Bennett and Lenski 2007; Chavhan et al. 2021). Nevertheless, fitness reductions in novel environments were generally observed across bacterial populations. Archaeal populations improve fitness in novel environments In contrast with S. ruber, selection yielded pleiotropic improvements among archaeal populations. However, fitness gains were not general across treatments. When comparing growth in the two selected environments, improvements were only observed for populations grown in rich media (Fig. 2b). This is true across intermediate and high- resource environments, where the fitness of selected populations was greater than ancestors grown between 5 and 10 grams of yeast extract but no difference was measured among populations at low resources (1 – 3 g of yeast extract) (Fig. 3b). In particular, selection in poor media led to the most significant fitness gains at intermediate resources compared to ancestors (Fig. 3b). Since previous studies have found that H. salinarum reaches a growth optimum at a range of salinities that encompasses both the rich and poor medias (Valentine 2007; Patil et al. 2023), we hypothesize that the low resources (rather than low salinities) in the poor media drove this response. The dip in fitness at low resources for ancestral populations also suggests that resource scarcity in the poor environment constituted a stronger selective pressure compared to the rich environment, possibly eliciting the greater corresponding change. This is consistent with previous experimental work with H. salinarum where most mutations following propagation at varying salinities were not associated with halotolerance (Bartha 2022). Similar to bacterial populations where the magnitude of adaptation did not match corresponding tradeoffs, pleiotropic improvements were observed for archaeal populations regardless of fitness gains in their selected environment. This indicates that the niche breadth of these halophiles can rapidly shift even in the absence of an adaptive response. Since most studies of halophilic archaea are either comparative (Stan-Lotter and Fendrian 2015) or focus solely on salinity tolerance (Sabet 2009), our results are among the first to 146 demonstrate the rapid emergence of fitness improvements in non-selected environments. Future work should disentangle the possible mechanisms underlying such niche expansions. Generality: Niche evolution diverges in shape but not absolute width A species’ niche breadth can be characterized both by the height of the fitness curve within an environmental gradient and the width of this curve defining its fundamental limits. In this study, we observe divergent shifts in the shape of fitness curves across taxa; selection generally resulted in tradeoffs for bacterial populations versus fitness increases for archaeal populations (Fig. 3). At the absolute edge of the niche, however, changes in niche width were parallel (Fig. 4). In particular, where no growth was observed among ancestral populations (at low-resources, 195 g NaCl, with 2.5 g of tryptone), all selected populations gained the ability to grow (Fig. 4). Selection in poor media led to the biggest fitness increases in both cases, expanding the absolute niche width for all bacterial and archaeal populations. This parallelism occurred across these low-resource conditions, where selected populations exhibited similar shifts in fitness relative to ancestors, regardless of taxa (Table 2). The degree of parallelism or divergence among populations’ response to selection appears to be correlated with their ancestral state. The fitness of both archaeal and bacterial ancestors were similar among low-resource conditions, and their evolutionary responses were subsequently alike (Fig. 3). Likewise, the ancestral fitness curves along a gradient of resources were highly disparate among taxa, possibly explaining their divergent responses (Fig. 3). In particular, the fitness of bacterial ancestors across this gradient aligns with expectations for a specialist population: high fitness measured among intermediate resources which tapered at each extreme (Fig. 3a). Archaeal ancestors were comparatively generalist in nature; the curve is more flat, with less extreme deviations among high and low resources (Fig. 3b). This difference indicates that preexisting variability in niche breadth may significantly affect adaptation in novel environments – a factor often 147 overlooked in comparative studies where the ancestral state is unknown (von Meijenfeldt et al. 2023) or studies focused on an extreme environmental axis (He et al. 2023; Vauclare et al. 2020). One indication that the preexisting variability in fitness curves shaped niche breadth evolution in these populations is that the two species are otherwise remarkably similar. Despite being divergent phylogenetically, H. salinarum and S. ruber converge physiologically, with different genes yielding similar phenotypes (Oren 2013). Lateral transfer among Halobacteriaceae and S. ruber has also led to shared genes and gene clusters whose effects span salt adaptation and cell biology generally (Oren et al. 2004; Mongodin et al. 2005). While these modular adaptive elements have led to phenotypic resemblance among the species, in this study we document ancestral differences in niche breadth that have not previously been observed. Correlations among ancestral niche breadth and responses to selection suggest that knowledge of prior plasticity may aid predictions of adaptation in novel environments, as has been suggested previously (Schaum and Collins 2014; Ghalambor et al. 2015). Dimensionality: Selective consequences differ across niche dimensions The concept of the niche in ecology has been integral yet controversial over the last century (Grinnell 1917; Elton 1927; Hutchinson 1957; Blonder et al. 2014; Carscadden et al. 2020; Chase and Leibold 2003). In particular, the multidimensional nature of niche evolution remains unclear (Bebber and Chaloner 2022). Here, we explored the niche devoid of interspecific interactions, using an experimental approach to isolate changes across environmental axes. We asked: Are the adaptive consequences of selection the same across niche dimensions? Niche evolution is multidimensional We observe changes across niche dimensions for bacterial populations selected in each environment. Selection in rich media led to fitness improvements across a range of 148 intermediate and high resources (Fig. 3a), however no change was observed along a salinity gradient at high resources (Fig. 4c), and the effect of salt on selected populations was generally context-dependent, occasionally leading to tradeoffs depending on resource availability (Fig. 4). Likewise, populations selected in poor media exhibited fitness tradeoffs across a range of novel resource levels (Fig. 3a), as well as in response to novel salt abundances (Fig. 4), suggesting that both axes of salinity and resource abundances shaped niche evolution. Similar to rich-selected populations, however, the effects of salt on poor-selected populations is context-dependent (Fig. 4a). Altogether, bacterial niche evolution appears to be multidimensional (i.e. observed across resource and salt levels), often leading to the same associated tradeoffs across niche dimensions. As discussed above, selection among archaeal populations corresponded with fitness improvements along a resource gradient (Fig. 3b), but changes observed at varying salinities were not always in the same direction. For instance, both selected populations benefited from higher salt concentrations in a high-resource environment compared to the ancestor, indicating that salinity was not where selection was operating. Previous work has found that archaeal lineages tend to be better-suited for stressful conditions, such as high salinity, acidity, or temperatures (Van De Vossenberg et al. 1998; Reed et al. 2013; Saralov 2019). It is therefore possible that the primary selective pressure governing prior evolution of H. salinarum was chronic energy stress (Valentine 2007) and thus the driver in this experiment was the makeup of other resources in the media (consistent with Bartha 2022). Our findings also demonstrate that the pleiotropic consequences of selection may not be the same across niche dimensions, which largely deviates from the multidimensional shifts we measured for bacterial lines. This divergence in dimensionality underscores the complexity of microbial niche evolution, even for comparisons among highly convergent species. This complexity aligns with previous experimental work, which has largely found inconsistency in the emergence of tradeoffs among replicated populations (Rodríguez- Verdugo et al. 2014; Jerison et al. 2020). Here, we find clarity by leveraging a comparative 149 experimental approach, where we find that ancestral niche breadth appears to underlie instances of divergence versus parallelism during subsequent evolution. Documenting the predictive power of ancestral variability is only possible given this experimental design: inclusion of two taxa, selected in two environments, measured across multiple niche dimensions. Our results also highlight the utility in incorporating extremophilic species – particularly from the less-studied Archaea domain – into experimental studies of niche evolution. V. Conclusion The nature of adaptation across several environmental axes is central to understanding patterns of diversification. Recently, an increasing interest in life’s limits has led to the characterization of microbial communities in a variety of extreme conditions. Key to investigating adaptation within these environments, however, is the exploration of corresponding fitness tradeoffs and their effects on niche breadth. Here, we use experimental evolution to investigate pleiotropic consequences of selection in rich and poor environments for a halophilic archaeon (Halobacterium salinarum) and bacterium (Salinibacter ruber), illustrating their divergence in niche breadth shape but not width. In doing so, this study is the first to demonstrate rapid niche evolution of halophilic species across multiple niche dimensions. We find that the preexisting variability in fitness curves for each species not only shifts as a consequence of selection but also helps provide a predictive lens for anticipating future trends. 150 References 1. Grinnell, J. (1917). The niche-relationships of the California Thrasher. The Auk, 34(4), 427-433. 2. Elton, C. (1927). Animal ecology. London: Sidgwick and Jackson. 260 p. EltonAnimal Ecology1927. 3. Hutchinson, G. E. (1957). Concluding remarks. In: Cold Spring Harb Symp Quant Biol. 415-427. 4. MacArthur, R., & Levins, R. (1967). The limiting similarity, convergence, and divergence of coexisting species. The american naturalist, 101(921), 377-385. 5. Qiao, H., Saupe, E. E., Soberón, J., Peterson, A. T., & Myers, C. E. (2016). Impacts of niche breadth and dispersal ability on macroevolutionary patterns. The American Naturalist, 188(2), 149-162. 6. Carscadden, K. A., Emery, N. C., Arnillas, C. A., Cadotte, M. W., Afkhami, M. E., Gravel, D., ... & Wiens, J. J. (2020). Niche breadth: causes and consequences for ecology, evolution, and conservation. The Quarterly Review of Biology, 95(3), 179- 214. 7. Peterson, A. T., Soberón, J., Pearson, R. G., Anderson, R. P., Martínez-Meyer, E., Nakamura, M., & Araújo, M. B. (2011). Ecological niches and geographic distributions. In Ecological niches and geographic distributions. Princeton University Press. 8. Sexton, J. P., Montiel, J., Shay, J. E., Stephens, M. R., & Slatyer, R. A. (2017). Evolution of ecological niche breadth. Annual Review of Ecology, Evolution, and Systematics, 48, 183-206. 9. Kassen, R. (2024). Experimental evolution and the nature of biodiversity. Oxford University Press. 10. Lynch, M., & Gabriel, W. (1987). Environmental tolerance. The American Naturalist, 129(2), 283-303. 151 11. Bennett, A. F., & Lenski, R. E. (2007). An experimental test of evolutionary trade-offs during temperature adaptation. Proceedings of the National Academy of Sciences, 104(suppl_1), 8649-8654. 12. Levins, R. (1968). Evolution in changing environments: some theoretical explorations (No. 2). Princeton University Press. 13. Travisano, M., & Lenski, R. E. (1996). Long-term experimental evolution in Escherichia coli. IV. Targets of selection and the specificity of adaptation. Genetics, 143(1), 15-26. 14. Lang, G. I., Murray, A. W., & Botstein, D. (2009). The cost of gene expression underlies a fitness trade-off in yeast. Proceedings of the National Academy of Sciences, 106(14), 5755-5760. 15. Jerison, E. R., Nguyen Ba, A. N., Desai, M. M., & Kryazhimskiy, S. (2020). Chance and necessity in the pleiotropic consequences of adaptation for budding yeast. Nature Ecology & Evolution, 4(4), 601-611. 16. Fahimipour, A. K., & Gross, T. (2020). Mapping the bacterial metabolic niche space. Nature communications, 11(1), 4887. 17. Maeno, S., Nishimura, H., Tanizawa, Y., Dicks, L., Arita, M., & Endo, A. (2021). Unique niche-specific adaptation of fructophilic lactic acid bacteria and proposal of three Apilactobacillus species as novel members of the group. BMC microbiology, 21(1), 41. 18. Reznick, D. (1985). Costs of reproduction: an evaluation of the empirical evidence. Oikos, 257-267. 19. Ostrowski, E. A., Rozen, D. E., & Lenski, R. E. (2005). Pleiotropic effects of beneficial mutations in Escherichia coli. Evolution, 59(11), 2343-2352. 20. Lee, M. C., Chou, H. H., & Marx, C. J. (2009). Asymmetric, bimodal trade-offs during adaptation of Methylobacterium to distinct growth substrates. Evolution, 63(11), 2816- 2830. 152 21. Jasmin, J. N., & Zeyl, C. (2013). Evolution of pleiotropic costs in experimental populations. Journal of evolutionary biology, 26(6), 1363-1369. 22. Rodríguez-Verdugo, A., Carrillo-Cisneros, D., González-González, A., Gaut, B. S., & Bennett, A. F. (2014). Different tradeoffs result from alternate genetic adaptations to a common environment. Proceedings of the National Academy of Sciences, 111(33), 12121-12126. 23. Leiby, N., & Marx, C. J. (2014). Metabolic erosion primarily through mutation accumulation, and not tradeoffs, drives limited evolution of substrate specificity in Escherichia coli. PLoS biology, 12(2), e1001789. 24. Rampelotto, P. H. (2013). Extremophiles and extreme environments. Life, 3(3), 482- 485. 25. Rampelotto, P. H. (2024). Extremophiles and Extreme Environments: A Decade of Progress and Challenges. Life, 14(3), 382. 26. Medina-Chávez, N. O., & Travisano, M. (2022). Archaeal communities: the microbial phylogenomic frontier. Frontiers in Genetics, 12, 693193. 27. Tobler, M., Kelley, J. L., Plath, M., & Riesch, R. (2018). Extreme environments and the origins of biodiversity: Adaptation and speciation in sulphide spring fishes. Molecular ecology, 27(4), 843-859. 28. Jeyasingh, P. D., Sherman, R. E., Prater, C., Pulkkinen, K., & Ketola, T. (2023). Adaptation to a limiting element involves mitigation of multiple elemental imbalances. Journal of the Royal Society Interface, 20(198), 20220472. 29. Snell‐Rood, E., Cothran, R., Espeset, A., Jeyasingh, P., Hobbie, S., & Morehouse, N. I. (2015). Life‐history evolution in the anthropocene: Effects of increasing nutrients on traits and trade‐offs. Evolutionary applications, 8(7), 635-649. 30. Jeyasingh, P. D., & Weider, L. J. (2005). Phosphorus availability mediates plasticity in life‐history traits and predator–prey interactions in Daphnia. Ecology Letters, 8(10), 1021-1028. 153 31. He, Q., Wang, S., Feng, K., Michaletz, S. T., Hou, W., Zhang, W., ... & Deng, Y. (2023). High speciation rate of niche specialists in hot springs. The ISME Journal, 17(8), 1303-1314. 32. Vauclare, P., Natali, F., Kleman, J. P., Zaccai, G., & Franzetti, B. (2020). Surviving salt fluctuations: stress and recovery in Halobacterium salinarum, an extreme halophilic Archaeon. Scientific Reports, 10(1), 3298. 33. Oren, A. (2013). Salinibacter: an extremely halophilic bacterium with archaeal properties. FEMS microbiology letters, 342(1), 1-9. 34. Kunka, K. S., Griffith, J. M., Holdener, C., Bischof, K. M., Li, H., DasSarma, P., ... & Slonczewski, J. L. (2020). Acid experimental evolution of the haloarchaeon Halobacterium sp. NRC-1 selects mutations affecting arginine transport and catabolism. Frontiers in Microbiology, 11, 535. 35. Angert, A. L., Bontrager, M. G., & Ågren, J. (2020). What do we really know about adaptation at range edges?. Annual Review of Ecology, Evolution, and Systematics, 51(1), 341-361. 36. Caruso, C. M., Martin, R. A., Sletvold, N., Morrissey, M. B., Wade, M. J., Augustine, K. E., ... & Kingsolver, J. G. (2017). What are the environmental determinants of phenotypic selection? A meta-analysis of experimental studies. The American Naturalist, 190(3), 363-376. 37. Aranda-Díaz, A., Willis, L., Nguyen, T. H., Ho, P. Y., Vila, J., Thomsen, T., ... & Huang, K. C. (2025). Assembly of stool-derived bacterial communities follows “early- bird” resource utilization dynamics. Cell Systems, 16(4). 38. Antón, J., Oren, A., Benlloch, S., Rodríguez-Valera, F., Amann, R., & Rosselló-Mora, R. (2002). Salinibacter ruber gen. nov., sp. nov., a novel, extremely halophilic member of the Bacteria from saltern crystallizer ponds. International journal of systematic and evolutionary microbiology, 52(2), 485-491. 39. Bebber, D. P., & Chaloner, T. M. (2022). Specialists, generalists and the shape of the ecological niche in fungi. The New Phytologist, 234(2), 345. 154 40. Futuyma, D. J., & Moreno, G. (1988). The evolution of ecological specialization. Annual review of Ecology and Systematics, 207-233. 41. Whitlock, M. C. (1996). The red queen beats the jack-of-all-trades: the limitations on the evolution of phenotypic plasticity and niche breadth. The American Naturalist, 148, S65-S77. 42. Lee, M. C., Chou, H. H., & Marx, C. J. (2009). Asymmetric, bimodal trade-offs during adaptation of Methylobacterium to distinct growth substrates. Evolution, 63(11), 2816- 2830. 43. Bono, L. M., Smith Jr, L. B., Pfennig, D. W., & Burch, C. L. (2017). The emergence of performance trade‐offs during local adaptation: insights from experimental evolution. Molecular ecology, 26(7), 1720-1733. 44. Chavhan, Y., Malusare, S., & Dey, S. (2021). Interplay of population size and environmental fluctuations: A new explanation for fitness cost rarity in asexuals. Ecology Letters, 24(9), 1943-1954. 45. Lenski, R. E. (2017). Experimental evolution and the dynamics of adaptation and genome evolution in microbial populations. The ISME journal, 11(10), 2181-2194. 46. Lenski, R. E., & Travisano, M. (1994). Dynamics of adaptation and diversification: a 10,000-generation experiment with bacterial populations. Proceedings of the National Academy of Sciences, 91(15), 6808-6814. 47. Bartha, Emese (2022) Investigating extremotolerant microbes in non-extreme environments and altering the salinity growth limits of halophiles. PhD thesis, University of Essex. 48. Levins, R. (1968). Evolution in changing environments: some theoretical explorations (No. 2). Princeton University Press. 49. Valentine, D. L. (2007). Adaptations to energy stress dictate the ecology and evolution of the Archaea. Nature Reviews Microbiology, 5(4), 316-323. 155 50. Patil, Divya & Vinod, Bhoomika & r v, Kavitha. (2023). Optimization of medium and media components for maximum biomass of Halobacterium salinarum NRC-1 using Response Surface Methodology. Indian Journal of Applied Microbiology. 25. 1-10. 51. Stan-Lotter, H., & Fendrihan, S. (2015). Halophilic archaea: life with desiccation, radiation and oligotrophy over geological times. Life, 5(3), 1487-1496. 52. Sabet, S., Diallo, L., Hays, L., Jung, W., & Dillon, J. G. (2009). Characterization of halophiles isolated from solar salterns in Baja California, Mexico. Extremophiles, 13, 643-656. 53. von Meijenfeldt, F. B., Hogeweg, P., & Dutilh, B. E. (2023). A social niche breadth score reveals niche range strategies of generalists and specialists. Nature Ecology & Evolution, 7(5), 768-781. 54. Oren, A., Rodríguez-Valera, F., Antón, J., Benlloch, S., Rosselló-Mora, R., Amann, R., ... & Russell, N. J. (2004). Red, extremely halophilic, but not archaeal: the physiology and ecology of Salinibacter ruber, a bacterium isolated from saltern crystallizer ponds. In Halophilic microorganisms (pp. 63-76). Berlin, Heidelberg: Springer Berlin Heidelberg. 55. Mongodin, E. F., Nelson, K. E., Daugherty, S., Deboy, R. T., Wister, J., Khouri, H., ... & Rodriguez-Valera, F. (2005). The genome of Salinibacter ruber: convergence and gene exchange among hyperhalophilic bacteria and archaea. Proceedings of the National Academy of Sciences, 102(50), 18147-18152. 56. Schaum, C. E., & Collins, S. (2014). Plasticity predicts evolution in a marine alga. Proceedings of the Royal Society B: Biological Sciences, 281(1793), 20141486. 57. Ghalambor, C. K., Hoke, K. L., Ruell, E. W., Fischer, E. K., Reznick, D. N., & Hughes, K. A. (2015). Non-adaptive plasticity potentiates rapid adaptive evolution of gene expression in nature. Nature, 525(7569), 372-375. 58. Grinnell, J. (1917). The niche-relationships of the California Thrasher. The Auk, 34(4), 427-433. 59. Elton, C. S. (2001). Animal ecology. University of Chicago Press. 156 60. Hutchinson, G. E. (1957, January). Population studies-animal ecology and demography-concluding remarks. In Cold Spring Harbor symposia on quantitative biology (Vol. 22, pp. 415-427). 61. Blonder, B., Lamanna, C., Violle, C., & Enquist, B. J. (2014). The n‐dimensional hypervolume. Global Ecology and Biogeography, 23(5), 595-609. 62. Chase, J. M., & Leibold, M. A. (2003). Ecological niches: linking classical and contemporary approaches. University of Chicago Press. 63. Van De Vossenberg, J. L., Driessen, A. J., & Konings, W. N. (1998). The essence of being extremophilic: the role of the unique archaeal membrane lipids. Extremophiles, 2, 163-170. 64. Reed, C. J., Lewis, H., Trejo, E., Winston, V., & Evilia, C. (2013). Protein adaptations in archaeal extremophiles. Archaea, 2013(1), 373275. 65. Saralov, A. I. (2019). Adaptivity of archaeal and bacterial extremophiles. Microbiology, 88, 379-401. 66. Rodríguez-Verdugo, A., Carrillo-Cisneros, D., González-González, A., Gaut, B. S., & Bennett, A. F. (2014). Different tradeoffs result from alternate genetic adaptations to a common environment. Proceedings of the National Academy of Sciences, 111(33), 12121-12126. 157 Supplementary Materials Suppl. Table 1. Media composition for the “poor” and “rich” environments. Suppl. Fig. 1. Growth curves for S. ruber and H. salinarum in rich and poor media. Optical densities of ancestral and selected populations were measured daily for five days in 0 1 2 3 4 5 6 7 8 9 10 11 0.0 0.2 0.4 0.6 Time (Days) A bs or ba nc e (O D 60 0) Bacteria Ancestor Bacteria Selected in Poor Media Bacteria Selected in Rich Media 0 1 2 3 4 5 0 1 2 3 Time (Days) A bs or ba nc e (O D 60 0) Bacteria Ancestor Bacteria Selected in Rich Media Bacteria Selected in Poor Media 0 1 2 3 4 5 0 1 2 3 Time (Days) A bs or ba nc e (O D 60 0) Archaea Selected in Poor Media Archaea Ancestor Archaea Selected in Rich Media 0 1 2 3 4 5 6 7 8 9 10 11 0.0 0.2 0.4 0.6 Time (Days) A bs or ba nc e (O D 60 0) Archaea Ancestor Archaea Selected in Poor Media Archaea Selected in Rich Media B ac te ria A rc ha ea Rich Media Poor Media a) b) c) d) Components Poor environment (g/L) Rich environment (g/L) Salts NaCl 195 250 Other salts MgSO4 7H2O 49.5 10 KCl 5 5 CaCl2 2H2O 1.25 0.2 MgCl2 6H2O 34.6 - NaBr 0.625 - NaHCO3 0.25 - Resources Yeast Extract 1 10 Tryptone - 2.5 158 rich media for bacterial (a) and archaeal populations (c) and every 48 hours for eleven days in poor media for bacterial (b) and archaeal populations (d) (error bars ± SEM). The environment had a large effect on optical density, with populations grown in rich media yielding significantly higher densities compared to poor media for all treatments (p-value <0.0001, nested ANCOVA). In bacterial populations, selection in poor media resulted in significant growth tradeoffs in the non-selected environment, with populations selected in poor media reaching lower densities when grown in rich media (p-value <0.0001, ANCOVA) and populations selected in rich media having lower densities when grown in poor media (p-value <0.0001), compared to ancestors and other selected populations. Optical density of archaeal populations also differed significantly by treatment in rich (p- value = 0.0205) and poor (p-value <0.0001, ANCOVA) media but with different relative effects. Both selected populations grew faster in rich media, compared to ancestors, whereas archaeal populations selected in poor media reached lower densities than other populations even when grown in its selected environment. Overall, the effects of past selection differed significantly between bacterial and archaeal populations grown in rich (p-value <0.0001, nested ANCOVA) and poor (p-value <0.0001, nested ANCOVA) environments, indicating that the emergence of tradeoffs was not uniform across species. Suppl. Fig 2. Co-cultures of archaeal ancestors and populations selected in poor media compared to their predicted fitness in poor media. Co-cultures were inoculated 159 with 50% ancestral cultures and 50% derived cultures selected in poor media, along with monocultures of each (4 replicates) and grown for 5 days in poor media (error bars ± SEM). Predicted fitness assumes equal competitive fitness among the two populations (dotted line), hence deviation from the prediction indicates greater competitive fitness of one population compared to the other. Monocultures of H. salinarum selected in poor media were significantly different from the predicted fitness of co-cultures (p-value = 0.0214, one-way ANOVA, Suppl. Table 15) but not from the fitness of experimental co- cultures (p-value = 0.1764, one-way ANOVA, Suppl. Table 16), indicating a greater similarity between the cultures selected in poor media and the co-cultures, compared to ancestors. This result suggests that while selection in poor media appeared to decrease the absolute fitness of derived populations (see Fig. 3d), the competitive fitness of these populations did not decline and instead may have increased. Suppl. Fig. 3. Growth curves and fitness values for archaeal populations grown in poor media with varying resources. Optical density was measured daily in four populations of H. salinarum selected in rich media for five days in poor media containing an additional 9g of yeast extract (10g total) and 2.5g of tryptone (circles), an additional 9g of yeast extract but no tryptone (triangles), and only 1g of yeast extract but 2.5g of tryptone (squares). Absolute fitness values were approximated as the area under the curve, shown in the lefthand box (error bars ± SEM). Fitness differed significantly among 0 1 2 3 4 5 0 1 2 3 4 Time (Days) O pt ic al D en si ty (O D 60 0) Growth and fitness depend on resource abundance Poor Media (+) Tryptone Poor Media (+) 9g Yeast Extract (-) Tryptone Poor Media (+) 9g Yeast Extract (+) Tryptone (-) Yeast Extract (-) Tryptone 0 2 4 6 8 10 Archaea A bs ol ut e Fi tn es s 160 treatments (p-value < 0.0001, one-way ANOVA, Suppl. Table 17), with the biggest fitness difference observed between populations grown with 10g versus 1g of yeast extract. Tryptone alone was insufficient to achieve fitness levels comparable to those in rich media. Results indicate that the major growth differences observed between populations in rich and poor media can be attributed to the disparity in the amount of yeast extract present. Suppl. Table 2. Analysis of Variance of absolute fitness for all archaeal and bacterial populations (Domain) grown in rich and poor media (Media) (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Domain [Treatment] 3 8.14811 3.7811 0.0181 * Media 1 194.25707 270.4355 <0.0001 * Treatment 2 4.01903 2.7976 0.0735 Model 6 208.87498 34.8125 48.4643 <0.0001 * Error 38 27.29585 0.7183 C. Total 44 236.17083 Suppl. Table 3. Analysis of Variance of absolute fitness for bacterial populations grown in rich media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Treatment 2 22.919446 15.3002 0.0013 * Model 2 22.919446 11.4597 15.3002 0.0013 * Error 9 6.740934 0.7490 C. Total 11 29.660380 161 Suppl. Table 4. Analysis of Variance of absolute fitness for bacterial populations grown in poor media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Treatment 2 0.56609179 22.1219 0.0005 * Model 2 0.56609179 0.283046 22.1219 0.00050* Error 8 0.10235843 0.012795 C. Total 10 0.66845022 Suppl. Table 5. Analysis of Variance of absolute fitness for archaeal populations grown in rich media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Treatment 2 5.1582572 2.57913 6.9420 0.0179 * Model 2 5.1581572 2.57913 6.9420 0.0179 * Error 8 2.9721877 0.37152 C. Total 10 8.1304449 Suppl. Table 6. Analysis of Variance of absolute fitness for archaeal populations grown in poor media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Treatment 2 0.11846181 2.1928 0.1741 Model 2 0.11846181 0.059231 2.1928 0.1741 Error 8 0.21608891 0.027011 C. Total 10 0.33455072 162 Suppl. Table 7. Analysis of Variance of absolute fitness for all archaeal and bacterial populations (Domain) grown in rich media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Domain [Treatment] 3 19.581560 11.4239 0.0002 * Treatment 2 8.888750 7.7786 0.0040 * Model 5 28.481010 5.69620 9.9695 0.0001 * Error 17 8.713122 0.57136 C. Total 22 38.194132 Suppl. Table 8. Analysis of Variance of absolute fitness for all archaeal and bacterial populations (Domain) grown in poor media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Domain [Treatment] 3 0.84171432 14.0970 <0.0001 * Treatment 2 0.24680954 6.2003 0.0101 * Model 5 0.9866015 0.197320 9.9141 0.0002 * Error 16 0.3184473 0.019903 C. Total 21 1.3050488 Suppl. Table 9. Analysis of Covariance of population growth (OD600) over time for population selection history (Treatment) grown in rich and poor media (Media). Source DF Sum of Squares Mean Square F Ratio Prob > F 163 Media 1 73.679243 253.7066 <0.0001 * Domain[Treatment] 3 0.853146 0.9792 0.4031 Time 1 16.172400 55.6879 <0.0001 * Treatment 2 0.810714 1.3958 0.2496 Model 7 75.40407 10.7720 37.0923 <0.0001 * Error 245 71.15076 0.2904 C. Total 252 146.55482 Suppl. Table 10. Analysis of Covariance of population growth (OD600) over time for all bacterial populations grown in rich media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Time 1 29.373454 295.2866 <0.0001 * Treatment 2 5.604439 28.1703 <0.0001 * Treatment * Time 2 0.227611 1.1441 0.3263 Model 5 36.411424 7.28228 73.2076 <0.0001 * Error 53 5.272143 0.09947 C. Total 58 41.683567 Suppl. Table 11. Analysis of Covariance of population growth (OD600) over time for all bacterial populations grown in poor media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Time 1 0.71506953 104.1932 <0.0001 * Treatment 2 0.38252137 27.8687 <0.0001 * 164 Treatment * Time 2 0.05946033 4.3320 0.0171 * Model 5 1.1570512 0.232410 33.7189 <0.0001 * Error 66 0.4529528 0.006863 C. Total 71 1.6100040 Suppl. Table 12. Analysis of Covariance of population growth (OD600) over time for all archaeal populations grown in rich media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Time 1 36.825639 371.0370 <0.0001 * Treatment 2 0.835844 4.2108 0.0205 * Treatment * Time 2 0.591096 2.9778 0.0602 Model 5 38.175024 7.63500 76.9265 <0.0001 * Error 49 4.863279 0.09925 C. Total 54 43.038303 Suppl. Table 13. Analysis of Covariance of population growth (OD600) over time for all archaeal and bacterial populations (Domain) grown in poor media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Time 1 1.3860001 175.6575 <0.0001 * Treatment 2 0.0982474 6.2258 0.0026 * Domain[Treatment] 3 0.6883728 29.0808 <0.0001 * Model 6 2.1512789 0.358546 45.4411 <0.0001 * 165 Error 131 1.0336368 0.007890 C. Total 137 3.1849157 Suppl. Table 14. Analysis of Covariance of population growth (OD600) over time for all archaeal and bacterial populations (Domain) grown in rich media (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Time 1 68.222550 647.1385 <0.0001 * Treatment 2 2.225502 10.5552 <0.0001 * Domain[Treatment] 3 3.965152 12.5374 <0.0001 * Model 6 74.452732 12.4088 117.7060 <0.0001 * Error 108 11.3855562 0.1054 C. Total 114 85.838295 Suppl. Table 15. Analysis of Variance of optical densities (OD600) after 5 days of growth in poor media for archaeal populations selected in poor media compared to predicted fitness of co- cultures. Source DF Sum of Squares Mean Square F Ratio Prob > F Treatment 1 0.02252503 0.022525 9.5465 0.0214 * Error 6 0.01415702 0.002360 C. Total 7 0.03668205 166 Suppl. Table 16. Analysis of Variance of optical densities (OD600) after 5 days of growth in poor media for archaeal populations selected in poor media compared to fitness of experimental co-cultures. Source DF Sum of Squares Mean Square F Ratio Prob > F Treatment 1 0.01003236 0.010032 2.3477 0.1764 Error 6 0.02563958 0.004273 C. Total 7 0.03567194 Suppl. Table 17. Analysis of Variance of absolute fitness for archaeal populations grown for 5 days in three different media compositions. Source DF Sum of Squares Mean Square F Ratio Prob > F Media 2 138.09000 69.0450 1395.495 <0.0001 * Error 9 0.44529 0.0495 C. Total 11 138.53530 Suppl. Table 18. Analysis of Variance of absolute fitness bacterial populations grown for 5 days across a gradient of yeast extract abundances (Yeast Extract) (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Yeast Extract 5 243.84501 63.5382 <0.0001 * Treatment[Yeast Extract] 12 71.99707 7.8167 <0.0001 * 167 Model 17 315.84209 18.5789 24.2054 <0.0001 * Error 54 41.44789 0.7676 C. Total 71 357.28997 Suppl. Table 19. Analysis of Variance of absolute fitness for archaeal populations grown for 5 days across a gradient of yeast extract abundances (Yeast Extract) (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Yeast Extract 5 188.40614 371.7461 <0.0001 * Treatment[Yeast Extract] 12 65.20694 53.6085 <0.0001 * Model 17 253.61307 14.9184 147.1784 <0.0001 * Error 54 5.47359 0.1014 C. Total 71 259.08667 Suppl. Table 20. Analysis of Variance of relative fitness for all populations selected in poor media grown across a gradient of yeast extract abundances, for both species (Domain). Source DF Sum of Squares Mean Square F Ratio Prob > F Domain 1 2.4403693 2.44037 52.9365 <0.0001 * Error 46 2.1205958 0.04610 C. Total 47 4.5609651 168 Suppl. Table 21. Analysis of Variance of relative fitness for all populations selected in rich media grown across a gradient of yeast extract abundances, for both species (Domain). Source DF Sum of Squares Mean Square F Ratio Prob > F Domain 1 0.1274128 0.127413 2.5815 0.1150 Error 46 2.2703818 0.049356 C. Total 47 2.3977946 Suppl. Table 22. Analysis of Variance of absolute fitness for ancestral populations grown for 5 days across a gradient of yeast extract abundances (Yeast Extract), for both species (Domain). Source DF Sum of Squares Mean Square F Ratio Prob > F Domain[Yeast Extract] 6 29.702470 84.1052 <0.0001 * Yeast Extract 5 89.470955 304.0141 <0.0001 * Model 11 119.17343 10.8339 184.0638 <0.0001 * Error 36 2.11895 0.0589 C. Total 47 121.29238 Suppl. Table 23. Linear regression analysis of absolute fitness of archaeal ancestors grown across a gradient of yeast extract abundances (Yeast Extract). Coefficients Estimate Std. Error t Ratio Prob>|t| 169 (Intercept) 3.3556283 0.240797 13.94 <0.0001 * Yeast Extract 0.3145503 0.029076 10.82 <0.0001 * Yeast Extract * Yeast Extract -0.039633 0.011372 -3.49 0.0022 * Adj. R2:0.865323, df = 3 Suppl. Table 24. Linear regression analysis of absolute fitness of bacterial ancestors grown across a gradient of yeast extract abundances (Yeast Extract). Coefficients Estimate Std. Error t Ratio Prob>|t| (Intercept) 5.3880676 0.462827 11.64 <0.0001 * Yeast Extract 0.2946295 0.055885 5.27 <0.0001 * Yeast Extract * Yeast Extract -0.140325 0.021857 -6.42 <0.0001 * Adj. R2:0.784271, df = 3 Suppl. Table 25. Analysis of Variance of absolute fitness for all populations grown at varying abundances of tryptone and NaCl (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Treatment 5 0.684042 2.2118 0.0615 Tryptone[Treatment] 6 16.276527 43.8568 <0.0001 * Salt[Treatment] 6 33.080993 89.1362 <0.0001 * Tryptone*Salt 1 24.302548 392.8969 <0.0001 * 170 Model 18 74.344109 4.13023 66.7730 <0.0001 * Error 77 4.762817 0.06185 C. Total 95 79.106926 Suppl. Table 26. Analysis of Variance of relative fitness for all bacterial populations grown in two high-resource environments with varying NaCl abundances (Media) (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Treatment 1 0.27206543 12.7494 0.0038 * Media 1 0.14187660 6.6486 0.0242 * Treatment * Media 1 0.22424502 10.5085 0.0071 * Model 3 0.63818705 0.212729 9.9688 0.0014 * Error 12 0.25607280 0.021339 C. Total 15 0.89425985 Suppl. Table 27. Analysis of Variance of relative fitness for all archaeal populations grown in two high-resource environments with varying NaCl abundances (Media) (selection history denoted as Treatment). Source DF Sum of Squares Mean Square F Ratio Prob > F Treatment 1 0.0021226 0.0443 0.8375 Media 1 1.2750423 26.6186 0.0004 * 171 Treatment * Media 1 0.0519679 1.0849 0.3221 Model 3 1.3271964 0.442399 9.2358 0.0031 * Error 10 0.047900 0.047900 C. Total 13