Similarities in biological processes can be used to bridge ecology and molecular biology

Abstract Much of the research in biology aims to understand the origin of diversity. Naturally, ecological diversity was the first object of study, but we now have the necessary tools to probe diversity at molecular scales. The inherent differences in how we study diversity at different scales caused the disciplines of biology to be organized around these levels, from molecular biology to ecology. Here, we illustrate that there are key properties of each scale that emerge from the interactions of simpler components and that these properties are often shared across different levels of organization. This means that ideas from one level of organization can be an inspiration for novel hypotheses to study phenomena at another level. We illustrate this concept with examples of events at the molecular level that have analogs at the organismal or ecological level and vice versa. Through these examples, we illustrate that biological processes at different organization levels are governed by general rules. The study of the same phenomena at different scales could enrich our work through a multidisciplinary approach, which should be a staple in the training of future scientists.


| INTRODUC TI ON
We work on several aspects of evolution using a range of methods.
Among other things, we are interested in how genes are born (Durand et al., 2019;, how within-genome dynamics affect evolution (Eberlein et al., 2019;Leducq et al., 2016), how genomes are molded in parasitic relationships (Hébert, Grambauer, Barber, Landry, & Aubin-Horth, 2017), and how simple parts combine to form complex molecular systems (Diss, Ascencio, DeLuna, & Landry, 2014;Landry, Levy, Abd Rabbo, Tarassov, & Michnick, 2013). Because our team consists of people trained in different disciplines, from biophysics to biochemistry, microbiology, and ecology, we like to consider problems that span several levels of organization in terms of size, time, and space. Interestingly, the integration of these disciplines has helped us realize that living systems often show similar patterns regardless of the organization level, from the molecular to the organismal to the ecological. Using some examples from our own fields, we aim to exemplify how researchers in different fields and subfields can borrow concepts from each other to better understand and describe their own subject, and that these fields are not so different after all.
When a concept emerges in biology, or any other field, researchers may struggle to describe it using the language of that field. Since it is a new concept, the language of the field may fail to fully describe it or make it easily understandable. The description and understanding of a new concept can therefore be aided by borrowing language from other fields, where a similar concept exists and has been described. This was the case for Jacques Monod and François Jacob in 1961 when they introduced the term "program" to explain the lingering problem of how one single cell can develop into a complex organism through a seemingly static genome (Jacob & Monod, 1961).
They borrowed the term program from the booming field of computational science, which had used it as a description of a sequence of commands given to a computer (Jacob, 1993;Keller, 2002). This was a computer science concept, turned genetics metaphor, turned concept again as the study of genetics went on, and it showcases the power of cross-disciplinary knowledge to describe and to understand biological phenomena. However, carrying over metaphors from different fields could also oversimplify complex ideas, mask important details, and be confused with true descriptions; they should therefore be used consciously with these limitations in mind, and we should perhaps be open to updating our metaphors with time (Avise, 2001;Olson, Arroyo-Santos, & Vergara-Silva, 2019). We keep these limitations in mind when discussing the analogies we propose.
Within biology, comparing the different levels of organization can reveal common principles. In particular, some models applied to the macroscopic world can be applied to analogous microscopic phenomena (Francino, 2005). We can then use knowledge generated at one level to develop hypotheses that will inspire research at other levels. By doing so, we also argue that separating life-science disciplines by the scale of the phenomena that are studied (as is done in many research institutions including our own, where microbiology and biology are taught in different programs and departments) might no longer be justified. In fact, new disciplines emerge by combining knowledge from different fields, which once happened with the advent of biochemistry and more recently with bioinformatics or synthetic biology. Since this allows for a more complete view of a given phenomenon, the training of future scientists should convey the importance of communication and exchange between fields.
In this article, we want to emphasize the importance of integrative science. We discuss how there are recurrent themes at the molecular and organismal levels, from how simple properties can give birth to systems with complex properties, to how new genes are born and how conflicts can arise between species and within genomes. We have written it with senior undergraduate students in mind and hope to inspire aspiring scientists (and established ones) to pursue exciting work in multidisciplinary and integrative research.

| EMERG ING PROPERTIE S: HOW DO COMPLE X SYS TEMS EMERG E FROM S IMPLE PA RTS?
Biologists often study the collective behavior of organisms, cells, or molecules. These collective behaviors are fascinating because they display properties that are not observable from the study of a single organism, cell, or molecule. Such emerging properties refer to new functionalities that arise from interactions between the elements of a system that would not be found in any of the parts acting in isolation (Johnson, 2006). Properties can emerge in an initially disordered system when interactions between individual elements promote organized behavior. To illustrate this, one can think of phospholipids. These amphipathic molecules self-organize to form a bilayer membrane in which the hydrophilic regions are in contact with the surrounding solvent, and the hydrophobic regions stay in the center (van Meer, Voelker, & Feigenson, 2008). This structure allows compartmentalization and only appears in aqueous solutions and in the presence of more than one molecule. The plasma membranes of cells follow this pattern of self-assembly by forming a phospholipid bilayer that delimits the entire cell ( Figure 1a). The consequences of these simple interactions are very important and relevant even for the origin of life; this property of self-assembly of phospholipids is likely to have spontaneously given birth to the first protocells after they formed vesicles around nucleic acids and proteins (Deamer, 2017;Deamer & Barchfeld, 1982).
At larger molecular scales, macromolecules also show complex behaviors from relatively simple pairwise interactions. One of the simplest forms is a chain of nucleotides, which on their own have no enzymatic activity but when organized into polymers with specific sequences can acquire enzymatic properties and become ribozymes (Doudna & Cech, 2002). Similarly, amino acids organized into proteins can acquire enzymatic activities. At a higher level, individual proteins frequently form dimers and higher-order oligomers because they have interfaces that have affinity for each other (Marianayagam, Sunde, & Matthews, 2004). Together, the properties of oligomers confer important regulatory and functional advantages over just having independent individual proteins (Bergendahl & Marsh, 2017). Many ligand binding sites are formed at the interface between interacting copies of the same protein (Abrusán & Marsh, 2018;Figure 1b). An example is the STING protein, which participates in an immunity pathway conserved from anemones to humans. Its ligands bind to the interface between the individual copies of STING and stabilize the conformation that allows downstream signaling (Kranzusch et al., 2015). The monomers alone do not bind to the ligands and therefore cannot participate in the signaling pathway without first dimerizing. Thus, simple rules that dictate interactions of identical molecules can lead to complex and more efficient functions.
Emerging collective behavior also arises in microorganisms.
Quorum sensing is a signaling mechanism that allows bacteria to regulate gene expression and coordinate physiological processes in response to fluctuations in cell population density (Miller & Bassler, 2001). Once a certain threshold of cell density is achieved, each cell reacts to each other and new emergent properties arise in the F I G U R E 1 Emerging properties in systems with increasing complexity. (a) Free phospholipids self-organize in aqueous solutions to form a bilayer. The pairwise rules are that the hydrophilic regions are attracted to the solvent, while the hydrophobic regions are repelled by it. This allows an efficient structure with a new functionality by encapsulating different materials, for instance, those of a cell. (b) The activity of some enzymes depends on their self-assembly as dimers because their ligand binding site is located at the interface between the two copies. (c) Quorum sensing is a system that is activated when a certain threshold of cell density is achieved. Each cell reacts to each other, and this sum of interactions produces what is known as a biofilm, a new colonial form. An emergent property that arises in the colony can be, for example, a greater tolerance to antibiotics. (d) The V-shaped migratory flight formation enables aerodynamic interactions between individuals so that they can conserve energy on long migration routes. This confers superior flight ability to the entire group compared to single individuals Density increase, quorum sensing colony ( Figure 1c). This allows some bacteria, such as Bacillus subtilis (Cohn, 1872;Ehrenberg, 1835) Similarly, tolerance can arise by slowing down metabolism, because numerous antimicrobial drugs target metabolic processes that occur during growth (Flemming et al., 2016). Likewise, Pseudomonas aeruginosa (Migula, 1900;Schroeter, 1872) colonies produce virulence factors (Smith & Iglewski, 2003) and Vibrio harveyi (Baumann, Baumann, Bang, & Woolkalis, 1980;Johnson & Shunk, 1936)  Pairwise interactions can also lead to a complex organization in macroorganisms, following the same principles as for macromolecules and individual cells. For instance, the V formation of flying migratory birds follows a complex pattern (Figure 1d). The V-shape improves aerodynamics throughout the group since each bird interacts with its neighbors by creating an upward flow behind the tips of its wings, which gives those who follow it an additional elevation (Nathan & Barbosa, 2008) and saves some energy. Another example is schools of fish in which complex patterns of collective behavior emerge in part due to relatively simple pairwise interactions within the group, for instance for swimming speed adjustment (Katz, Tunstrøm, Ioannou, Huepe, & Couzin, 2011).

| EMERG ENT FE ATURE S OF NE T WORK S
Networks are mathematical objects that represent elements of a system as nodes and their pairwise connections as edges. The comparisons of such networks from different systems have helped identify key emergent properties that define, for instance, how they respond to external changes. Ecological interaction networks (or species interaction networks) are generated by monitoring the interactions (represented by "edges" of the network) between species (represented by "nodes" of the network) (Pocock, Evans, & Memmott, 2012). Interactions between species can be trophic (e.g., predator-prey interactions) or nontrophic (e.g., pollination).
By compiling the interactions between pairs of species, a global network can be created (see schematic in Figure  The degree distribution can be fitted to a power law equation (Figure 2b,c). Networks that follow the power law in this way are called scale-free networks. A peculiar property of the scale-free network is that the power law exponent (i.e., the value representing the slope between the log-scaled frequency of the nodes and corresponding degrees) usually lies between 2 and 3. Interestingly, in networks with such distributions, the removal or perturbation of a randomly selected node is unlikely to perturb the whole network.
Indeed, the removal of a highly connected node is usually required to severely perturb a scale-free network, but this is unlikely since hubs are not very common. These analyses revealed similar properties for networks that from the outset may seem to have no similarity (Barabási & Pósfai, 2016). This property provides inherent robustness to ecological and protein-protein interaction networks and consequently to the corresponding complex systems. From an evolutionary perspective, models suggest that the ways in which genes duplicate and change their interactions through time could be enough to produce such an organization (He & Zhang, 2005;Lynch & Force, 2000).
In order to understand the dynamics of ecological systems, network architecture provides an efficient medium because one can estimate how the removal of a node will likely affect other elements of the network (Delmas et al., 2018). Indeed, several studies show that network topologies (e.g., modularity, nestedness, connectedness) are associated with the robustness of ecosystems (Fortuna et al., 2010). In particular, the stability of ecological systems is of interest to ecologists who can use network-based approaches in their work (Landi, Minoarivelo, Brännström, Hui, & Dieckmann, 2018). For instance, in some habitats, plant species show high connectedness (Pocock, Evans, & Memmott, 2012). Therefore, the extinction or removal of a plant species in that habitat can presumably have a high impact on the robustness of the system by changing the network topology.
The topology of networks can also be used to understand the robustness of a cell. In some species, genes can be broadly categorized as "essential" or "nonessential" for cell survival. Gene essentiality often correlates with their connectedness in protein-protein interaction networks, in a way similar to the connectedness of important plant species mentioned above. Genes with functions that are essential for survival are more likely to be network hubs than nonessential genes (He & Zhang, 2006). Given the important implications of network architecture, there have been progressive attempts at genome-wide loss-of-function screenings of essential genes through RNAi (Liu et al., 2015) and CRISPR-Cas9 (Marceau et al., 2016) screens, which confirm that network architecture is a powerful way of representing the relationships among genes.
Overall, these examples illustrate that the robustness of complex systems at the ecological and molecular levels is partly attributable (or at least correlated) to the topological properties of their underlying networks, which themselves depend on rather simple pairwise interactions among their components. System properties therefore transcend levels of organization.

| E VOLUTI ON OF NOVELTIE S IN B IOLOG IC AL SYS TEMS
Biological networks are not static throughout evolution. They can change by adding new elements or removing existing ones. In the case of ecological networks, new species can be added by speciation events. One species can diverge into several new ones (Kocher, 2004;Meier et al., 2017), and different species can hybridize and create new species (Eberlein et al., 2019;Leducq et al., 2016;Mavárez et al., 2006;Soltis & Soltis, 2009). These principles present parallels for the emergence of functional proteins. New proteins most often arise from genes that have been duplicated (Aury et al., 2006;Bomblies & Madlung, 2014;Guan, Dunham, & Troyanskaya, 2007;Hakes, Pinney, Lovell, Oliver, & Robertson, 2007;Zhang, 2003). Although their functions may diverge from that of their ancestor, these new proteins often inherit properties from them, including their interactions with other proteins. Similarly, new species can inherit properties from their progenitors, like their trophic relationships with other species. In this section, we will discuss the processes by which novel elements appear at different organization levels.
Gene duplication produces an identical copy of a preexisting gene, usually through errors in DNA replication, cell division, unequal crossing over during recombination, or retrotransposition events (Otto & Whitton, 2000;Reams & Roth, 2015;Zhang, 2003).
The duplicates, or paralogs, are identical at first, but they diverge as they accumulate mutations. As a result, they can divide the ancestral gene's functions so that they can be separately optimized (Baker, Hanson-Smith, & Johnson, 2013;Force et al., 1999;Lambert, Cochran, Wilde, Olsen, & Cooper, 2015;Lynch & Conery, 2000)  The way gene duplication paves the way for new functions has some analogy with speciation by allopatry, that is, speciation as a consequence of physical separation (Figure 3; Mayr, 1942Mayr, , 1947. In this classical mechanism of speciation, a population is separated into two groups by a change in the environment, such as the appearance of a mountain range (Barrera-Guzmán, Milá, Sánchez-González, & Navarro-Sigüenza, 2012; Chaves & Smith, 2011;Gutiérrez-Pinto et al., 2012), or in more recent history, a highway (Thompson, Rieseberg, & Schluter, 2018). While the two groups would be very similar at first, the separation prevents gene flow. Through time, random mutations and recombination would lead to the appearance of different variants in each of the two groups. In fact, they could face selection on different traits, and develop different phenotypes (Barrett et al., 2019) or even become separate species by reproductive isolation, due to, for example, the incompatibility between the gene and protein networks of the two species (Dobzhansky, 1934;Mayr, 1942Mayr, , 1947Muller, 1942). Similarly, after gene duplication the two paralogs are identical at first, but can accumulate separate mutations (Force et al., 1999). Indeed, paralogs might face selection for different properties, as in the case of the freeze resistance proteins discussed above (Deng et al., 2010). One of the limitations of this analogy between speciation and gene duplication is that, while gene flow is limited between the allopatric populations, paralogs recombine with each other with varying frequencies (depending on the locus), and can sometimes transfer variants unilaterally through gene conversion (Chen, Cooper, Chuzhanova, Férec, & Patrinos, 2007;Dumont, 2015). Nevertheless, paralogs can reach a level of sequence divergence that allows them to minimize recombination (Harpak, Lan, Gao, & Pritchard, 2017), reminiscent of how species become reproductively isolated.
As is the case for new species, the relationships between paralogs This behavior can also be seen at the community level in ecological equivalents, that is, organisms that serve similar functions in different communities (Hubbell, 2006;Lincoln, 1998  However, despite their inherent similarity, not all paralogs are able to compensate for each other. In fact, some proteins need to interact with their paralog to function correctly (Baker et al., 2013;Boncoeur et al., 2012;Bridgham et al., 2008;DeLuna, Springer, Kirschner, & Kishony, 2010;Diss et al., 2017). Interestingly, similar relationships are also seen at an ecological level with closely related species that become mutualistic. This is the case for the cichlid species Etroplus maculatus (Bloch, 1795) and Etroplus suratensis (Bloch, 1790) that evolved a cleaning symbiosis (Wyman & Ward, 1972), the cichlid species Archocentrus nigrofasciatus (Günther, 1867) that benefits from Hypsophrys nicaraguensis (Günther, 1864) to avoid predation (Lehtonen, 2008), and the ant species Dolichoderus debilis (Emery, 1890) and Crematogaster levior (Longino, 2003) that share nests and forage together (Forel, 1898;Jarau & Hrncir, 2009). Although closely related, these duplicated genes or species cannot replace one another because they depend on each other. How these relationships evolved is still being investigated, but a necessary condition for evolving them is to maintain the interaction between the proteins or species. Research suggests that keeping similar traits increases the chance of maintaining the interactions. In the cichlids example, the cleaned species mimics a posture shown by adult members of the other species, which prompts the cleaning activity (Wyman & Ward, 1972). As for the ant species, they secrete similar molecules, which allow for the recognition of their ally and keep them from attacking each other (Menzel et al., 2013;Menzel, Linsenmair, & Blüthgen, 2008). Finally, a higher overall protein sequence similarity between paralogs favors the retention of their interactions with the same partners, including with one another (Ispolatov, Yuryev, Mazo, & Maslov, 2005 Just like speciation and gene duplication, phenotypic changes arise through the modification of preexisting material. Novel phenotypes allow organisms to fill unoccupied niches; for example, mammals have taken to both the sea and the sky over the course of their evolution. To our eyes in the present day, their innovation is apparent and can even be species defining. For example, bats' wings and whales' fins share ancestry with the limbs of quadrupedal animals (Cooper & Tabin, 2008;Teeling et al., 2005;Thewissen, Cooper, George, & Bajpai, 2009), but now they match the performance of other limbs that serve similar functions in their environments, such as bird wings and fish fins. In the genome, new genes can also emerge from unexpected places (Figure 3). In the first decade of this century, genes that had evolved from previously nongenic sequences were described (de novo genes) (Begun, Lindfors, Kern, & Jones, 2007;Levine, Jones, Kern, Lindfors, & Begun, 2006). This was a major shift in the field of novel genes, as de novo genes were deemed practically impossible until the 21st century (Tautz, 2014). Since then, de novo genes have been found in the genomes of many different organisms, from yeast (Cai, Zhao, Jiang, & Wang, 2008)  It can be difficult to understand how some of these phenotypes arose. What use is an eye before it is fully functional? What use is a wing if it cannot be used for flight? However, these are cases where rudimentary forms of the phenotype yield a selective advantage; eyes that cannot form images can still sense light differences (Lamb, Collin, & Pugh, 2007), and wings that cannot propel can still be used for a controlled descent (Kaplan, 2011). Having meaningful intermediate states is required for the evolution of organismal traits as it is for the evolution of protein sequences and function (Smith, 1970).
Unfortunately, the picture is not yet as clear with de novo genes; the mechanisms by which a new gene can emerge from noncoding regions are not fully understood. It is difficult to envision how a random DNA sequence that is expressed into a protein could have anything but a negative fitness effect. However, an elegant solution to this problem is found in the mounting evidence that a majority of the genome is transcribed, if at low levels (Clark, Amaral, Schlesinger, & Mattick, 2011;Havilio, Levanon, Lerman, Kupiec, & Eisenberg, 2005;Kapranov et al., 2002). In fact, it has been shown that large parts of the "noncoding" genome are translated at low levels (Ji, Song, Regev, & Struhl, 2015;Ruiz-Orera, Verdaguer-Grau, Villanueva-Cañas, Messeguer, & Mar Albà, 2018;Wilson & Masel, 2011). This pervasive transcription and translation allow intergenic regions to be molded by natural selection and increase their probability to gain a function in the cell (Wilson & Masel, 2011).

| CONFLI C TS IN CELLUL AR SYS TEMS AND IN THE ENVIRONMENT
Evolutionary novelties are often retained if they confer advantages to an organism. However, an advantage for one organism might result in a disadvantage for another one, which leads to conflict.
Conflicts may occur when the interests of two or more interacting parties drive their fitness in opposite directions (Queller, 2014).
Although the term conflict primarily derives from ecology, it has been extensively adopted in molecular biology to describe interactions between fragments of genetic material and biomolecules.
Interactions can be both between-species and within-species, and between-cells and within-cells. The only requirement for an element to trigger the conflict is to be able to replicate and to have heritable variation, which means that it applies to genes and organisms. One reason why conflicts deserve attention is that they can stimulate the evolution of complex molecular and organismal traits that would otherwise not have evolved. Indeed, a stable equilibrium is rarely established, which often leads to the constant evolution of traits, new adaptations, increased diversity, and eventually the evolution of more complex systems.

| INTER S PECIFIC CONFLIC T
A conflict between species typically includes interactions between prey and predator, or host and parasite. As formalized in the Red Queen hypothesis (Van Valen, 1973), the antagonistic pressures exerted by the interacting parties on each other accelerate their evolution, while retaining a stable fitness. Such conflict-driven evolution can manifest in different ways, for example, by improving the ability to recognize the opponent, by developing "weapons" to fight it, by confusing it through mimicry, and many others (Queller & Strassmann, 2018). Prey-predator dynamics can affect sensory systems, behavior, and life-history traits of both species to improve each other's recognition, as it is in the case of bats and moths; bats use an echolocation system to track moths, whereas moths have developed sensory organs to hear bat calls (Ter Hofstede & Ratcliffe, 2016). At the cellular level, in a conflict between a host and pathogen, proteins of different organisms evolve to recognize each other. Organisms have developed systems to detect foreign cells and molecules that depend on the binding abilities of macromolecules such as proteins.
Host proteins affected by pathogens ("host factors") determine the strength of infection based on the underlying network of interacting proteins. Pathogens usually target highly connected proteins, which leads to perturbation of many cellular functions and more severe consequences for an organism. Specifically, viruses target host proteins that can connect to any other protein in the network via a small number of intermediate interactions (Liu et al., 2015;Wuchty, 2014).
In this way, pathogens and parasites use the network properties of cell hosts to their advantage. Accelerated evolution caused by conflicts can be detected by the elevated rate of protein evolution (positive selection). This phenomenon has been observed at the binding sites of interacting proteins (McLaughlin & Malik, 2017). For instance, in mammals, two paralogs, TRIM5a and TRIM22, recognize retroviral proteins. Since retroviruses evolve rapidly, these proteins are expected to also evolve rapidly to maintain their ability to recognize viral proteins. Positive selection within the two domains of each protein responsible for antiviral specificity was shown to accelerate their evolution, revealing that the conflict may accelerate the divergence of paralogous proteins (Sawyer, Emerman, & Malik, 2007). Finally, human proteins that interact with viruses were shown to, on average, evolve under stronger positive selection than those that do not interact with viruses (Enard, Cai, Gwennap, & Petrov, 2016).
Mimicry is a strategy that escalates conflicts both at the molecular and organismal levels. A mimic evolves to resemble its opponent or opponent's components (e.g., eggs or proteins) and evade recognition. This is famously exemplified by the common cuckoo, Cuculus canorus (Linnaeus, 1758), which lays its eggs in other species' nests, removes host eggs, and parasitizes upon the parental care of the host species. The cuckoo's egg color and pattern evolve to match those of the host eggs, whereas the host bird sensory system evolves to better recognize parasitic eggs (Stoddard & Stevens, 2011). Similar coevolutionary patterns can be observed at a much smaller scale, for instance, between host cells and their pathogens. K3L is a rapidly evolving protein in poxviruses that mimics and competes with the primate protein eIF2alpha. elF2alpha is a component of the immune system that triggers an immune response when bound by the host's protein kinase R, suppressing virus growth (Elde, Child, Geballe, & Malik, 2009). The virus improves the resemblance of K3L to eIF2alpha in order for K3L to be bound by protein kinase R instead, abolishing the immune response. However, the host's protein kinase R in turn evolves to avoid binding to the virus protein while still binding to the conserved native one (Figure 4).
Interspecific conflict often leads to the evolution of new adaptations, such as weapons, behaviors, or protective structures. Toxin and antitoxin production is a common strategy of defense or attack, from vertebrates (Hanifin, Brodie, & Brodie, 2008) to insects and microbes (Pedrini et al., 2015). Parasites usually evolve refined adaptations to lure and infect their host and complete their reproductive cycle, which includes changing the behavior of the attacked host (de Bekker et al., 2014;Dubey, 2014). Fungal pathogens, for instance, present a large repertoire of shields, weapons, structures, and strategies during infection to avoid phagocytosis (Erwig & Gow, 2016).

| INTR ACELLUL AR CONFLI C TS
Conflicts do not only exist between cells and organisms, but also among the genes encoded in the genome of a single cell. Just like interactions between species are central to the definition of an ecosystem (Willis, 1997), interactions between genes in a genome are essential for the function of the cell (Tarassov et al., 2008).
While there is little doubt that a large proportion of these interactions evolved because they were adaptive, some exist only to cope with genetic conflicts. Conflicts among genes often arise because the inheritance mode of some genes violates one or several of the principles of Mendelian inheritance (Werren, 2011). This raises the possibility that such a gene would be inherited more often than expected and thus have higher fitness. Natural selection would act on the gene itself to favor its spread, which could be associated with a decrease in the fitness of the host. As a result, the host and the selfish gene will initiate an evolutionary arms race, analogous to those occurring at the intercellular level.
The selfish spread of a gene can be compared to the spread of invasive species in an ecosystem, which is largely disconnected from the fitness of the other species in the ecosystem (Figure 5a) and may eventually jeopardize the maintenance of the ecosystem itself. Models of ecological networks that integrate trophic interactions between species predicted that some properties of species, like generalism, can predict their chances of success in an invasion (Romanuk et al., 2009).
Thus, non-Mendelian inheritance is a key property within a genome that predicts the selfish spread of a gene, while ecological properties can predict species invasiveness at the ecosystem level. Conflicts impose a selective pressure on "host" components (genes in a genome or species in an ecosystem) to evolve mechanisms that limit the spread of the "selfish" genes or species. Adaptive evolution to invaders within ecosystems can involve various classes of traits, spanning morphological, physiological, behavioral, and life-history traits (Strauss, Lau, & Carroll, 2006). Similarly, selfish genetic elements pressure host genes to evolve repression mechanisms, as exemplified below. In this way, one can imagine both genomes and ecosystems as coherent systems harboring some interactions that may only serve to buffer the detrimental effects caused by their cheating components.
One type of non-Mendelian transmission pattern that can lead to conflict is when chromosomes do not segregate in equal ratios to the gametes during meiosis (Werren, 2011). One of several mechanisms that can cause this relies on the asymmetry of female meiosis. Only one out of four meiotic products becomes a gamete during female meiosis, enabling the rise of selfish chromosomes that increase their odds of being transmitted to the oocyte (Henikoff, Ahmad, & Malik, 2001). Chromosomes that exhibit such a segregation distortion exploit the meiotic spindle machinery to favor their inheritance to the oocyte (Akera et al., 2017;Hewitt, 1976). This distortion motivated the elegant centromere drive hypothesis (Henikoff et al., 2001), which posits that a conflict exists between female and male meiosis. The evolution of "stronger" centromeres in asymmetrical female meiosis would be deleterious in male meiosis, which is symmetrical and might suffer more from nondisjunction. Empirical evidence of the unexpectedly fast and adaptive evolution of centromeric DNA repeats and histone proteins, respectively (Kursel & Malik, 2018), are remarkably consistent with an arms race resulting from this conflict.
Centromere drive is one among many types of meiotic drive.
Another type is caused by killer meiotic drivers, which selectively eliminate meiotic products that do not harbor the drive element or that contain the target locus conferring killer sensitivity (Bravo Núñez, Nuckolls, & Zanders, 2018). Spore killing in some ascomycete fungi (Turner & Perkins, 1991)

and segregation distorters in
Drosophila (Larracuente & Presgraves, 2012) are examples of drive systems in which gametes harboring the killer locus eliminate sensitive gametes before they can develop or fertilize other ones. These examples of male meiotic drive bear a certain analogy with siblicide.
As an extreme form of sibling conflict, siblicide has been documented in many species (Furness, Morrison, Orr, Arendt, & Reznick, 2015). For instance, in many bird species, individuals within a brood can cause the death of their siblings, either because of aggressive behavior or because of competition for food (Mock, 1984;Redondo, Romero, Díaz-Delgado, & Nagy, 2019).

F I G U R E 4
Mimicry evolving from interspecific conflict has similar features at different levels of organization. (a) Birds do not experience fitness costs if they are able to distinguish their own eggs (top row of eggs) from the parasitic eggs of cuckoo (bottom row of eggs). Cuckoo eggs evolve to mimic the host egg, making birds unable to distinguish and remove the parasitic eggs. In turn, hosts evolve improved recognition of mimic eggs. (b) Host's protein kinase R (PKR) triggers an immune response in primates upon binding to eIF2alpha (top row), causing cell growth arrest, and limiting production of virus particles. Virus protein K3L mimics eIF2alpha and binds to PKR, blocking its function. The binding surface of PKR evolves to recognize the virus while at the same time retaining the ability to detect the conserved structure of eIF2alpha. In response, virus proteins evolve to better mimic eIF2alpha. Figure  Another type of deviation from Mendelian transmission is the spread of transposable elements (TEs) within genomes. If we consider the completion of a single cell cycle through a mitotic division, each canonical gene on each chromosome of a given genome is replicated exactly once, and both resulting cells inherit one copy of this gene. In contrast, TEs can replicate more than once within a single cell cycle, with copies of themselves invading new loci on chromosomes (McClintock, 1950). It soon appeared clear that like all mutation types, TEs mostly have deleterious effects upon insertion into the host genome, as illustrated by the classic example of TEs contributing to hybrid dysgenesis in Drosophila (Rubin, Kidwell, & Bingham, 1982). Population genetic models also support a scenario in which selection against transposition is the main determinant of TE frequency patterns in natural populations (Charlesworth, Sniegowski, & Stephan, 1994).
TEs will often encode genes that are necessary but not sufficient for their own replication. For example, retrotransposons encode genes that are essential for their own reverse transcription and integration into the host genome (Wicker et al., 2007).
However, the initial production of retrotransposon mRNAs relies on the host transcription machinery. Thus, a conflict can arise from the host's need to express its own genes and the retrotransposon's need to replicate itself. Starting from an equilibrium state between retrotransposon expansion and repression from the host, we can imagine that a retrotransposon mutant favoring its own replication would pressure host genes to evolve stronger repression in response to the deleterious effects of a retrotransposition expansion. In humans, this hypothetical scenario is illustrated by the conflict between some retrotransposon subfamilies and KRAB zinc finger proteins that bind dsDNA and recruit a transcriptional repressor complex. It was shown that the human zinc finger protein ZNF91 is efficient in silencing SVA retrotransposons, while a reconstructed ancestral ZNF91 gene sequence is not (Jacobs et al., 2014) (Figure 5b, left). This repression likely imposes a cost on the host, as it also abrogates the transcription of host genes in the vicinity of SVA insertions. TE repression imposing a cost F I G U R E 5 Conflicts within genomes and ecosystems. (a) Conflict between the host genome and selfish genetic elements (SGEs, left) or between native and invader species (right) can be schematically described by similar fitness functions. In both cases, the "host" (black) is advantaged when the "invader" (grey) is at low frequency. Inversely, the "invader" gains fitness with higher frequency. In both cases, we can imagine that very high SGE copy numbers or large invader population sizes can drive the extinction of the host species or disruption of the native ecosystem, respectively. Figure inspired by Queller and Strassmann (2018). (b) The conflict between retrotransposons and host transcription repressors in hominids led to mutations (star and triangle symbols) accumulating in a manner consistent with an arms race. Left: The ZNF91 KRAB zinc finger protein acquired mutations in the common ancestor of gorilla, chimp, and humans that allow it to repress the activity of the SVA retrotransposon. Right: Subfamilies of L1 retrotransposons acquired a mutation that allows them to escape repression mediated by ZNF93 by repressing neighboring host genes is also hypothesized in the case of TE methylation in Arabidopsis, as regions near genes are deprived in methylated TEs (Ahmed, Sarazin, Bowler, Colot, & Quesneville, 2011). Additionally, the repressing ability of the related human zinc finger protein ZNF93 against a subfamily of L1 retrotransposons was counteracted by a deletion in the ZNF93 binding site of the L1 sequence (Jacobs et al., 2014) (Figure 5b, right). These lines of evidence support the hypothesis of an arms race taking place between KRAB zinc finger proteins and retrotransposons in hominid evolution.
In summary, arms races and conflicts are common themes that emerge from the comparison of selfish components within genomes and ecosystems. Thus, it can be expected that these conflicts lead to the emergence of complex systems that cannot be explained by adaptive evolution of host features alone, but rather by compensatory evolution in the face of cheaters.

| DISCUSS ION/CON CLUS ION
Molecular biology and ecology have analogous properties that can be exploited in teaching and research. We expanded on these ideas by discussing how molecular phenomena closely resemble what is seen at the organismal level. These parallels call for deeper integration of the fields since models from one field can inspire models for analogous processes in the other. For example, networks can be used to represent interactions between molecules and between species equally well; new functions and species can emerge from modification of the existing ones; and arms races have been described at the species and molecular levels. While we focused on these examples that are close to our fields of research, there are potentially many more.
Having concepts carry over between fields allows models to be transferable. As described above, networks have been used at both the molecular and ecological levels. The well-established framework of networks has contributed to the advancement of both fields to solve analogous problems, such as the identification of genes involved in diseases (Yu, Kim, Xiao, & Hwang, 2013) and species critical to ecosystem conservation (Harvey, Gounand, Ward, & Altermatt, 2017) based on their positions and connectedness in networks. These models of conservation benefit from the comparison and the principles of stoichiometry. Proteins that are highly connected in a network are also highly abundant (Saeed & Deane, 2006), just like the optimal population sizes of a given species in an ecosystem depend on their specific roles and their interactions with other species (Harvey et al., 2017;Sterner & Elser, 2002). Elsewhere, the study of gene duplication has benefitted from comparisons to speciation. For instance, models on how several species emerge from an ancestral one (adaptive radiation) have been used to describe how gene families emerge by gene duplication. While this model's predictions match well the evolution of olfactory receptor families (Francino, 2005), further research could provide more information on how new functions are acquired through gene duplication. Ultimately, these cases show how our understanding of specific processes can benefit from models based on analogous phenomena.
Traditionally, molecular biology and ecology are taught separately as if they were somehow disconnected. Considering the similarities between molecular biology and ecology, a case can be made for a more integrative approach in teaching. This kind of organization could potentially lead to advancement of both fields by bringing students to think about the overarching properties of biological systems. We believe that students' curricula and activities could be modified to facilitate better communication between these fields and to focus on commonalities among systems rather than on the differences. From a pedagogical point of view, narratives from one level can be borrowed to help teach key concepts at other levels, depending on the training of the people involved.
As a result, future professionals and researchers would be able to come up with original ideas to study and understand nature better as a whole.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this article as no new data were created or analyzed in this study.