A fundamental question in biology is how the connections between populations of organisms, particularly of the same species, impact their ecology and evolution. However, the nature of such connections, or “connectivity,” is characterized differently depending on the subdiscipline or research question (Lowe and Allendorf, 2010; Kool et al., 2013; Keeley et al., 2022). Ambiguity in definitions can lead to misunderstandings between scientists attempting to work across subdisciplines or manage different components of an ecosystem. Yet, integrative work is essential to the current practice of ecology and evolutionary biology (Habel et al., 2015), as well as the successful management of ecosystems under rapid environmental change (Vos et al., 2008). The nature of connections among populations and species have important implications for their present persistence and future adaptations. We present a unifying framework for the study of population connectivity in ecology and evolutionary biology to facilitate communication and collaboration between subdisciplines without requiring subdisciplines to adopt a universal vocabulary.

At its most basic level, connectivity describes the relationship between populations or groups of organisms. The degree of connectivity is “the outcome of dependencies between populations or individuals” (Kool et al., 2013). Broadly, connectivity might be defined as the movement of individuals between groups or species (UN CMS, 2020), the flow of genes between groups (Habel et al., 2015), or even indirect interactions between groups across a landscape (Sheaves, 2009). Population connectivity concerns the first two definitions – the flow of organisms or genes. For specific research questions, all of these connectivity definitions describe real phenomena and are therefore useful. However, inconsistent definitions pose problems when researchers attempt to collaborate or synthesize information across different fields (Lowe and Allendorf, 2010; Kool et al., 2013).

When attempting to work across disciplines or with new sources of data, inconsistent definitions create misunderstandings and unnecessary friction. The process of collaboration is “laden with semantic issues that at best slow down the process, and at worst, exclude some from the conversation” (Pennington, 2008). For example, many stock assessments assume that fish stocks (groups of fish captured within an explicit geographic area) are discrete, spatially circumscribed units exhibiting internally homogenous dynamics (Secor, 2005). Mismatch between the management definition of a stock and the ecological, demographic, and, especially, the genetic relationship among groups of fish caught within that area can result in suboptimal management (Lipcius et al., 2008; Cadrin et al., 2014; Spies and Punt, 2015; Kerr et al., 2017; Harrison et al., 2020). Similarly, disagreements surrounding the definition of “species” have led to conflict over listing species under the United States Endangered Species Act (Waples, 1991; Crifasi, 2007; Carolan, 2008; Levin et al., 2021). Outside of management applications, collaborations across disciplines are frequently characterized by circular conversations about terminology disagreements. This situation is particularly exacerbated when fields are highly interconnected with overlapping terminology, as is the case in the subdisciplines of ecology and evolutionary biology (Dewulf et al., 2007; Habel et al., 2015).

Here, we address the confusion resulting from inconsistent definitions of population connectivity by developing a single unifying framework. Rather than focusing on generating specific terminology or definitions, we instead focus on identifying the inferences we gain about connectivity when we view it from the different timescales used in various fields. By focusing on the inference rather than the terminology, we hope to facilitate collaborations and data integration efforts across subdisciplines. First, we summarize current definitions of population connectivity and provide a glossary of relevant terms. Next, we introduce our population connectivity framework, which focuses on the relationships between different measures of population connectivity and time, and highlight the specific inferences provided by each. Finally, we describe an example application, applying the connectivity framework to summer flounder, Paralichthys dentatus, to contextualize our framework.

Though it is difficult to pinpoint the exact origin, the idea of connectivity between populations likely arose in patch dynamics theory beginning in the 1970s (e.g., Levins, 1969; Reddingius and den Boer, 1970; Levin and Paine, 1974; Levin, 1976; Roff, 1974). These early population models (e.g., Levins, 1969) recognized the interplay between migration and extinction in connected heterogenous landscapes, but migration was explicitly tied to the movement of organisms. Merriam (1984) introduced the concept of “landscape connectivity,” which noted that movement among habitat patches was not merely a function of an organism’s attributes, but also the quality of the landscape through which they moved. Formative debates within the field of landscape ecology concerned whether connectivity exists at the habitat patch-level definition, or whether connectivity is an emergent property at the landscape scale (Merriam, 1984; Taylor et al., 1993; Tischendorf and Fahrig, 2000; Moilanen and Hanski, 2001; Moilanen and Nieminen, 2002).

As early as the mid-20^th century, geneticists described populations in terms of “allelic divergence” or “gene frequency” differences to better understand the evolutionary processes at play in a system (Wright, 1931, 1943; Speith, 1974; Allendorf and Phelps, 1981). Wright’s (1931, 1949) foundational ideas about the genetics of isolated populations are arguably an antecedent of metapopulation theory (Hastings and Harrison, 1994; Harding and McNamara, 2002), where populations are conceptualized as “populations of populations” or collections of population groups that interact via immigration and emigration (Levins, 1969). Interactions among metapopulations have implications for population persistence and evolution (Hanksi, 1994; Hanski and Gaggiotti, 2004; Hanski, 2011). During this period in evolutionary biology, the presumed movement of individuals between populations and their subsequent contribution to reproduction was often described as “gene flow” and quantified through F-statistics or their derivatives (see Glossary or Section 2.2). Recognizing the importance of the landscape for gene flow, some studies began to use “connectivity” to describe how physical barriers to dispersal led to genetic isolation and drift (Soulé and Simberloff, 1986; Simberloff and Cox, 1987; Green, 1994; Hastings and Harrison, 1994). Attempts to understand the scale of physical dispersal and biogeography readily took advantage of novel genetic tools for measuring genetic divergence (Hedgecock et al., 1988). For those using a genetics toolbox, the term “connectivity” came to describe past or present gene flow across space (Benzie and Ballment, 1994; Mills and Allendorf, 1996; Roberts, 1997).

With these conflicting terms, it became important to distinguish between different types of population connectivity (Waples and Gaggiotti, 2006; Lowe and Allendorf, 2010). Yet, delineating what constitutes a population is not always straightforward (Funk et al., 2005) and determining the “connectivity” of poorly defined units can be even more complicated. For example, estimates of connectivity using genetics have been used to design marine reserves and designate management units (Roberts, 1997; Palsbøll et al., 2007; Jenkins and Stevens, 2018). In many cases, landscape attributes or physical-biological models of dispersal correlated well with early genetic studies of marine fish species where high dispersal capacity manifested in high gene flow over large spatial scales (Rosenblatt and Waples, 1986; Shulman and Bermingham, 1995). Studies like these support the notion that “open” populations with high immigration and emigration rates can be genetically well-mixed. However, biological oceanographers using coupled biophysical models and fish behavior studies have found many fish species to be comprised of highly structured populations, even in the face of high dispersal (Cowen et al., 2000, 2006). As genetic tools were further refined, it was possible to detect genetic structure at fine scales, despite high dispersal capacity (Doherty et al., 1995; Shulman and Bermingham, 1995; Ruzzante et al., 2006). These studies revealed that populations may be more “closed” to demographic exchange than previously thought. These debates on whether populations are fundamentally “open” or “closed” indicate a rift between the gene flow and movement-based concepts of connectivity.

Attempts have been made to generate clear distinctions between conceptions of connectivity (famously Lowe and Allendorf, 2010), as genetic data alone provides little information on organism movements across a landscape (Lowe and Allendorf (2010) term this “demographic connectivity”). Tischendorf and Fahrig (2000) examined the usage of the term “landscape connectivity” in 33 different studies and identified three major areas of confusion: (1) functional connectivity based on organism movements vs. structural connectivity based on habitat contiguity, (2) patch isolation vs. landscape connectivity, and (3) corridors vs. connectivity. Calabrese and Fagan (2004) critiqued the “inconsistently defined concept” of connectivity and developed a framework for classifying connectivity metrics according to data requirements, spatial scales and information yield. Waples and Gaggiotti (2006) reviewed commonly used definitions of “population,” differentiating between the ecological paradigm, which emphasizes demographic cohesion, and the evolutionary paradigm, which emphasizes reproductive cohesion. Despite these attempts, the use of the ambiguous term “connectivity” in both ecological and genetic studies continues apace: As of the writing of this manuscript there are 710,000 studies with “ecology” and “connectivity” as keywords, 127,000 studies with “genetic connectivity” or “genomic connectivity,” and only 2,610 using “demographic connectivity” (via Google Scholar).

As there are already many definitions and multiple, previous efforts to generate a universal vocabulary, we instead focus this manuscript on distinguishing the different concepts of population connectivity by their underlying processes–ecology or evolution. For our purposes, we have divided conceptions of population connectivity into two groups: ecological population connectivity (hereafter “ecological connectivity”) and evolutionary population connectivity (hereafter “evolutionary connectivity”). The specific terms themselves are not important – indeed, evolutionary connectivity is synonymous with genetic connectivity in most senses. What is important is the process of interest and the questions motivating researchers. Below we outline ecological and evolutionary conceptions of connectivity by first describing the fields, questions, and techniques used to study connectivity, and we then show how these conceptions have essential differences related to time.

Given the different ways of looking at population connectivity (i.e., ecologically and evolutionarily), how can we conceptualize the various connections among populations? Is there a framework we can use across subdisciplines which does not necessitate a universal vocabulary but can help synthesize inferences gained from different approaches? As described above, over the course of a lifetime, organisms will interact directly or indirectly with other individuals in a multitude of ways, from simply sharing space to consuming each other. Organisms may also interact through reproduction, thus passing genetic information across generations and maintaining gene flow between their populations. While the ecological or evolutionary processes resulting in population connectivity are continuous, we propose that focusing on the timescale of the interaction will result in more fruitful collaborations when working across biological subdisciplines. Therefore, we conceptualize population connectivity across discrete generational timescales, where the timescales together represent a time continuum. Just as a ripple propagates across a body of water from its point of origin, we can understand how biological interactions impact population connectivity within and across generations when we use this generational framework (Figure 1). Observations of the ripple near its point of origin encompass short generational timescales and ecological connectivity whereas observations of the ripple further away from its origin represent longer generational timescales and evolutionary connectivity. By focusing on a particular generational timescale, researchers may make inferences about specific interactions or processes, such as competition or reproduction, but observations over greater and greater generational timescales (and space in the case of a ripple) will result in a more holistic understanding of population connectivity.

For example, the northern elephant seal (Mirounga angustirostris) population experienced an enormous decline along the eastern Pacific: less than 20 individuals remained by the late 1800s (Hoelzel et al., 1993). Determining the connectivity of this population today is dependent on the timescale of interest. From an evolutionary perspective, all elephant seals along the west coast of North America stem from one single, bottlenecked group. Over 100 years later, genetic estimates show there is essentially no genetic variability (Fst ~ 0) within the population (Abadía-Cardoso et al., 2017). Elephant seals tend to return to their natal beaches, but their long generation time (~8 years; estimate from LeBoeuf and Reiter, 1988) means there has not been enough time for any significant genetic divergence between them on a population scale. Genetic markers which focus on specific subpopulations might be able to distinguish between natal beaches, but differences between subpopulations are small (Abadía-Cardoso et al., 2017; as is the case with the blue and yellow seals on the left in Figure 1 despite the population as a whole being a single connected unit). Ecologically, however, the interactions that individual elephant seals have with conspecifics and other species vary depending on their natal beach location, as well as sex and age. For example, while adult elephant seals tend to remain near their breeding shore, juvenile elephant seals instead wander broadly, hauling out on beaches up and down the coast (this may be revealed by tagging data, as shown on the blue elephant seal in the center of Figure 1). The interactions between different elephant seal subpopulations therefore depend on when a researcher looks for a connection; competitive interactions, disease transmission, etc. could all happen between natal beaches during the juvenile phase, even if the adults remain largely isolated. The amount of “population connectivity” therefore depends on the timescale of interest (within or across generations) and the underlying process (ecological or evolutionary).

In our generational framework, connectivity can be subdivided into three timescales: (1) connectivity within a generation (center of Figure 1), (2) connectivity between parent and offspring (across one parent–offspring generation, right side of Figure 1), and (3) multigenerational connectivity (2+ generations, left side of Figure 1). Connectivity within a generation has ecological consequences (e.g., resource competition, habitat use, foraging, etc.) and may have implications for gene flow and evolutionary connectivity. Connectivity between parent and offspring can have ecological or evolutionary consequences. Finally, connectivity on longer timescales can have evolutionary consequences, which may ultimately have impacts on ecological processes (eco-evolutionary feedbacks).

While the degree to which populations are connected across space and time may change, connectivity at the shortest timescales (ecological scales) radiates outward to influence connectivity at longer timescales (evolutionary scales). Thus, population connectivity along the timescale continuum is nested, in that it both depends upon previous timesteps and may influence subsequent timesteps. But measures of connectivity at different timescales are not perfect reflections of each other. Populations can be both ecologically and evolutionarily connected at a given moment. Evolutionarily connected populations must have had some ecological connectivity in the past, but exactly when that occurred will vary. Ecologically connected populations of related species may have once interbred, but due to reduced gene flow and eventual speciation, may now be connected only through competition. Ecological connectivity occurs on a continuum, as individuals of a species persist throughout time. In contrast, evolutionary connectivity occurs at discrete time points; once gene flow occurs, the level of evolutionary connectivity cannot change until the next mating event. Direct observations of ecological connectivity are bound to the present while evidence of evolutionary connectivity can persist through time.

These three timescales of connectivity also capture how connectivity is measured. For instance, occurrence tags track individuals over part or all of their lives, and thus can only directly capture connectivity at the timescale of less than one generation – ecological connectivity. Parent–offspring connectivity (ecological and evolutionary connectivity) can be inferred from genetic parentage analysis, where genetic tags are characterized in the parents and “recovered” in the offspring. Parent–offspring relationships can also be deduced based on distinct visual markings and repeat observations (Bain, 1990). Both of these methods track connectivity across one generation of parent and offspring, thus functioning as “intergenerational tags.” Multigenerational connectivity is tackled by the broad field of population genetics, which provides insight into connectivity patterns on longer timescales than parent–offspring connectivity (i.e., 2+ generations, evolutionary connectivity). Assuming adequate sampling, each approach provides inferences about aspects of connectivity at different timescales, resulting in a more complete picture of connectivity.

This generational framework for population connectivity clarifies the limitations of different sampling approaches. Any method which samples within a generation makes several assumptions about connectivity across generations. For example, in addition to the common assumption that tagged individuals are reflective of an entire population, tracking studies often also assume that present behavior reflects past behavior. Similarly, sampling methods which target past generations (i.e., population genetic methods) are used to hypothesize on ecological connections that cause the observed patterns, but they may differ dramatically from present patterns. For example, is a genetic bottleneck a result of population reduction due to disease or physical separation from a larger population (i.e., a founder effect)? To understand connectivity from a more holistic perspective, multiple sampling approaches must be employed.

This generational framework allows us to ask synthetic questions about the causes and consequences of connectivity. By thinking of connectivity generationally, we can ask how specific ecological connectivity processes impact the mating pool of a single generation (Figure 2). Across generations, this mating pool effect has implications for the subsequent (i.e., F1 and following) generations, with eventual consequences for evolutionary connectivity. Processes that increase the availability of prospective mates should, all things being equal, forge new connections via gene flow and increase the genetic diversity within a population. These processes could be direct, such as new interactions with a population of the same species, or indirect, such as reduced competition leading to an increase in available mates. Over the long term, increased connectivity should decrease the prevalence of local phenotypes and their associated genotypes, and potentially increase the occurrence of hybridization (i.e., more overall genetic diversity, but less genetic adaptation to particular local environments). In contrast, processes which decrease the mating pool within a generation will generally reduce overall genetic diversity, but could possibly increase the prevalence of local adaptation, though the outcome of these various scenarios will depend on the interplay between natural selection and genetic drift. These processes could include the presence of a new predator or the reduction of available nutrients. Crucially, not all increases to the mating pool result in increased evolutionary connectivity – only new gene flow between distinct populations creates increased connectivity (defining what counts as a distinct population is a perpetual exercise and outside of the scope of this paper).

We can also use this generational perspective to examine how evolutionary connectivity may or may not impact an organism’s ecological connectivity (Figure 3). We know that evolutionarily connected populations must contain some shared genetic material and, within a focal population, this shared material may be maintained or removed through natural selection or random chance (i.e., genetic drift; Griffiths, 2005). Evolutionary connectivity has consequences for ecological connectivity if the presence of shared genetic material alters a species’ pattern of occurrence within the present generation. For this to occur, and assuming a strong enough selection coefficient and a large enough population size, the genetic material must first impact some trait which has ecological importance – if the trait is neutral, the impact will not be felt evolutionarily or ecologically (Mariani and Bekkevold, 2014; Xuereb et al., 2021). Secondly, trait impacts must be relevant for the present generation. For example, a trait expanding tolerance to cold may be evolutionarily important, but if current temperatures do not drop below a meaningful minimum temperature threshold, the trait is not expected to change the population’s current distribution. In contrast, if gene flow introduces variation in a trait relevant to current conditions, the distribution of a population within its range may shift, potentially altering the degree of direct or indirect contact between the focal population and its potential interactors. Thirdly, and relatedly, to impact ecological connectivity, a trait must inherently be under selection (i.e., the frequency of the trait is expected to change in subsequent generations). If all of this is true, specifically if introduced variation is functional, presently ecologically important, and under selection (i.e., not neutral), such evolutionary connectivity will impact ecological connectivity via altered occurrence, and therefore interactions, of the focal species.

Like the mating pool in Figure 2, ecological connectivity reflects changes in the distribution of organisms in space and time (i.e., occurrence). Evolutionary connectivity that alters occurrence patterns within the present generation alters a species’ ecological connectivity. Of course, what precisely counts as the “present” generation depends on a species’ life history and timescale. Many species have overlapping generations, so evolutionary connectivity may alter occurrence patterns for a species’ offspring, which could then have consequences on the pattern of occurrence for older individuals in the population. It is important to remember that both ecological and evolutionary connectivity occur at the population level; individual patterns of occurrence or isolated hybridization do not, by themselves, indicate meaningful connections between populations.

In general, our generational framework aims to (1) explicitly identify the timescale of a research question, (2) identify data sources or points of view relevant to the research question, and (3) incorporate inferences gained from other conceptions of connectivity, while placing research findings within broader ecological and/or evolutionary contexts. As an example, we will use a research question focused on the population connectivity of summer flounder, Paralichthys dentatus.

Identifying the timescale of our inferences involving connectivity allows us to place our findings more clearly within the context of evolutionary and/or ecological processes. The generational framework introduced here serves as a tool to synthesize information about population connectivity. Understanding any natural process requires multiple points of view and no single definition of connectivity will adequately encompass the entirety of any complex natural system. Given this, our framework provides a structure to help researchers collaborate across subdisciplines without inventing new terminology. By allowing population connectivity to have context-specific meaning and focusing on the timescales of connectivity processes, it becomes easier to see which types of data and points of view are relevant for particular research questions.

Like any model, this framework serves as a starting place rather than an end point for understanding populations. Species interactions are complex. Conspecifics within the same population may have unexpected interactions (i.e., predation), challenging assumptions about ecological interactions and their evolutionary connectivity outcomes. For most species, multiple generations exist simultaneously, so the distinct “one generation” and “2+ generation” categories are not always clear. Ecological interactions between different generations may have complex evolutionary outcomes and vice versa. Investigating how discrete or overlapping generational structures may alter these inferences is one potential application of this generational framework.

While the framework discussed here is clearly grounded in population connectivity, population-level connections are not the only types of connections between organisms. Within ecology, connectivity can also refer to abiotic ecosystem components, including the transport of matter or energy within and between habitats (Pringle, 2001; Hillman et al., 2018; Hilty et al., 2020; Keeley et al., 2022). An oft-cited example of this is the influx of marine-derived nitrogen to upland riparian streams during salmon spawning (Helfield and Naiman, 2001). Many species coevolved with this seasonal influx of nutrients and shifts in salmon biomass have widespread ecological consequences (Carlson et al., 2011). The generational framework does not explicitly include these ecosystem connections, however, inferences from such studies might be incorporated if they were clearly assigned a particular timescale.

Only by integrating across many fields of study can we gain a holistic understanding of connectivity within natural systems. Although studies of population connectivity from multiple perspectives are rare (Cooke et al., 2008), the few studies that do exist have yielded important insights. Cross disciplinary collaborations can be challenging, but successful collaborations start with a shared set of ideas and goals (Knapp et al., 2015). We hope the generational framework presented here can serve as a foundation for conducting and communicating population connectivity research across subdisciplines.

AC, JH, TD, RG, JT, JC, EP, JG, and RB conceived and designed the manuscript content and scope. AC, JH, TD, RG, JC, and JT performed background research and framing. AC, JH, TD, RG, and JT wrote the manuscript. JG, EP, RB, TD, RG, and JT provided critical reviews. AC, JH, TD, and RG edited the manuscript. JG, EP, and RB supervised the project. All authors contributed to the article and approved the submitted version.

This project was funded by the Research Coordination Network for Evolution in Changing Seas (NSF-OCE 1764316).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.