Eco-evolutionary dynamics in changing environments: integrating theory with data

Introduction

While the debate on whether the ecological processes are separated from evolutionary processes was very active in the second half of the last century (Antonovics, 1976a; Hutchinson, 1965; Reznick, 2013), it is now broadly accepted that the “non-evolutionary answer to the [ecological] question of why an animal is abundant in some parts of its range and rare in others [and what determines a species’ border] is of necessity incomplete” (Orians, 1962). Population growth rate reflects both the mean adaptation to the current environment and the variance around it, while the rate of adaptation depends on the changing genetic composition and the selective pressure in the interacting populations (Pimentel, 1961). Species’ interactions continuously change via evolution and depend on the ecological context (Barraclough, 2015; Brockhurst & Koskella, 2013; Dostál, 2024; Holt, 2005; Post & Palkovacs, 2009; Sakarchi & Germain, 2023).

The feedback between ecology and evolution is important over a wide range of timescales: from short (contemporary) to long (millions of generations) (Fronhofer et al., 2023; Hendry, 2017). Ecological dynamics is a very broad term, encompassing changes in community structure and species’ (biotic and abiotic) interactions as well as simple population dynamics, and these ecological processes have a fundamentally different complexity. While local population dynamics tend to be fast, other ecological processes such as community assembly over large spatial scales can be slow, even substantially slower than the evolutionary processes—as evidenced by similar niches being occupied by very different species across continents (Antonovics, 1976a; Darwin, 1859). Therefore, the conjunction between temporal and spatial scales is central to many eco-evolutionary processes: community assembly (Leibold et al., 2022), speciation, co-evolution and co-existence in sympatry (Irwin & Schluter, 2022) and species’ range evolution (Barton, 2024). Notably, temporal and spatial variation fundamentally influence variability both within and between species, which, in turn, interacts with community dynamics.

It appears that two fairly distinct domains of eco-evolutionary dynamics coexist with limited interaction, perhaps reflecting both the historical separation of ecology into functional, population and community ecology (see also Orians, 1962) and the relative emphasis on ecology vs. evolutionary genetics. One domain focuses on the evolution of species’ interactions, often in communities with a complex network structure (Johnson & Stinchcombe, 2007; Loeuille, 2010; Montoya et al., 2006; Segar et al., 2020; Thompson, 2019), where with only rare exceptions (Cosmo et al., 2023), within-species variation tends to be neglected or, in theoretical models, fixed on the phenotypic level (Bolnick et al., 2011; Violle et al., 2012). The second domain focuses on the feedback between population regulation and evolution by including its underlying genetics (i.e., genetic architecture and/or variance): typically with a single focal species evolving in its biotic and abiotic environment, while population dynamics reflects the degree of adaptation. This area includes a lot of research on species’ range margins (Antonovics, 1976b; Barton, 2001, 2024; Hanski, 2012; Pease et al., 1989; Polechová, 2018), and evolutionary rescue in changing environments (Bell & Gonzalez, 2009; Carlson et al., 2014; Gomulkiewicz & Holt, 1995; Uecker et al., 2014). The limited communication between the domains of (i) community eco-evolutionary dynamics and (ii) eco-evolutionary dynamics of a focal species with defined population regulation and genetic architecture is detrimental beyond the semantic ambiguity of “eco-evolutionary dynamics”: it suppresses research on the feedback amongst variability within-species (including its genetic architecture), population regulation, and community and ecosystem dynamics (Holt, 2005; Matthews et al., 2011). The separation is even more striking when considering spatial structure (Urban et al., 2008); and we devote a full section to this topic.

A few studies have addressed the feedbacks between evolution, demography, and community dynamics in experimental systems (Bassar et al., 2010; Becks et al., 2012; Palkovacs & Post, 2009; Yoshida et al., 2003), in the wild (Farkas et al., 2013) and theoretically (Abrams, 2000; Fussmann et al., 2000; Kendall et al., 1999; Loeuille, 2010; Patel et al., 2018). With the advance of genomics in the recent decades, it has become possible to study the feedback between community dynamics and complex genetics of adaptation (Becks et al., 2012; Crutsinger et al., 2006; Rudman et al., 2018; Sinclair-Waters et al., 2024). This extends our ability to study adaptation in the absence of a prior knowledge of the traits that are fundamental to community dynamics, and also enables to detect signatures of adaptation over a wider range of timescales. Nevertheless, the more complex the genetic architecture, the harder it is to detect selection at any locus: both as with decreasing effect of any single allele, selection (and thus the rate of change) decreases, and because the fitness effect of the associations with other loci may be changing through time (due to epistasis or at the infinitesimal limit of many loci of small effect, Barghi et al., 2020; Barton et al., 2017). Furthermore, in contrast to manipulative experiments with a known driver of selection, with increasing ecological complexity, it becomes harder to find the proximate drivers of selection (Sinclair-Waters et al., 2024), which may change through time, further complicating inference (Merilä et al., 2001).

Ecological processes tend towards generating more efficient (e.g., in resource acquisition) and stable communities (Hutchinson, 1959; MacArthur, 1955), and the strength and variability in species’ interactions is one of the core factors affecting community stability (Barraclough, 2015; Bolnick et al., 2011; MacArthur, 1955; Des Roches et al., 2018). What variability within-species does this imply, on both phenotypic and genetic levels? Phenotypic variance influences ecological processes such as population regulation and species’ interactions, while genetic variance, its heritable component, together with the selection gradient, determines the speed of evolution (Fisher, 1958; Lande, 1976). When genetic variance is too small relative to the rate of temporal and spatial change, this leads to a failure to adapt and thus limits a species’ niche through time and space (Barton, 2024; Lynch & Lande, 1993; Pease et al., 1989; Polechová et al., 2009).

As remarked by Antonovics (1976b), an answer to the question of what limits adaptation and/or genetic variability that does not include an explicit feedback between eco-evolutionary dynamics and genetic variability is insufficient, bordering on tautology: if genetic variance is limiting, what keeps genetic variance too small? Genetic variance evolves in changing environments both due to spatially and temporally variable selection (Felsenstein, 1976) and decreases due to effect of genetic drift in small or fluctuating populations (Wright, 1931). Notably, genetic variance increases beyond simple mutation–selection balance due to gene flow across heterogeneous environments, and there is an interaction amongst dispersal, evolutionary dynamics, and population dynamics (Fouqueau & Roze, 2021; Pisa et al., 2019; Polechová, 2018; Whitlock & Barton, 1997; Yeaman & Otto, 2011).

As well as determining the rate of evolution (Fisher, 1958; Frank & Slatkin, 1992), the associated variance in fitness imposes a cost (Lynch & Lande, 1993), thus reducing the growth rate. Furthermore, there is often a trade-off between efficiency in current conditions and breadth of the niche: first, due to direct costs/benefits of bet-hedging (Seger & Brockmann, 1987), and second, a large niche breadth, if not correlated with genetic variability, decreases adaptability (Auld et al., 2010; Whitlock & McCauley, 1999). Typically, eco-evolutionary feedbacks are studied conditional on the existing genetic architecture (Yamamichi & Ellner, 2016; Yamamichi, 2022). Yet, what genetic architecture do we expect to prevail, depending on the species’ eco-evolutionary history? In general, larger grain, such as discrete optima or multiple hosts, tends to favour larger-effect loci and evolution of linkage-disequilibrium (Gilbert & Whitlock, 2017; Holt & Barfield, 2011; Yeaman & Whitlock, 2011; Yeaman, 2015). In contrast, adaptation to conditions changing smoothly through time and/or space tends to be less sensitive to the size of the allelic effects (Matuszewski et al., 2015; Polechová, 2018).

This special issue stems from our symposium at ESEB 2022 (Prague), driven by our desire to foster dialogue between researchers focusing on theory, experiments, and natural history—as well as integrate a wider range of eco-evolutionary dynamics. This issue thus covers eco-evolutionary dynamics in the broad sense, from (i) adaptation and phenotypic plasticity in complex environments, focusing on single species, and (ii) the importance of gene flow for eco-evolutionary dynamics in spatially structured populations and communities, to (iii) the feedback between adaptation, population regulation and community dynamics. In the second section, we also revisit the concept of biological species and discuss that completion of speciation to zero gene flow is not inevitable (Barraclough, 2024; Butlin, 2024), highlighting that both local and global adaptation can be shared between good biological species, and that speciation is fundamentally an eco-evolutionary process. The speciation process has both spatial and temporal dimensions, and is inherently coupled with population dynamics (Harvey et al., 2019). Typically, nascent species both exchange genes and compete for similar resources (Barton, 2010; Germain et al., 2021; Irwin & Schluter, 2022). Furthermore, species’ interactions, in particular symbiosis or parasitism, can influence or even drive speciation (De Vienne et al., 2013; Dorey & Schiestl, 2024). Thus, considering speciation as a (point-) ecological process analogous to mutation (Govaert et al., 2019; Vellend, 2010) is not necessarily appropriate. We stop short of phylogenetic time scales and macroevolutionary patterns (Segar et al., 2020; Urban et al., 2020), although the connection between evolutionary processes and macroecological patterns is undoubtedly an exciting topic.

Overview

Adaptation and phenotypic plasticity in complex environments

As environment changes in time and space, selection acts on many traits simultaneously. The effect of simultaneous selection on several traits will depend on how these traits are genetically correlated with each other, their effect on fitness (Antonovics, 1976b) and the kind of selection (such as directional vs. stabilising, Chevin et al., 2010b; Matuszewski et al., 2014; Schneemann et al., 2024; Welch & Waxman, 2003). As long as these correlations are constant or changing slowly, we can still understand the current limits of species’ adaptation based on the overall fitness load (Haldane, 1957; Kirkpatrick & Barton, 1997; Lynch & Lande, 1993), components of which can be estimated from transplant experiments. Yet, in order to understand adaptive potential in changing environments, disentangling the structure of selective pressure can be helpful: for example, while the combined stressors may induce fitness costs that prevent adaptation, adaptation may be possible under uncorrelated and/or weaker selection. Adaptive phenotypic plasticity diminishes the fitness load by weakening selective gradients, and an increase in plasticity may thus enable adaptation where demographic stress and genetic drift would otherwise have led to extinction (Bradshaw, 1965; Hendry, 2016; Pigliucci, 2005; Via & Lande, 1985). Theory predicts that the evolution of plasticity (Chevin & Hoffmann, 2017; Chevin et al., 2010a) can aid adaptation, yet empirical demonstrations of evolvable plasticity have been sparse (Dostál, 2022; King & Hadfield, 2019). Choy et al. (2024) have demonstrated that when Drosophila melanogaster were subjected to a combined thermal–nutritional stress, there was substantial heritable variation in plasticity for different life-history traits, while the genetic effects on plasticity were largely uncorrelated between different life-history traits. Interestingly, in a meticulous design separating adaptation to wet and/or warm environment of Festuca rubra via changes in plasticity (G $\times$ E effects) from direct genetic effects, Münzbergová et al. (2024) found significantly more markers associated with changes in plasticity in response to the climate-driven selection than ones associated with trait mean.

Increase in phenotypic plasticity, reducing the multidimensionality of the stressors, and reducing the mean selection pressure (perhaps while increasing variance) can all reduce the fitness load (Haldane, 1957) to a level more favourable to adaptation. Godinho et al. (2024) investigated an aspect of this problem: is adaptation to a stressor (cadmium) easier or harder in homogeneous or heterogeneous environments? The authors assessed adaptation in spider mites feeding on cadmium-rich plants, control plants or their mix: while after 12 generations, adaptation to cadmium was only present in the heterogeneous environments, the differences lapsed in the later generations. The lack of detected adaptation is attributed to low heritability of the fitness traits, yet the authors still observed a trade-off in fitness for spider mites grown in the cadmium-rich environments when returned to the control plants—implying that adaptation may have happened in other traits. Whether adaptation is easier or harder in heterogeneous environments depends on details of selection, genetic architecture, and dispersal patterns—which we discuss in the next section.

Eco-evolutionary dynamics in spatially structured communities

While both the ecological dynamics of metacommunities across spatial scales (Leibold et al., 2004) and the joint evolutionary and population dynamics of a focal species are fairly well studied (Sexton et al., 2009), there is still very little overlap beyond pairwise species’ interactions (Urban et al., 2008). Below we concentrate first on the feedback between population regulation and evolution in spatially structured environments with gene flow, including evolution of genetic architecture. We then extend the discussion to another neglected angle: persistent gene flow between (nascent) species. Speciation with gene flow is a slow process (Endler, 1977; Nosil, 2012) but it is often assumed that a complete reproductive barrier eventually forms in sexually reproducing (recombining) species (Butlin & Smadja, 2018; Kulmuni et al., 2020)—and if such a barrier does not form, it is because selection is too weak to complete reinforcement (Bímová et al., 2011; Bank et al., 2012). Could low gene flow between species be under recurrent positive selection, possibly conditional on particular genomic architecture (Barraclough, 2024; Butlin, 2024)? How could such a process affect adaptation to changing environments and coexistence (Barraclough, 2024)? We close by discussing coevolution in complex communities during range expansions, both in native and invaded habitats (Dostál, 2024; Lustenhouwer et al., 2024).

Dispersal, population dynamics, and local vs. global adaptation

The importance of spatial structure and density regulation in evolutionary ecology is well recognised (Antonovics, 1976a; Haldane, 1956). The field is, however, somewhat divided between (i) metapopulation/metacommunity dynamics, where dispersal probability into a particular subpopulation is independent of physical distance (Hanski, 1998) vs. (ii) consideration of continuous spatial scales, often with diffusive dispersal (Barton, 2001; Pease et al., 1989; Richardson et al., 2014). The synthesis by Barton (2024) puts the results of these structurally different models into mutual context—from source-sink type models and classic metapopulations with discrete demes and implicit space to explicit spatial structure with local dispersal. Furthermore, Barton (2024) explains how genetic architecture affects the outcome and, linking ecological genetics with information theory, discusses what kind of genetic architecture is expected to evolve under different population structures and local vs. global selection pressure. Barton (2024) shows that in populations with discrete optima, adaptation is favoured when selection per locus is strong relative to migration: hence, larger allelic effects or linkage between loci enable adaptation. With smooth environmental change, even high levels of gene flow are often beneficial due to both an increase in genetic variance (and influx of beneficial alleles) and a decrease in genetic drift while adaptation is less sensitive to the effect size (Barton, 2024; Sexton et al., 2024). Barton’s synthesis also highlights that strict separation of timescales can be problematic: although the rate of change in population size is much faster than the change in the trait mean, and genetic variance evolves even more slowly, the coupling of these processes in the theoretical analysis qualitatively affects the conclusions.

The swamping effect of migration increases both with increasing asymmetry in density across populations and the grain of the environmental change. Yet a low level of migration is still beneficial as it potentially brings in adapted genotypes (Gomulkiewicz et al., 1999; Uecker et al., 2014). When conditions vary sharply, dispersal and the resulting gene flow will aid adaptation only in genetically impoverished populations. Durkee et al. (2024) show that unless gene flow is very strong, it may not have any negative effects: in their study, adaptation to insecticide in experimental populations of Tribolium was largely independent of the immigration rate (0, 1, or 5 migrants per generation coming into a population of $N=50$⁠).

The synthesis of Barton (2024) and the studies of Sexton et al. (2024) and Durkee et al. (2024) emphasise the complex feedbacks amongst spatial structure, migration, selection, and genetic drift. A useful measure to assess the strength of genetic drift is the inverse of the effective population size $N_e$⁠, which determines the rate of change of the genetic composition of a population caused by genetic drift. Whether a globally favourable mutation spread, or a deleterious mutation is eliminated, is controlled by the product of the total effective population size and the intensity of selection (⁠$N_{e}s$⁠, Charlesworth, 2009; Wright, 1931, 1933). Note that the total $N_e$ increases as migration decreases as it takes longer for the alleles sampled at random from the whole (meta)population to coalesce (Slatkin, 1991); though the effect reverses when within-deme fluctuations are large (Whitlock & Barton, 1997). While total $N_e$ is a useful metric to assess the fixation times and probabilities of both deleterious or favourable mutations (Whitlock, 2003), the reality is more complicated. First, increasing subdivision (decreasing migration rate) will amplify the effect of genetic drift in a local population (Whitlock & Barton, 1997). Second, the outcome depends on the underlying genetic architecture as the effect of drift on a single locus does not extend to polygenic adaptation. While single locus theory requires $N_{e}s\mathit{>>}1$ for selection to be effective relative to drift, this is not the case for polygenic traits. Even if the effect of selection acting on each individual allele is weak, adaptation in a polygenic trait can still be efficient and occurs via changes in associations between loci (Barghi et al., 2020; Barton et al., 2017). Third, we may be interested in whether diversification across continuous spatial scales is possible: can a cline form along an environmental gradient? In this case, the relevant measure of genetic drift is the inverse of the neighbourhood size (Wright, 1943, 1946), which can be thought of as the number of individuals within one generation’s dispersal range (Charlesworth et al., 2003) (and thus, the strength of genetic drift decreases with increasing dispersal distance). In the context of diversification across continuous space, clines steepen and become increasingly rugged as genetic drift (measured by the inverse of neighborhood size) increases relative to selection (Nagylaki, 1978; Polechová & Barton, 2011; Slatkin and Maruyama, 1975)—and eventually will fix. Yet, before the variation is lost, apparent “noise” across the loci may still contribute to adaptation in the (statistical) context of all the other loci as in the classic infinitesimal model (Barton, 2024).

In general, sensitivity of adaptation to migration, selection and drift depends on the type of metapopulation structure and its optima, and favours different genetic architecture. Estimated local effective population size across a population is a good indicator of the expected relative efficacy of selection: in this issue, Reynes et al. (2024) show that local $N_e$ estimated from temporal samples indeed positively correlates with the number of detected candidate loci underlying adaptation to the changing surface sea temperature. Diminishing the impact of genetic drift strengthens resilience of populations in changing environments, and doing so typically stabilizes demographic fluctuations as well.

While the exact conditions may differ, strong genetic drift will lead to an eventual collapse of local adaptation in source-sink situations (Gomulkiewicz et al., 1999; Sachdeva et al., 2022), classic metapopulations (Olusanya et al., 2023; Szép et al., 2021) as well as in continuous space with local dispersal (Polechová, 2018; Polechová & Barton, 2015)—unless preceded by a demographic collapse (Lande, 1988). Sexton et al. (2024) argue that protecting marginal or declining populations from dispersal is likely to be harmful for the adaptive potential of such populations, particularly under escalating climate change. The benefits of local dispersal often outweigh the costs (Aitken & Whitlock, 2013). In some cases, introgression of adaptive alleles from already adapted populations may be beneficial even if these are substantially genetically distant: in the face of accelerated change, an increase in fitness variance may be more important than the increase in its mean.

Gene flow across a species’ boundary

In the previous section, we considered the dynamics of species’ range limits as the corollary of the limits to adaptation of focal species, treating the environment as passive. This is obviously an oversimplification (Case et al., 2005) but before we turn to interspecific interactions, we would like to focus on adaptation that is shared across a species’ boundary. For many species, there is evidence of contemporary introgression from co-occurring species, typically (but not necessarily) phylogenetically closely related (Bohutínská & Peichel, 2023; Grant & Grant, 1989; Mallet et al., 2016). The question is, does this boundary typically close (either by reinforcement or as a side effect of divergence, Abbott et al., 2013; Bisschop et al., 2020), or is there evidence for a stable, low level of gene flow, perhaps just in parts of the genome? If the species’ boundary is blurred, is that because selection strength is too low to be effective, or could there be selection against cessation of gene flow (in particular genomic regions)? This is an intriguing possibility explored in a perspective by Barraclough (2024) and the commentary by Butlin (2024). For example, globally changing conditions or fluctuating environments may impose selection where a more widely shared gene pool can be beneficial; selection may also be recurrently favouring transfer of genes adapted to a novel environment from a species already adapted to it (Satokangas et al., 2023). Such introgression would typically be highly asymmetric (Baird et al., 2023). The ease of introgression will, of course, depend on genetic architecture, epistasis (Martin et al., 2019), and whether the transfer includes the whole genome or parts of it. At the moment, while there is ample evidence of instances of adaptive introgression (Hodson et al., 2022; Malinsky et al., 2018; Roux et al., 2016), the evidence for selection favouring recurrent gene flow between species is scarce (Nocchi et al., 2024). Currently, it remains an open possibility that a reinforcement of speciation to zero gene flow between the nascent species (Barton, 2020; Kulmuni et al., 2020) is not the only expected outcome and that repeated introgression in particular genomic regions may be favoured across some phylogenetically more distant species, perhaps following a rapid range expansion (as hypothesised by Dostál (2024)).

Coevolution in spatially structured communities

Coevolution is an integral part of the dynamics of species’ ranges and niches (Case & Taper, 2000; Ricklefs, 2010; Mestre et al., 2023)—and we see it accelerated in the ecosystems affected by rapid range expansions. Dostá (2024)’s review stresses that both native and invasive species adapt following an invasion; a textbook example is the so-called boom-bust dynamics of invasive species (Elton, 2020; Simberloff & Gibbons, 2004), though its extent and prevalence may be exaggerated in the literature (Strayer et al., 2017). Very few examples of current coevolution following an invasion are well documented. In particular, the studies of Alliaria show that invasion is followed by (i) a complex co-evolution that may involve accumulation of pathogens of the invader (Lankau, 2012), (ii) evolution to a less harmful form of the invader via decreased production of (costly) toxins (Lankau et al., 2009) and (iii) adaptation of the native species, such as evolution of tolerance to the toxins of the invader (Blossey et al., 2021). The last point has also been observed in Veronica chamaedrys, which evolved its shade tolerance in response to the invader, Heracleum giganteum (Dostál, 2022). Thus while the interaction would not be compatible with coexistence at first, this may change over time as the ecological community evolves (Dostál, 2024; Yamamichi et al., 2022).

Another important feedback upon range expansion involves coevolution between plants and soil microbes. Lustenhouwer et al. (2024) studied the effect of soil microbiota on growth of stinkwort Dittrichia graveolens across both the native and exotic range, and demonstrated a complex pattern of co-adaptation (or lack of it) with no evidence of pathogen release during range expansion. For plant seeds from the native range, the growth rate was on average higher in the soil of its own micro-site, irrespective of the core vs. edge origin and destination.

Feedback amongst adaptation, population, and community dynamics

Genetic variance fundamentally determines the time scale of evolution in eco-evolutionary dynamics (besides affecting ecology) (Bolnick et al., 2011; Lande, 1976; Patel et al., 2018): the greater the genetic variance and the strength of selection, the faster the speed of evolution and, hence, the larger the relative importance of evolution to ecology (sensu population dynamics). Yet, very few studies experimentally assess the importance of genetic variance for eco-evolutionary dynamics (Yoshida et al., 2003). Hermann et al. (2024) estimate how changes in the ecological and evolutionary predictor variable contribute to changes in the response variable, predator growth rate (Ellner et al., 2011; Hairston Jr et al., 2005). They show that, with a higher variance in a prey-defence a prey-competitiveness (growth rate in the presence vs. absence of predator) trait, there is indeed a higher contribution of evolution relative to ecology than when this variance is smaller. The variance was manipulated by combining clones with different values of these traits, and both for the prey Chlamydomonas reinhardtii and the obligatorily asexual predator Brachionus calyciflorus, the reproduction during the experiment remained asexual. The speed of evolution is, therefore, a function of the selective advantage of the fitter clone, an advantage that increases with phenotypic distance (Orr & Unckless, 2014; Yamamichi & Miner, 2015). The results, while displaying substantial fluctuations in the estimated numbers, qualitatively match the predictions of a deterministic model of the system.

Within-species’ genetic variation not only directly affects speed of evolution, it is also a core factor affecting community dynamics. It is often postulated that higher (phenotypic) diversity of core species supports higher ecological diversity (Hughes et al., 2008; Vellend & Geber, 2005) and that there may be a positive feedback between them, but the experimental evidence is scarce (Papadopoulou et al., 2011). In their pioneering work, Sinclair-Waters et al. (2024) apply GWAS to a community trait in the wild: rather than studying genetic associations with a trait of the focal species, they assess the association between genetic diversity within-species and the co-occurring arthropod communities. While they find a positive association with community traits, the path analysis reveals that most of the variation is explained by host-plant traits, thus highlighting the difficulties of disentangling causal effects in complex eco-evolutionary studies.

Summary

In this special issue, we aim to bring together eco-evolutionary studies that emphasise the feedback between population regulation and evolution of traits and their genetic basis, with research focused on feedback between evolution of species’ interactions and community dynamics. In particular, we focus on the importance of the eco-evolutionary feedback in spatially structured populations. Despite this being an expanding area of research, the integration of evolution (of focal species’ traits as well as of species’ interactions) with population regulation and community dynamics remains limited. Yet, it may substantially improve our understanding of adaptation dynamics in changing environments, with implications for both population and community resilience—and ultimately enable a deeper insight into the fundamental principles shaping patterns of species’ diversity.

With the technological advances including modern molecular genetic methods, it is now plausible to test novel predictions that include the genetics of complex traits associated with a focal-species’ characteristics or community structure (community traits). We find the following directions particularly interesting:

(1) What factors determine whether evolution stabilizes or destabilizes ecological dynamics? In particular, how do the predictions depend on ecological factors such as network structure, as well as on evolutionary factors such as within-species variance, frequency of recombination, and genetic architecture? How does the evolution of the genetic basis affect the dynamics?

(2) Spatial and temporal variability influence the above processes and create non-trivial feedbacks; a notable example is the maintenance of genetic variance and its effect on ecological dynamics. How do the predictions change when spatial structure (with heterogeneous selection, local population dynamics, and dispersal), temporally changing environment (directional or fluctuating), and their interaction are taken into account?

(3) When does selection favor recurrent introgression in the context of speciation and under rapidly changing conditions, where we expect both increased demographic fluctuations and more accidental mixing due to range shifts, as well as more shared adaptations? How does this depend on the current genetic architecture and its evolution?

To determine causality, a testable and sufficiently complex theory, as well as clever experimental designs, are necessary, along with a good understanding of organismal, population, and community biology.

Funding

This research was funded by the Austrian Science Fund (FWF), project doi: 10.55776/P32896, Institutional Identifier: 501100002428, grant number: P32896 and L.F. acknowledges the support of the NOMIS-ISTA Fellowship Program.

Acknowledgements

We would like to thank Nick Barton, Roger Butlin, Stuart Baird, Patrik Nosil, and Jason Sexton for their insightful comments on the earlier drafts, and to John Carchrae for his valuable contribution in refining phrasing and enhancing clarity. For open access purposes, the author has applied a CC BY public copyright license to any author-accepted manuscript version arising from this submission.

Conflicts of interest

None declared.