The Princeton Guide to Evolution

IV.3

Geographic Variation, Population Structure, and Migration

Ophélie Ronce

OUTLINE

1. The causes of spatial structure in genetic diversity

2. Individuals and their genes move around

3. Gene flow shapes patterns of spatial genetic structure

4. Evolution in spatially structured populations

5. Implications for conservation

Species are rarely genetically homogeneous sets of individuals, and genetic diversity is not distributed randomly through space within their ranges. Spatial patterns in genetic diversity can be observed at many different scales. At a very fine scale, within a continuous population, the genetic similarity between two individuals generally declines with increasing distance between them. For instance, in the annual plant Medicago truncatula, two individuals growing within 0.5 m of each other in an old field were found to be on average about 10 times more likely to carry identical variants for some highly polymorphic DNA sequence than two individuals separated by 7 m in the same field. Genetic similarity often varies along environmental gradients. Distribution of genetic diversity can also be patchy. Spatial patterns of genetic variation can be manifest at very broad geographical scales. Both genetic drift and selection can explain the emergence of spatial patterns in genetic diversity. Movement of individuals through space, resulting in gene flow, shapes these patterns. In turn, the nonrandom distribution of genetic diversity through space has multiple consequences for the evolution of mating systems, life histories, and more generally, fitness. Finally, spatial genetic structure and its evolutionary consequences bear many implications for the conservation of biodiversity in the context of global changes.

GLOSSARY

Cline. A gradient of continuous variation through space for a genetic or phenotypic character within a species

Dispersal Kernel. The probability density that an individual initially at coordinates (0,0) is found at coordinates (x,y) after dispersal.

Gene Flow. The partial mixing between different populations.

Genetic Rescue. Increase in a population’s mean fitness due to the introduction of genetically divergent individuals.

Gene Swamping. Lack of significant response to selection because gene flow is too high.

Habitat Selection. The nonrandom distribution of individuals after dispersal across habitats.

Heterosis. Higher fitness of progeny born to parents originating from different populations rather than from the same population.

Isolation by Distance. Decreasing genetic similarity with increasing distance; due to shared ancestry when dispersal is limited.

Local Adaptation. Higher fitness of resident genotypes in their native environment relative to that of immigrant genotypes in that same environment.

Metapopulation. A set of discrete populations connected by dispersal.

Migration Load. Decrease in mean fitness of a population because of immigration.

Outbreeding Depression. Lower fitness of progeny born to parents originating from divergent populations than from related populations.

Wahlund Effect. The higher proportion of homozygotes due to local mating compared with that expected in a well-mixed population with the same genetic diversity.

1. THE CAUSES OF SPATIAL STRUCTURE IN GENETIC DIVERSITY

Genetic Drift

At a fine spatial scale, when offspring move a short distance from their parents, the formation of local pedigree structures such that nearby individuals are highly related to each other is a basic explanation for patterns of genetic isolation by distance, as observed in M. truncatula. Similar arguments hold at different spatial scales: if exchanges between populations are rare, two individuals found in the same population are more likely to share a common ancestor in the relatively recent past (in that same population) than are two individuals in different populations, whose common ancestor probably dates back to an earlier time. When lineages have been separated for a longer time, the additional time has allowed for mutations to appear and develop differences in gene copies that initially descended from the same ancestor, which can explain the greater divergence between genes sampled in different populations rather than in the same population.

This process is accentuated by small population size. When the number of individuals reproducing locally is not large, it is more likely that two local inhabitants share a common ancestor in a recent past. Many individuals will then carry identical gene copies that have not been altered by mutation since they descended from the same ancestor. Consanguinity then results in decreased genetic diversity and increased homozygosity at the local scale. At the extreme, local genetic diversity can be entirely lost. As this random drift process is blind, it is likely that different genes have spread stochastically to fixation in different populations when the latter are isolated from one another (see chapter IV.1).

The demographic history of populations thus has much potential to shape spatial genetic patterns. Many species have undergone relatively rapid spatial expansion after the retreat of glaciers in the quaternary. Colonization of new areas by a reduced number of founders affects spatial patterns of genetic diversity, because genotypes found in the new part of the range are only a small sample of the diversity found in the original range. The spatial spread of chloroplast variants at the continental scale has therefore been used to infer the various recolonization routes of oak trees from the distinct glacial refuges in southern Europe, with good agreement with the pollen fossil records.

Selection

The other main explanation for spatial genetic structure is that selection favors different genotypes in different locations. In the mosquito Culex pipiens, genes conferring resistance to insecticide are obviously strongly selected for in areas near the coastline where pesticides are used. The mutated protein that is the insecticide target, however, functions less well than the original one, so that mosquitoes sensitive to the insecticide outperform the resistant ones in nontreated areas in the north. Coloration in the walkingstick Timema cristinae confers differential crypsis depending on the host plant. The unstriped morph is more cryptic on Ceanothus, whereas the striped morph is more cryptic on Adenostoma. The less cryptic morph on each host decreases in frequency within each generation, as expected if it is subject to higher predation. In other cases, such as the latitudinal cline in wing size in D. subobscura, the agent of selection is less clear. D. subobscura is of European origin and was introduced into both North and South America. The convergent evolution of such similar genetic clines de novo in both continents after introduction, however, suggests that variable selection (somehow linked to temperature), not drift, is responsible for the formation of the pattern.

When new mutants are obtained by mutagenesis, their relative fitness can be assayed in different environments (e.g., different gene deletions in yeast grown with different sugars). Such experiments have been conducted in a number of rapidly reproducing organisms (e.g., bacteria, viruses, fruit flies) and have revealed that the effect of genetic variation on fitness is in general dependent on the environmental context. Given the existence of adequate genetic variation, selection pressures varying in space should result in the emergence of patterns of local adaptation. Reciprocal transplantations of different genotypes across sites often show that local genotypes outperform foreign genotypes in their environment of origin (e.g., in about 70 percent of the sites in data compiled from 35 transplantation experiments with various plant species).

The fitness of different genotypes may also vary from one locality to the other, not because of extrinsic environmental differences, but because of intrinsic variation in the genetic composition of the local populations. This is the case in particular in presence of genetic incompatibilities, such that crosses between some genotypes are partially sterile. When for historical reasons different incompatible genotypes are more frequent in different locations, rare genotypes incompatible with the locally dominant genotype then suffer from a large fecundity disadvantage, which can maintain strong spatial patterns in genetic diversity.

Habitat Choice

A last cause of spatial genetic structure is when habitat choice is dependent on an individual’s genotype. For instance, preference for different host plants is in part genetically determined in some butterflies. Female butterflies with innate preference for some host plant will lay their clutches on patches of that host plant, and the corresponding genotype will be more frequent in such patches, even in the absence of any differences in performance of the different genotypes on that plant. Heavier fledglings of great tits preferentially settle in the less crowded parts of Wytham Wood, while lighter birds settle in denser areas. Phenotype- or genotype-dependent dispersal could thus be a source of phenotypic and genetic divergence among sites.

2. INDIVIDUALS AND THEIR GENES MOVE AROUND

Movement Does Not Amount to Gene Flow

Movement is an essential feature of life. Even sessile organisms such as plants and corals have evolved adaptations like mobile larvae or winged fruits to facilitate movement at some stage in their life cycle. Movement serves many functions, such as foraging for food, finding mates, colonizing new territories, and escaping predators or deteriorating environmental conditions. Not all types of movement affect gene flow: after having spent most of their lives in the sea, salmon undertake long and dangerous journeys to return to mate in the same river where they were born. Despite very large distance movement, there is therefore little genetic mixing among pools of fish from various localities. Conversely, dispersal movements by which individuals leave their natal site to reproduce elsewhere, or to reproduce in different locations in different attempts, are relevant for genetic exchanges across space; moreover, effective gene flow requires that dispersing individuals succeed in spreading their genes in that new location. For instance, colonial naked mole rats are highly xenophobic, and immigrants are often killed when they intrude on a new colony. Depressed or enhanced reproductive success of migrants hence alters patterns of gene flow.

Evolutionary Consequences of Movement beyond Gene Flow

If movement does not necessarily result in gene flow, movement still causes individuals to experience different environmental conditions, and thus different selection pressures. Salmon have, for instance, evolved the capacity to adjust their physiology to different salinity levels. Movement also affects population dynamics, the distribution of population sizes through space, and spread rates, with many consequences on genetic diversity, which are partly distinct from gene flow issues. For instance, experimental populations of the small plant Cardamine hirsuta experience frequent crashes and local extinction when the proportion of seed dispersed between patches of plants is either too small or too large. Genotypes found at the very edge of the range in species undergoing rapid spatial expansion (e.g., in invasive species) can benefit from the demographic wave of spread, increasing in frequency even if they have weak negative effects on fitness. In outcrossing plants, seed dispersal and pollen dispersal both contribute to gene flow between localities, but seed dispersal has distinct demographic consequences, allowing in particular the colonization of new areas. Gene flow can therefore be partly uncoupled from the other evolutionary consequences of dispersal.

Heterogeneity in Dispersal and Gene Flow

Despite the general ubiquity of adaptation to facilitate movement, there is a large heterogeneity among species in the spatial extent of resulting gene flow. Even within a set of highly mobile species such as birds, recovery of ringed individuals shows that among 75 species breeding in UK, the mean natal dispersal distance varies from about 2 km in the Dunnock (Prunella modularis) to about 70 km in the Grey Heron (Ardea cinerea). This heterogeneity is also found within species. For instance, within several insect species, some individuals have fully functional wings while many others carry atrophied wings. The spatial extent of gene flow is often summarized by the mean distance between the location where a parent reproduced and that where its offspring reproduced. Given the highly stochastic nature of movement, dispersal is, however, better described by the entire distribution of distances between offspring and parents. Such dispersal kernels are often highly asymmetrical, with many short-distance dispersal events and a few long-distance events; for instance, viable airborne tree pollen can move over hundreds to thousands of kilometers in some conifers.

Complex Patterns of Gene Flow

It is tempting to describe the movement of living organisms by analogy with the random diffusion of molecules; however, patterns of movement in nature are much more complex. Even for organisms relying on passive dispersal by wind or water, there are often strong patterns of directionality. Most pollen that fertilized seeds in a Swedish population of Pinus sylvestris was found to have originated from higher latitudes by 1 to 2 degrees, a finding that might be due to climatic conditions that year. Dispersal has often been found to be influenced by an organism’s perception of its own internal condition and of its environment. Even in the small ruderal plant Crepis sancta, the proportion of seeds equipped with a parachute-like structure facilitating wind dispersal increases when the mother plant is grown in stressing conditions. Habitat selection is a pervasive feature of many organisms that has the potential to strongly shape the patterns of gene flow through space.

3. GENE FLOW SHAPES PATTERNS OF SPATIAL GENETIC STRUCTURE

Gene Flow Makes Genetic Patterns Vary More Smoothly in Space

The partial mixing of genes across locations tends to blur the spatial genetic structure generated by drift and selection. In particular, genetic similarity among pairs of individuals due to pedigree structure declines more slowly with increasing distance when gene flow occurs on greater spatial scales, and when effective population density is higher. Consistently, genetic similarity generally declines faster with distance in herbaceous plants than in trees, as the latter disperse their genes farther. Patterns of isolation by distance are often used to estimate the spatial scale of gene flow. Such indirect estimates from genetic spatial patterns (e.g., an average distance of 123 m between parent and offspring in the damselfly Coenagrion mercuriale) can agree well with more direct demographic estimates of dispersal distance (128 m traveled within a lifetime in that same species). Gene flow makes genetic clines broader than the spatial scale over which selection changes through space. While the transition between pesticide-treated and untreated areas is relatively sharp, for C. pipiens the changes in resistance frequency through space are quite smooth. Cline shape results from the tension between divergent selection, which enhances genetic differentiation, and the homogenizing effect of gene flow, which erases it. Variation in cline shape can therefore be used to estimate selection intensity and the spatial extent of gene flow.

Migration Load

In the walkingstick T. cristinae, the frequency of the less cryptic morph on a given host plant is much higher when a neighboring population uses the alternate host plant than when such patches of alternate host are found much farther away. Gene flow can thus introduce maladapted individuals in populations and constrain local adaptation. Gene flow can entirely prevent adaptation to local selection pressures in small environmental pockets surrounded by larger areas of a different habitat. Such gene swamping occurs when the force of gene flow is much greater than that opposed by selection: local adaptation polymorphism is lost, and genes advantageous in the dominant habitat spread to fixation. Depending on the spatial scale of gene flow, environmental pockets have a critical size below which adaptation to these specific environmental conditions is lost.

Different Genetic Characters Show Different Patterns of Spatial Variation within the Same Species in the Same Set of Localities

Substantial genetic divergence for genes involved in local adaptation can be maintained when the rest of the genome is homogenized by gene flow. In temperate forest trees, for instance, there is very little divergence among most DNA sequences found in different locations, across very large distances, because of long-distance pollen flow and large population sizes; there is, however, much evidence for fine-scale adaptation to local climate in forest tree species, with marked genetic clines along latitude or altitude for frost resistance or the timing of flowering.

Gene Flow at Range Margins

The constraining effect of gene flow on local adaptation is expected to be stronger in small populations at the periphery of the range, where many migrants are received from core populations adapted to different environmental conditions. Asymmetrical gene flow would make marginal populations more genetically similar to core populations that would be optimal in their own environment. When transplanted into various common gardens, populations of Pinus sylvestris from marginal locations with extreme climate indeed grow better in milder conditions, closer to the core, than in their original locations. The relative role of evolutionary constraints linked to gene flow, lack of genetic diversity, interspecific competition, and other demographic asymmetries in explaining the evolution of range limits (i.e., failure to adapt to environmental conditions outside the range) is still debated.

Gene Flow Allows the Genetic Cohesion of a Species across Space

By constraining genetic divergence between interbreeding populations, gene flow can be seen as the glue binding the collection of populations constituting a species. In particular, gene flow allows the spread of new favorable mutations across the species range. The delta-32 mutation in the CCR5 gene in humans confers resistance to infection by HIV and other diseases. This mutation is found in Europe and western Asia, with a high frequency in northern Europe, and is thought to have originated about 3000 years ago. Analysis of spatial patterns for the frequency of the delta-32 mutation suggests that it has spread as a result of rapid dispersal (more than 100 km per generation) and strong selection but has had insufficient time to expand to the entire range of the human species. The constraining effect of gene flow on divergence explains why the speciation process is often initiated in a geographic context that leads to disruption of gene flow (allopatric speciation).

The strength of the normalizing effect of gene flow in maintaining species integrity and preventing speciation can, however, be questioned. Gene flow at the scale of the range is very rare in some species, such as between populations of the snail Cepea nemoralis found in different valleys of the Pyrenees Mountains. Stabilizing selection, rather than gene flow, might then be instrumental in maintaining some phenotypic uniformity at the scale of the species range. Conversely, some cases of phenotypic divergence, and of further evolution of reproductive isolation, have been documented in geographic contexts where gene flow was initially not absent (parapatric or sympatric speciation).

Gene Flow Affects Levels of Genetic Diversity within Populations

Dispersal shapes the distribution of genetic diversity through space, by increasing the proportion of total diversity contained within rather than between populations. Accordingly, in selfing plants, which lack pollen dispersal, much diversity is contained between populations rather than within populations. Addition of a relatively moderate number of immigrants each generation (one migrant or more) suffices to maintain within each population a large fraction of the total genetic diversity contained in the whole metapopulation.

The maintenance of high levels of variation for traits closely linked to fitness remains a paradox where stabilizing selection is acting to reduce variation within populations. Genetic diversity for adaptive traits within populations should increase with gene flow from divergent populations, up to the point at which polymorphism is lost because of gene swamping. If gene flow between differently adapted populations is a persistent source of genetic variation, there should be strong correlations between genetic diversity within a population and the amount of heterogeneity in the environment around that population. In a study of 142 populations of lodgepole pine, variation for growth indeed correlated with climatic heterogeneity in the region of origin of populations.

Gene Flow Can Advance Adaptation to Changing Environments

By replenishing genetic variance eroded by drift and selection within populations, gene flow can facilitate adaptation to new environmental conditions. In particular, this is the case when selection varies in both time and space: genetic variation that was favored in some other part of the range may become useful in a new location. The evolutionary arms race between parasites and their hosts provides an example of a case where increased migration actually improves local adaptation. When host populations have evolved resistance against infection by their local parasites, introduction of new genetic variants that are less well recognized by the defense system of the local host could allow the parasite population to overcome host resistance more quickly. Experimental coevolution of viruses with their bacterial host Pseudomonas fluorescens in microcosms confirmed such prediction. Local viruses were more infectious on their local host than were foreign viruses, but only when a fraction of viruses were regularly transferred between cultures.

4. EVOLUTION IN SPATIALLY STRUCTURED POPULATIONS

Spatial proximity generally means that individuals are likely to mate, compete, or more generally interact with each other. Genetic resemblance between individuals that are close spatially has therefore many evolutionary consequences.

Different Behaviors and Life History Traits Are Selected For

Ecological interactions with neighbors can greatly affect the fitness of an individual, for example, in competition for the same resource pool, interfering agonistically or, conversely, providing help. Genetic resemblance among neighbors generates some association between the genotype of an individual and the phenotypes of individuals affecting its fitness, with many consequences for the evolution of traits involved in such interactions. For instance, some strains of the yeast Saccharomyces cerevisiae consume glucose very quickly but with a low energetic yield (selfish strains), while other strains consume glucose more slowly but with higher yield (prudent strains). In well-mixed cultures, selfish strains readily invade populations of prudent strains. When competing in a spatially structured metapopulation, prudent strains resist the invasion by selfish strains, because patches of prudent strains more efficiently turn resources into population growth. Prudent yeasts do better when they are surrounded by other prudent yeasts, while selfish yeasts do worse when their competitors are also selfish. Spatial genetic structure sets the stage for kin selection to similarly alter the evolution of cooperation, life histories, and dispersal.

Genetic Load can Increase or Decrease with Population Structure

Frequent mating between genetically similar parents results in an increased proportion of gene copies being carried by offspring in homozygous form, compared to what would be expected under random mating (Wahlund effect). Many mutations have innocuous or weak effects when present in a single copy in an individual, but are strongly deleterious in homozygous form. In the French-speaking Canadian population of Saguenay–Lac-Saint-Jean in northeastern Quebec, a population with few founders and historically isolated because of religious and linguistic issues, carriers of recessive genetic disorders, very rare or unknown in other human populations, can represent up to 4 percent of the local population. Thus, consanguinity due to spatial structure can in the short term greatly depress the mean fitness of the population when deleterious mutations previously hidden in heterozygous form are expressed. In the long term, however, consanguinity helps purge such mutations, potentially ameliorating the genetic load (see chapter IV.5). Purging occurs only when enough genetic diversity remains for selection to sieve from. Loss of local genetic variation means that selection within populations is inefficient in removing badly adapted genes. Genetic load can therefore either decrease or increase in the long term in spatially structured populations, compared to a reference well-mixed population. This depends on the extent of spatial genetic structure, the distribution of mutation effects on fitness, and the exact details of the life cycle, which govern the ways in which competition within spatial entities contributes to change in gene frequencies across generations.

Crosses between Populations Produce Fitter Individuals

The rare plant Ranunculus reptans has a fragmented distribution, with many small isolated populations. Plants whose parents originate from different populations produce more seeds than plants produced by crosses within the same population. This pattern of heterosis, or increased vigor of hybrids, is frequently observed and is used in agriculture, for instance, in corn production. In R. reptans, crosses between populations produce fitter individuals when the parental populations are small and characterized by low genetic diversity, but not when they are large and diverse. The random fixation of different weakly deleterious recessive mutants in distinct small isolated populations could therefore explain the higher fitness of heterozygous hybrids, in which such mutations are masked.

Crosses between Distant Populations Produce Less Fit Individuals

Conversely, hybrids between distant populations may suffer from outbreeding depression. Outbreeding depression can result from loss of local adaptation, when hybrids have intermediate phenotypes, which make them poorly adapted to their parent environment, or when genetic divergence among populations having evolved in isolation results in genetic incompatibilities. Arabidopsis thaliana is a selfing plant with a very large range and very few likely genetic exchanges among distant populations. The gene HPA, with essential function for plant development, is duplicated in A. thaliana. Plants from different geographical origin (Colombia vs. Cape Verde Islands lines) carry different mutations, disrupting the expression of either one of the duplicates of HPA, while the alternate duplicate gene is still functional. In a cross of the two lines, recombinant offspring carrying the two mutated loci in homozygous form have no functional copy of HPA and abort as embryos. Similar epistatic interactions are often found to depress the fitness of progeny from crosses between distant populations, with deleterious effects visible after several generations of interbreeding.

5. IMPLICATIONS FOR CONSERVATION

Global changes alter patterns of gene flow, as well as their putative consequences. In particular, gene flow may be critical to helping a species adapt to climate change, when genetic variation already exists somewhere in the range that could foster more rapid adaptation to warming temperature. For many species, habitat loss and fragmentation have resulted in both a drastic reduction in local population sizes and disrupted gene flow between populations. Increased inbreeding in small isolated populations can increase their extinction risk, as was found in the Glanville fritillary butterfly.

Genetic rescue has been proposed as a management option that seeks to increase fitness in a population by introducing unrelated individuals. The Florida panther population had dropped to fewer than 20 individuals in 1995; this small isolated population had accumulated many phenotypic defects, including low sperm quality, suggesting that deleterious mutations had drifted to fixation. Eight females from Texas were translocated in Florida in 1995 and since then the panther population has risen to more than 100, with much reduced frequency of phenotypic defects in panthers of Texas ancestry; however, the success of such genetic rescue must be balanced by the potential risks of translocating individuals across large distance, including the spread of diseases, outbreeding depression, and the swamping of local adaptation. Recurrent introduction of one migrant per generation in small populations with high genetic load is considered sufficient to replenish genetic variation and counter the effects of genetic drift, while still allowing the preservation of local adaptation.