Self-Fertilization

1. Reproductive Assurance

1.1. Pre-Mendelian Observations

1.1.1. Darwin’s Experiments

Before anyone knew what a gene was, naturalists and breeders had a strong working knowledge of the effects of inbreeding. Charles Darwin, in his characteristically obsessive fashion, spent over a decade meticulously documenting these effects in plants. His 1876 book, The Effects of Cross and Self Fertilisation in the Vegetable Kingdom, is a monument to experimental design, showing that the offspring of cross-pollination were almost always more vigorous than the offspring of self-pollination.

In one of his most extensive experiments, Darwin worked with the morning glory, Ipomoea purpurea, for eleven years, raising ten generations of self-pollinated and cross-pollinated plants. His method was simple but effective:

He would take a single plant and designate some of its flowers for self-pollination (covering them with a net to keep insects out) and others for cross-pollination (using pollen from a different, unrelated plant).
He would then plant the seeds from both groups on opposite sides of the same pot, ensuring they grew under identical conditions.
He measured their height and tracked their survival and fertility.

The results were stark. The cross-pollinated plants were consistently taller, healthier, and more fertile. The self-pollinated lines became progressively weaker, a phenomenon he called a loss of “constitutional vigour.” This was a direct, quantitative demonstration of what we now call inbreeding depression.

1.1.2. Animal Breeders’ Folk Theories

Animal breeders had reached similar conclusions through trial and error centuries earlier. While figures like Robert Bakewell in the 18th century famously used close inbreeding to fix desirable traits in livestock—creating “prepotent” animals that reliably passed on their characteristics—they also knew it was a risky strategy. Breeders observed that sustained inbreeding often led to “degeneracy”: reduced fertility, higher infant mortality, and a general lack of robustness. The folk model was one of “blood purity.” Inbreeding could concentrate “good blood,” but it could just as easily concentrate “bad blood,” leading to disaster. This was a purely empirical observation of a statistical pattern, lacking a mechanistic explanation until the rediscovery of Mendel’s work.

1.2. The Logic of Selfing

If inbreeding is so bad, why does selfing exist at all? Because sometimes, the alternative to inbreeding isn’t healthy outcrossing; it’s not reproducing at all.

1.2.1. A Toy Model of Pollinator Failure

Consider a plant population that is entirely dependent on a specific pollinator to reproduce. Now, imagine this pollinator is unreliable and only shows up in 10% of reproductive seasons.

An obligate outcrosser (a plant that must be cross-pollinated) will fail to produce any seeds in 9 out of 10 seasons. Its long-term fitness is crippled by this bottleneck.
Now, introduce a mutant allele that allows for self-fertilization. In the 10% of seasons with pollinators, this plant can still outcross. But in the 90% of seasons without pollinators, it can self-pollinate and produce seeds.

Even if these self-pollinated seeds suffer from severe inbreeding depression (e.g., have only 50% of the viability of outcrossed seeds), the ability to reproduce at all in those nine barren seasons provides an enormous fitness advantage. The mutant allele for selfing will rapidly spread through the population. This guarantee of producing offspring, even at a reduced quality, is known as reproductive assurance. It is the single most consequential short-term advantage of self-fertilization and the primary reason it evolves.

1.2.2. The Transmission Advantage

There’s a sneakier, arithmetic advantage to selfing, often called the “cost of outcrossing” or, more accurately, the transmission advantage. From the perspective of a single gene in a hermaphroditic organism, outcrossing is wasteful.

When an organism outcrosses, it contributes only 50% of the genes to its offspring (one set of chromosomes). The other 50% comes from its partner.
When an organism self-fertilizes, it acts as both mother and father. It contributes 100% of the genes to its offspring.

An allele that promotes selfing, all else being equal, gets twice the representation in the next generation compared to an allele that promotes outcrossing. This provides a strong selective pressure favoring the evolution of selfing, which must be counterbalanced by the fitness costs of inbreeding depression.

1.3. Edge Cases

The concept of “selfing” has some interesting edge cases that test our definitions.

1.3.1. Geitonogamy

Geitonogamy is the transfer of pollen between different flowers on the same plant. Genetically, this is identical to self-pollination of a single flower; the resulting offspring are just as inbred.

Ecologically, however, it’s a completely different beast. A plant with a single, self-pollinating flower doesn’t need to invest in attracting pollinators. But a plant that engages in geitonogamy often does. It produces nectar and showy flowers to attract an insect, which then flies from one flower to another on the same plant. The plant pays all the costs of advertising for outcrossing but gets none of the genetic benefits. This is a coordination failure: the pollinator, acting as a selfish agent, minimizes its travel time by staying on one resource-rich plant, inadvertently causing the plant to inbreed against its own long-term interests.

1.3.2. Mixed Mating Systems

Many organisms don’t commit to just one strategy. They engage in mixed mating, switching between selfing and outcrossing depending on environmental conditions. The freshwater snail Physa acuta is a classic example.

These hermaphroditic snails prefer to outcross when possible. But if an individual finds itself isolated or in a very low-density population, finding a mate is difficult. In this scenario, it switches to self-fertilization as a form of reproductive assurance. The snail’s mating system is thus a conditional strategy, tuned by an environmental parameter: the probability of finding a mate. This flexibility allows it to thrive, both as a colonizer of new habitats (where it can self-fertilize to establish a population from a single individual) and in dense, established populations (where it can outcross to avoid inbreeding depression).

2. The Arithmetic of Inbreeding

The genetic consequences of self-fertilization are not mysterious; they are a matter of simple, relentless arithmetic. Gregor Mendel, ironically, chose to work with pea plants (Pisum sativum) precisely because they are almost exclusively self-fertilizing. This created “true-breeding” lines, allowing him to uncover the mathematical laws of inheritance that, in turn, perfectly explain the mechanics of selfing.

2.1. Mendelian Mechanics

2.1.1. Halving Heterozygosity

Let’s start with a single individual that is heterozygous at a particular gene locus, with genotype Aa. This is generation 0 (G_0). The proportion of heterozygotes in our “population” is 100%.

What happens when this individual self-fertilizes? According to Mendel’s law of segregation, it produces gametes with A and a alleles in equal proportion. The resulting offspring (G_1) will have genotypes in the classic Mendelian ratio:

25% AA (homozygous dominant)
50% Aa (heterozygous)
25% aa (homozygous recessive)

The proportion of heterozygotes just dropped from 100% to 50% in a single generation.

Now let’s go to the next generation (G_2). The homozygous individuals (AA and aa) can only produce homozygous offspring when they self-fertilize (AA produces only AA, aa produces only aa). The heterozygous individuals (now 50% of the population) will again produce offspring with 50% heterozygosity.

The proportion of heterozygotes in G_2 will be 50% (the proportion of heterozygotes in G_1) * 50% = 25%.

This pattern holds indefinitely. After n generations of selfing, the proportion of heterozygotes (H_n) is given by the formula:

H_n = H_0 * (1/2)^n

Where H_0 is the initial proportion of heterozygotes. With each generation, the proportion of heterozygotes is halved, and the proportion of homozygotes (AA and aa) correspondingly increases, eventually approaching 100%.

2.1.2. The Genomic Lock-in

This relentless march to homozygosity has a profound effect on the entire genome. Recombination (crossing over) generates novel combinations of alleles, but it can only do so at loci where an individual is heterozygous. If a chromosome is homozygous from end to end, recombination is functionally invisible; swapping identical segments of DNA changes nothing.

As selfing rapidly eliminates heterozygosity, it dramatically reduces the effective rate of recombination. Genes that are physically located on the same chromosome become more tightly linked than they would be in an outcrossing population. The chromosome begins to be inherited as a single, indivisible block, or “haplotype.” This creates non-random associations between alleles at different loci, a state known as linkage disequilibrium. This is a critical failure mode, as it prevents selection from acting on individual alleles independently. A beneficial allele that arises on a chromosome carrying a handful of slightly bad alleles may be unable to escape its detrimental neighbors.

2.2. Formalizing Inbreeding

The concepts intuited by breeders were formalized in the early 20th century by pioneers of population genetics like Sewall Wright.

2.2.1. Wright’s Inbreeding Coefficient, F

Sewall Wright developed the inbreeding coefficient, F, to quantify the effects of non-random mating. F is defined as the probability that the two alleles at any given locus in an individual are identical by descent (IBD).

This means the two alleles are not just the same state (e.g., both A), but are physical copies of the very same allele from a recent common ancestor. F ranges from 0 (for an individual whose parents are completely unrelated in a theoretically infinite population) to 1 (for a completely homozygous individual produced by a long line of selfing).

2.2.2. Calculating F

F can be calculated by tracing the paths of alleles through a pedigree.

Self-fertilization: Consider an offspring produced by a selfing parent. The inbreeding of the offspring (F_offspring) depends on the inbreeding of the parent (F_parent). The standard formula is F_offspring = 0.5 * (1 + F_parent). If the parent is not inbred (F_parent = 0), the probability that two of its gametes carry alleles that are IBD is 1/2. Thus, the offspring’s inbreeding coefficient is F = 0.5.
Full-sibling mating: Now consider the offspring of a brother-sister mating. They share two parents (the grandparents of the offspring). There are two paths by which an allele from a grandparent can end up in the offspring in two copies: one through the mother and one through the father. For each grandparent, the probability of this happening is (1/2)^3 = 1/8. Since there are two grandparents, the total probability is 1/8 + 1/8 = 1/4. Thus, the offspring has F = 0.25.

This confirms that self-fertilization is the most extreme form of inbreeding possible.

2.2.3. F and Heterozygosity

There is a direct mathematical relationship between the inbreeding coefficient F and the level of heterozygosity in a population. The observed heterozygosity (H) is reduced in proportion to F compared to the expected heterozygosity under random mating (H_0, which is 2pq from the Hardy-Weinberg principle).

H = H_0 * (1 - F)

This formula elegantly links the pedigree-based concept of identity by descent (F) to the population-level observation of reduced genetic diversity (H). When F=0 (no inbreeding), H = H_0. When F=1 (complete inbreeding), H = 0.

3. The Genetic Load

Why is the reduction in heterozygosity so often a bad thing? The phenomenon of inbreeding depression—the reduced survival and fertility of inbred offspring—is one of the most widely observed patterns in evolutionary genetics. It happens because inbreeding exposes the “bad code” that all genomes carry.

3.1. Models of Inbreeding Depression

For decades, two competing hypotheses sought to explain the mechanism of inbreeding depression.

3.1.1. Dominance vs. Overdominance

The Dominance Hypothesis: This is the now-favored explanation. It posits that populations harbor a multitude of deleterious recessive alleles. These are mutations that are harmful, but only when present in two copies (e.g., genotype aa). In large, outcrossing populations, these alleles are rare and mostly exist in heterozygous individuals (Aa), where their harmful effects are masked by the functional dominant allele (A). Selfing and other forms of inbreeding increase the frequency of homozygotes (aa), unmasking these deleterious alleles and reducing the fitness of the individual.
The Overdominance Hypothesis: This hypothesis claims that heterozygotes (Aa) are intrinsically superior in fitness to both homozygotes (AA and aa). This is also known as heterozygote advantage. In this view, inbreeding depression is caused by the loss of these superior heterozygous genotypes as the population becomes more homozygous. A classic, though rare, example is the sickle-cell allele in humans, where heterozygotes are resistant to malaria.

3.1.2. Evidence from Mutation Accumulation

The dominance hypothesis is now considered the primary driver of inbreeding depression for the majority of traits in most species.

The key evidence comes from mutation accumulation experiments. In these experiments, researchers create many parallel lineages of an organism (like Drosophila or C. elegans) and allow them to reproduce under conditions of minimal natural selection. This allows mutations, both beneficial and deleterious, to accumulate randomly. After many generations, these lines are analyzed. The results consistently show that the vast majority of mutations with a fitness effect are deleterious, and these deleterious effects are, on average, partially to fully recessive. There is very little evidence for widespread overdominance. Inbreeding depression, therefore, seems to be caused not by the loss of intrinsically superior heterozygotes, but by the expression of a vast underlying reservoir of bad recessive code.

3.2. Quantifying the Load

The “bad code” carried by a population is its genetic load. This can be defined as the proportional reduction in the mean fitness of a population compared to a hypothetical optimal genotype.

3.2.1. Mutation-Selection Balance

In any large population, new deleterious mutations are constantly arising at a certain rate (μ). Natural selection constantly works to remove them at a rate determined by the selection coefficient (s), which measures the fitness reduction of the deleterious genotype.

For a deleterious recessive allele (a), selection only “sees” the aa homozygote. In a large, outcrossing population, the equilibrium frequency of this allele (q) is reached when the rate of new mutations equals the rate of removal by selection. This is the mutation-selection balance, and the equilibrium frequency is approximately:

q = sqrt(μ / s)

Because the allele is recessive, it can “hide” in heterozygotes, so its frequency is determined by the square root of the mutation/selection ratio. For a weakly deleterious allele (small s), the equilibrium frequency can be surprisingly high.

3.2.2. First-Generation Shock

What happens when our large, outcrossing population is suddenly forced to self-fertilize? The allele frequencies (p and q) don’t change in the first generation, but the genotype frequencies do. The proportion of aa homozygotes, which was q^2 under outcrossing, abruptly increases as Aa heterozygotes (with frequency 2pq) self-fertilize to produce 1/4 aa offspring.

The expressed genetic load skyrockets. All those deleterious recessive alleles that were hiding in heterozygotes are suddenly exposed to selection. The mean fitness of the population plummets. This immediate and drastic reduction in fitness is the primary barrier to the evolution of selfing. A lineage must somehow survive this initial shock to reap the long-term benefits of purging.

3.2.3. Lethal Equivalents

A practical proxy used to quantify genetic load is the concept of lethal equivalents, a term coined by Morton, Crow, and Muller in 1956. One lethal equivalent is defined as a set of deleterious alleles that, if made homozygous, would cause one death on average. This could be a single allele that is 100% lethal when homozygous, or two different alleles that are each 50% lethal, and so on.

Studies measuring lethal equivalents typically involve controlled inbreeding experiments and tracking survival rates.

In humans, estimates suggest each individual carries, on average, 1 to 2 lethal equivalents. This is the genetic basis for the universal taboo against incest.
In Drosophila, the number is often estimated to be between 1 and 2 per individual as well.

This proxy can be biased. It often only measures mortality before reproductive age and may not capture effects on fertility. Furthermore, it assumes that the fitness effects of different deleterious alleles are independent, ignoring potential epistasis (interactions between genes) which could make the combined effect better or worse than the sum of its parts.

4. Self-Incompatibility

Given the severe costs of inbreeding, it’s not surprising that many organisms have evolved mechanisms to prevent it. In plants, the most common system is self-incompatibility (SI), a form of genetically controlled mate recognition that allows a plant to reject its own pollen. Think of it as a distributed molecular security system designed to enforce outcrossing.

4.1. Molecular Recognition

SI systems are typically controlled by a single genetic locus, the S-locus, which is highly polymorphic (meaning it has dozens or even hundreds of different alleles in a population). The basic rule is: if the S-allele of the pollen matches an S-allele in the pistil (the female part of the flower), fertilization is blocked. The specific molecular implementation of this rule, however, differs.

4.1.1. Gametophytic vs. Sporophytic SI

The key distinction lies in whose genotype determines the pollen’s “identity.”

Gametophytic Self-Incompatibility (GSI): The pollen’s phenotype is determined by its own haploid genotype (the gamete).
Sporophytic Self-Incompatibility (SSI): The pollen’s phenotype is determined by the diploid genotype of the parent plant that produced it (the sporophyte).

Feature	Gametophytic SI (GSI)	Sporophytic SI (SSI)
Pollen Identity	Determined by the pollen’s own haploid S-allele.	Determined by the diploid S-genotype of the parent plant.
Example Family	Solanaceae (tomatoes, tobacco)	Brassicaceae (cabbage, broccoli)
Site of Inhibition	Pollen tube growth is arrested inside the style.	Pollen germination is blocked on the stigma surface.
Mechanism Analogy	Each pollen grain carries its own passport.	All pollen grains from a plant carry copies of the parent’s two passports.

4.1.1.1. The GSI Mechanism

In a GSI system, a pollen grain with allele S1 cannot fertilize a style from a plant with genotype S1S2. Why?

The pistil produces and secretes proteins called S-RNases into the style. Each S-RNase corresponds to a specific S-allele (e.g., S1-RNase and S2-RNase). When a pollen tube begins to grow down the style, it takes up these S-RNases. The pollen, being haploid, expresses its own S-allele product, an F-box protein.

The current model suggests that the F-box protein acts as an antidote. It recognizes and degrades all non-matching S-RNases.

An S1 pollen grain on an S1S2 style takes up both S1-RNase and S2-RNase.
Its S1 F-box protein successfully neutralizes the S2-RNase.
However, it cannot neutralize the matching S1-RNase.
The active S1-RNase then acts as a cytotoxin, destroying the pollen’s RNA and arresting its growth.

4.1.1.2. The SSI Mechanism

In an SSI system, all pollen from an S1S2 plant is rejected by an S1S2 stigma, regardless of whether a specific pollen grain carries the S1 or S2 allele. How does the parent’s diploid genotype control the pollen’s identity?

The key is the pollen coat. This outer layer is not produced by the haploid pollen grain itself, but is deposited onto its surface by the diploid tapetum tissue of the parent plant. This coat contains proteins encoded by both of the parent’s S-alleles.

The male determinant is a protein called SCR (S-locus Cysteine-Rich protein).
The female determinant is a receptor kinase on the stigma surface called SRK (S-locus Receptor Kinase).

When pollen from an S1S2 plant lands on an S1S2 stigma:

The pollen coat contains both S1-SCR and S2-SCR proteins.
The stigma has both S1-SRK and S2-SRK receptors.
The binding of a matching pair (e.g., S1-SCR to S1-SRK) triggers a signaling cascade inside the stigma cell that prevents the pollen grain from hydrating and germinating. Fertilization is blocked before it even starts.

4.2. Game Theory of S-Alleles

4.2.1. Negative Frequency-Dependent Selection

There is extremely strong negative frequency-dependent selection on S-alleles. This means that an allele’s fitness is inversely proportional to its frequency in the population.

A rare S-allele is a golden ticket. Pollen carrying it can fertilize almost any plant it lands on, because very few plants will share that rare allele.
A common S-allele is a curse. Pollen carrying it will be rejected by a large fraction of the plants in the population, dramatically reducing its mating success.

This dynamic creates a constant pressure that favors rare alleles and punishes common ones, leading to the stable maintenance of dozens or even hundreds of different S-alleles in a single population. It is one of the most powerful balancing selection mechanisms known in nature.

4.2.2. Invasion of the Self-Compatible Mutant

What happens if a mutation breaks the SI system? A plant with a “self-compatible” (SC) allele can now self-fertilize. Despite causing inbreeding depression, this mutant can successfully invade and even take over the population under specific ecological conditions:

Pollinator Scarcity: As modeled in section 1.2.1, if pollinators are unreliable, the reproductive assurance offered by selfing can outweigh the costs of inbreeding.
Colonization Events: When a single individual colonizes a new habitat (like an island), it has no one to mate with. An SC mutant can establish an entire population from a single founder. This makes self-compatible species, which can reproduce with a single individual, more likely to establish sexually reproducing populations after long-distance dispersal than self-incompatible species. This idea is known as Baker’s Law, articulated by Herbert Baker in 1955:

With self-compatible individuals a single propagule is sufficient to start a sexually-reproducing colony, making its establishment much more likely than if the chance growth of two self-incompatible yet cross-compatible individuals sufficiently close together spatially and temporally is required.

4.2.3. Multilevel selection conflict

The breakdown of SI is a classic example of a conflict between selection at different levels.

Individual-level selection: Favors the SC mutant for its short-term benefit of reproductive assurance. An individual plant that can guarantee seed production, even if inbred, outcompetes one that fails to reproduce at all.
Population-level viability: Favors the maintenance of outcrossing via SI. Outcrossing maintains the genetic variation that is crucial for long-term adaptation to changing environments, like evolving parasites (see the Red Queen Hypothesis).

The transition to self-compatibility is a short-term individual win that often leads to a long-term population-level loss. It is a failure of alignment between the interests of the individual and the long-term viability of the collective.

5. Purging the Genetic Load

If inbreeding depression is so severe, how does any species successfully make the transition to selfing? The answer is that sustained inbreeding, while initially costly, can eventually “purge” the genetic load that causes the problem.

5.1. How Purging Works

5.1.1. Exposing Deleterious Alleles

Purging works by leveraging the mechanism that causes inbreeding depression: the increase in homozygosity. As deleterious recessive alleles are forced out of their heterozygous hiding places and into homozygous genotypes, they become visible to natural selection. Individuals carrying the aa genotype have lower fitness and are less likely to reproduce, so the a allele is gradually removed—or purged—from the population. An outcrossing population can tolerate a large genetic load because it’s mostly hidden. A selfing population cannot; it is forced to clean up its genetic code.

5.1.2. The Efficiency of Purging

The efficiency of purging depends directly on the strength of selection against the deleterious allele. In a simple model with one deleterious recessive allele, the speed of purging is a function of the selection coefficient s.

If s is large (e.g., s = 1 for a recessive lethal allele), purging is extremely fast and efficient. The aa homozygotes are produced, they die, and the a allele is rapidly eliminated.
If s is small (e.g., s = 0.01 for a weakly deleterious allele), purging is very slow and inefficient. The fitness difference between the normal homozygote and the deleterious one is so slight that selection has a hard time “seeing” it.

5.1.3. The Problem of Weakly Deleterious Alleles

This leads to a crucial insight: purging is highly effective against the most severe mutations but much less so against weakly deleterious ones.

Strongly deleterious alleles (lethals, steriles) are purged quickly. A selfing lineage can effectively rid itself of its worst genetic baggage.
Weakly deleterious alleles, however, pose a major problem. Each one has only a tiny effect on fitness, making selection against it weak. Genetic drift can easily overpower such weak selection, especially in the smaller effective population sizes typical of selfing lineages.

The consequence is that while a selfer may purge its lethal equivalents, it can accumulate a large burden of many weakly deleterious mutations. The cumulative effect of these thousands of tiny defects can still result in a significant overall fitness reduction.

5.2. Evidence for Purging

5.2.1. Comparative Studies

A powerful way to test for purging is to compare the magnitude of inbreeding depression in closely related species or populations that differ in their mating system. For example, the plant Arabidopsis lyrata has both selfing and outcrossing populations. When individuals from both types of populations are experimentally inbred, the naturally selfing populations consistently show lower levels of inbreeding depression. This suggests their history of inbreeding has already purged a significant portion of their genetic load. The self-compatible model plant Arabidopsis thaliana is a highly successful selfer that is thought to have undergone a severe population bottleneck, which likely helped purge its genetic load and enabled its global spread.

5.2.2. Genomic Signatures

With modern genome sequencing, we can look for the footprint of purging directly in the DNA. Genome sequencing studies comparing selfing and outcrossing lineages generally confirm the theoretical predictions:

Selfing lineages show a lower frequency of predicted strongly deleterious recessive mutations compared to their outcrossing relatives. The purge is effective against the big-ticket problems.
However, selfing lineages often show an excess of weakly deleterious mutations. The reduced effective population size that accompanies selfing makes selection less efficient overall, allowing these small-effect mutations to accumulate through drift.

This creates a mixed genomic picture: selfers are “cleaner” in terms of severe genetic diseases but “messier” in terms of accumulated minor defects.

6. The Evolutionary Dead End

The long-term evolutionary fate of selfing lineages appears to be bleak. The transition to self-fertilization is often described as an “evolutionary dead end.” The idea, first articulated by G. Ledyard Stebbins in 1957, is that while selfing provides short-term advantages, it ultimately leads to a higher rate of extinction, making it an unsustainable strategy over geological timescales.

6.1. Mechanisms of Extinction

6.1.1. Loss of Effective Recombination

As discussed in section 2.1.2, the extreme homozygosity in selfing lineages cripples the effectiveness of recombination. Genes become tightly linked, and the genome is inherited in large, non-recombining blocks. This has severe downstream consequences.

6.1.2. Muller’s Ratchet and Adaptation Rate

The lack of effective recombination makes selfing lineages vulnerable to the same problems that plague asexual organisms:

Muller’s Ratchet: Without recombination to separate alleles, deleterious mutations accumulate irreversibly. The class of individuals with the fewest mutations can be lost by chance and can never be recreated. The genome slowly degrades over time.
The Fisher-Muller Effect: Adaptation is slowed because beneficial mutations that arise in different individuals cannot be easily combined into a single, highly-fit genotype. The lineage must wait for multiple beneficial mutations to occur sequentially in the same line, a much slower process.

6.1.3. The Red Queen’s Race

Reduced genetic variation also makes selfers a sitting duck for co-evolving parasites. The Red Queen hypothesis posits that organisms are in a constant evolutionary arms race with their pathogens. Outcrossing and recombination are crucial weapons in this race, as they constantly generate novel combinations of resistance genes. Selfing lineages, with their low genetic diversity, produce genetically uniform offspring. Once a parasite evolves to overcome the defenses of one individual, it can potentially wipe out the entire population.

6.2. Evidence for the “Dead End”

6.2.1. Asymmetric Transitions

The strongest evidence for the dead end hypothesis comes from phylogenetic studies. If selfing were a stable, long-term strategy, we would expect to see transitions from outcrossing to selfing and back again at roughly equal rates. This is not what we see.

Phylogenetic trees consistently show that transitions to selfing are very common and have occurred independently hundreds of times across the plant kingdom.
Transitions from selfing back to outcrossing are extremely rare, if they occur at all.

Modern analyses, such as Igic et al. (2008), have confirmed this strong asymmetry. The evolutionary path appears to be a one-way street.

6.2.2. Irreversibility Ratchet

Why is the transition irreversible? Once a lineage becomes self-compatible and predominantly selfing, the complex genetic machinery of self-incompatibility is no longer maintained by selection. The S-locus genes accumulate mutations and degrade into non-functional pseudogenes. The high diversity of S-alleles, which is essential for the system to work, is also rapidly lost. Re-evolving such a complex, polymorphic recognition system from scratch appears to be evolutionarily impossible.

6.2.3. Nearest True Thing

The modern consensus is a refinement of Stebbins’ original idea: Selfing is a highly successful short-term strategy that leads to long-term failure. It’s a winning strategy for losers.

Selfing evolves frequently because its short-term advantages (reproductive assurance, transmission advantage) are immediate and powerful, especially for colonizing new environments. This is why we see so many selfing species.
However, the long-term genomic consequences (loss of recombination, accumulation of deleterious mutations, inability to adapt) lead to higher rates of extinction.

The result is a “tippy” distribution on the tree of life: most selfing species are found at the tips of the phylogeny, indicating they are of recent evolutionary origin. The older selfing lineages have simply gone extinct. Selfing isn’t an evolutionary dead end in the sense that it causes immediate doom, but rather in the sense that it puts a lineage on a highly canalized trajectory with a single eventual outcome: extinction.

Tags