Fitness and adaptation

Concept

Fitness and adaptation

T-018Home BU-104Threads evolution

Statement

Reproductive success shapes the traits of a population.

Why it matters

natural-selection already established that differential survival and reproduction changes trait frequencies across generations; fitness-adaptation supplies the quantitative measure — relative fitness — that makes "differential" precise, and shows both how adaptation accumulates from repeated selection and why it never reaches an idealised, unconstrained optimum. Without a formal fitness measure, statements like "trait X is favoured" remain qualitative; with one, the rate and direction of expected change can be calculated directly.

The result also directly clarifies a common confusion: fitness is not a fixed, intrinsic property of an organism or a species, but a relative, environment-specific quantity, which is why the same trait can be adaptive in one environment and maladaptive in another, and why evidence-common-descent's vestigial structures (retained, historically useful features, now reduced) are exactly what a fitness-driven but historically constrained process predicts, rather than a sign of imperfect design refutin adaptation.

Hypotheses

Fitness is defined relative to other genotypes present in a specific environment, not as an absolute or species-wide property.Without this qualifier, "fitness" would have no well-defined comparison point; the same genotype can have high relative fitness in one environment and low relative fitness in another, so any statement about fitness implicitly assumes a stated environment and comparison group. The trait variation underlying fitness differences is at least partly heritable.If fitness differences existed but were not heritable (\(h^2=0\)), selection would still act every generation, but nothing would accumulate from one generation to the next — offspring trait distributions would not shift toward the fitter parents' values, and no lasting adaptation would result. Trade-offs and developmental/genetic constraints (e.g. antagonistic pleiotropy) mean selection converges toward a local, not necessarily global, fitness optimum.A gene or trait that increases fitness via one function may simultaneously decrease it via another; selection cannot freely search all conceivable trait combinations independently, since many are linked by shared underlying genetic or developmental machinery.

Proof

w_i = \frac{\text{genotype } i\text{'s average reproductive output}}{\text{reference genotype's average reproductive output}}, \qquad s_i = 1-w_i

Relative fitness \(w_i\) rescales a genotype's reproductive output against a chosen reference, and the selection coefficient \(s_i\) measures its relative disadvantage (or, if negative, advantage); this rescaling is what makes fitness comparable across genotypes and, in principle, across different studies of the identical population. A

p_{t+1} = \frac{p_t(1+s)}{1+s\,p_t}

For a haploid locus with a favoured allele of selective advantage \(s\) at frequency \(p_t\), this standard one-locus selection recursion gives the next generation's frequency; because \(s>0\) makes the numerator grow faster than the denominator whenever \(p_t<1\), the favoured allele's frequency rises each generation, converging toward fixation (\(p\to1\)) as long as heritability holds (Hypotheses). B

\text{Because fitness effects are environment-dependent (Hypotheses), the direction and strength of selection on a given trait shift as the environment shifts.}

A trait conferring high relative fitness under one set of conditions can confer low relative fitness under different conditions, so adaptation tracks whatever environment a population is currently in, rather than converging on a single, environment-independent "ideal" form. A

\text{Trade-offs (e.g. antagonistic pleiotropy, resource-allocation constraints) mean increasing fitness via one trait often decreases it via another linked trait.}

Because many traits share underlying genetic or developmental machinery, selection acting to improve one function frequently degrades a linked function; the resulting adaptation is therefore a compromise that balances competing demands, a local optimum given the existing constraints, rather than an unconstrained, all-traits-maximised design. B

R = h^2 S

The breeder's equation: the response to selection \(R\) (change in a trait's mean per generation) equals the heritability \(h^2\) multiplied by the selection differential \(S\) (how much the selected parents' trait mean differs from the population mean); response is therefore bounded by however much heritable variation is actually present, not by an unconstrained search over all conceivable trait values (directly reinforcing Hypotheses' heritability requirement). B

Result

R = h^2 S, \qquad p_{t+1}=\frac{p_t(1+s)}{1+s\,p_t}

Reading. Adaptation is the accumulated, generation-by-generation response of a population's heritable trait distribution to differential relative fitness, with both its rate (via \(s\), \(h^2\)) and its ultimate extent (via trade-offs and available heritable variation) set by measurable quantities.

Scope. Predicts direction and approximate rate of change under selection on existing heritable variation in a roughly constant environment; does not predict an unconstrained global optimum, since trade-offs and a shifting environment (Hypotheses, Step 3–4) generally prevent one from being reached or even existing as a fixed target.

Corollaries & converses

speciation can result when geographically or otherwise isolated populations adapt toward diverging local fitness optima, accumulating enough difference that they can no longer interbreed.
sexual-selection is a specific, sometimes opposing, component of overall fitness (mating success specifically, rather than survival) that can pull a trait in a direction actively working against straightforward survival-based natural selection.
evidence-common-descent's vestigial structures are a direct prediction of Step 4's trade-off/constraint logic: a structure useful to an ancestor, retained in reduced or repurposed form because eliminating it entirely is developmentally costly or unnecessary, rather than evidence against adaptive fitness-driven change.

Fails without

Drop heritability of the fitness-relevant variation (Hypotheses): selection would still act on the phenotypic variation present in every generation, but since offspring would not reliably resemble their fitter parents for the trait in question, the population's trait distribution would not shift over successive generations — by the breeder's equation (Step 5), \(R=h^2S=0\) whenever \(h^2=0\), regardless of how strong selection (\(S\)) itself is.
Drop trade-offs/constraints (t3 Hypothesis): without them, one would predict convergence toward a single, universally "optimal" organism across all lineages and environments; instead, the actual diversity of adaptive solutions observed across taxa, together with the imperfect, historically constrained structures evidence-common-descent documents, directly contradicts an unconstrained-optimum prediction.

Common errors

Treating fitness as an absolute or intrinsic property of a species or organism, rather than a relative, environment- and comparison-group-specific quantity (Hypotheses).
Using "fit" in its colloquial sense (strong, healthy, physically capable) rather than its technical sense (relative reproductive output, Step 1) — a physically weaker or smaller genotype can have higher fitness if it reproduces more successfully in the relevant environment.
Assuming adaptation implies perfection or an optimal design, ignoring the trade-offs and historical/developmental constraints of Step 4 that generally prevent any trait from being simultaneously maximised across every function it affects.
Assuming any trait currently present in a population must itself be adaptive; some traits are byproducts of selection on a different, linked trait (via pleiotropy or developmental constraint) rather than independently favoured for their own effect.

Discussion

R.A. Fisher formalised fitness within a rigorous population-genetic framework in the early 20th century, including his 1930 fundamental theorem of natural selection relating a population's rate of fitness increase to its additive genetic variance in fitness. Classic observed cases of selection acting on measurable fitness differences — the shift in peppered moth colour-morph frequencies during industrial-era England, and beak-size changes tracked across drought years in Darwin's finches on the Galápagos — are among the most frequently cited direct demonstrations of fitness differences producing measurable adaptive change within observable timescales.

Because relative fitness depends on the specific environment and the specific set of competing genotypes present, a genotype's fitness rank can, and sometimes does, reverse entirely if the environment shifts (e.g. a seasonal or year-to-year change in which beak size is favoured); adaptation therefore tracks a moving, not fixed, target over sufficiently long timescales.

Common misconception: that natural selection and adaptation always produce the objectively "best possible" trait value given enough time. Because of Step 4's trade-offs, Step 3's shifting environment, and Step 5's dependence on whatever heritable variation happens to be present, selection generally converges only to a local, constraint-bounded optimum, not a theoretically ideal, unconstrained one.

Worked examples

s=0.10,\quad p_0=0.01

A favoured allele starts rare (frequency 1%) with a 10% selective advantage (\(w=1.10\) relative to the alternative allele). Applying the recursion of Step 2 of the Proof repeatedly traces out how the allele frequency changes generation by generation. A

p_1 = \frac{0.01(1.10)}{1+0.10(0.01)} = \frac{0.011}{1.001} \approx 0.01099

A single generation produces only a small absolute increase from an initially rare allele; repeating the recursion (not shown generation-by-generation here) shows the proportional rate of increase accelerates as \(p_t\) rises through intermediate frequencies, before decelerating again as \(p_t\) approaches fixation — the frequency change is not linear in time, even though \(s\) itself is held constant throughout. B

\text{Selection on a rare favoured allele accelerates before decelerating as } p\to1

Reading. Even a constant, modest selective advantage produces a characteristic sigmoid (S-shaped) trajectory of allele-frequency change over time, slow at both very low and very high starting frequency and fastest through intermediate frequencies.

Scope. Applies to single-locus selection at constant \(s\) in a large (drift-negligible) population; more complex genetic architectures and fluctuating \(s\) require correspondingly more elaborate models.

Problems

Two genotypes have average reproductive outputs of 4.5 and 5.0 offspring respectively. Taking the second as the reference genotype, compute the relative fitness and selection coefficient of the first.
Solution
\(w_1 = 4.5/5.0 = 0.90\). \(s_1 = 1-w_1 = 0.10\): the first genotype has 10% lower relative fitness than the reference.
A trait shows strong heritability (\(h^2=0.6\)) and experiences a selection differential of \(S=2\) units in a given generation. Predict the response to selection, and explain what would happen instead if \(h^2\) were 0.
Solution
\(R=h^2S=0.6\times2=1.2\) units of change in the trait mean in the next generation. If \(h^2=0\), then \(R=0\) regardless of \(S\): selection would still act (parents with extreme trait values would still be favoured), but since the trait would not be reliably inherited by offspring, no shift in the population's trait mean would occur.
A bird species' beak size is favoured to be larger in drought years (when only large, hard seeds are available) and favoured to be smaller in wet years (when small, soft seeds dominate). Explain, using the Hypotheses, why this does not represent a contradiction in the concept of fitness.
Solution
Fitness is defined relative to a specific environment, not as a fixed, universal property of a trait value (Hypotheses). Large beak size has higher relative fitness specifically in the drought-year environment (better able to process hard seeds), while small beak size has higher relative fitness in the wet-year environment (more efficient on small, soft seeds); the reversal reflects the environment-dependence built into the definition of relative fitness (Step 3 of the Proof), not an inconsistency in the concept itself.