{
  "id": "evolutionary-fitness",
  "title": "Evolutionary Fitness",
  "lang": "en",
  "source_lang": "en",
  "summary": "Evolutionary fitness measures reproductive contribution to future generations in a specified environment and underlies explanations of natural selection.",
  "category": "science",
  "subcategory": "biology",
  "type": "concept",
  "aliases": [
    "Darwinian fitness",
    "biological fitness"
  ],
  "generated_at": "2026-11-03",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/evolutionary-fitness/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/evolutionary-fitness/",
      "json": "https://aiwiki.page/zh/evolutionary-fitness.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Evolutionary fitness is the reproductive contribution of an organism, genotype, or other biological entity to subsequent generations, assessed in a specified environment and over a specified interval. It is the central comparative quantity in natural selection: when heritable variants differ in their expected reproductive contributions, their frequencies can change systematically. Fitness therefore concerns evolutionary success, not simply strength, health, longevity, or complexity. An organism that survives exceptionally well but leaves no descendants may have lower fitness than one that lives briefly and reproduces abundantly.\n\nFitness connects the mechanisms of genetics with the dynamics of evolution. Depending on the question, it may be expressed as expected offspring number, relative reproductive success, population growth rate, or contribution to a lineage’s distant future. These measures are related but are not interchangeable under all conditions. Their interpretation depends on generation time, inheritance, population structure, environmental variation, and interactions among organisms.\n\nThe concept also has important limits. Fitness is not a permanent property of a species or an intrinsic ranking of organisms: a variant advantageous in one setting may be disadvantageous in another. Distinguishing expected fitness from realized reproductive outcomes, and identifying the appropriate unit and timescale of comparison, are essential to its use in evolutionary explanation."
    },
    {
      "heading": "Historical development and conceptual foundations",
      "text": "Charles Darwin developed the theory of natural selection in On the Origin of Species in 1859 without the mathematical definition of fitness used in modern population biology. His account emphasized variation, inheritance, competition, and differential success in leaving offspring. Herbert Spencer introduced the phrase “survival of the fittest” in 1864, and Darwin incorporated it into the fifth edition of the Origin in 1869. Despite its wording, survival alone was never sufficient to capture the evolutionary mechanism: survival matters because it can affect reproduction.\n\nThe development of population genetics in the early twentieth century made fitness quantitatively explicit. Ronald A. Fisher, J. B. S. Haldane, and Sewall Wright constructed models linking Mendelian inheritance to changes in variant frequencies. Their work showed how modest differences in survival or fertility could produce substantial evolutionary change over many generations. Fitness became a parameter describing the expected reproductive performance of genetic types rather than a loose synonym for biological superiority.\n\nAn important distinction subsequently emerged between realized reproductive success and expected fitness. Two genetically identical organisms in similar circumstances can leave different numbers of offspring because of chance events. One may die in an accidental disturbance while another reproduces. Their observed outcomes differ, but those outcomes need not establish a difference in their underlying reproductive capacities. A propensity interpretation treats fitness as an expectation over relevant possible outcomes, conditional on the organism’s characteristics and circumstances.\n\nThis distinction helps address the charge that natural selection is circular: if “the fittest” meant only “those that reproduce most,” saying that the fittest reproduce most would add little explanation. Evolutionary studies instead seek independently testable mechanisms. A difference in beak structure, for example, may alter the ability to process particular seeds; food acquisition may then affect survival and reproduction. The proposed connection can be examined before or independently of the final reproductive outcome.\n\nFitness nevertheless requires a reference context. It can be assigned to an individual, a genotype, an allele, or a lineage, but each assignment answers a different question. Genotype fitness typically averages over individuals carrying that genotype. Allele fitness may average over the different genotypes and genetic backgrounds in which the allele occurs. These averages can change even when the allele itself does not, because its partners, competitors, and environment change. Consequently, a statement that a variant is “fitter” is incomplete unless the comparison and relevant conditions are specified."
    },
    {
      "heading": "Absolute fitness, relative fitness, and selection coefficients",
      "text": "In a simple population with nonoverlapping generations, absolute fitness, commonly denoted \\(W_i\\), is the expected contribution of a member of type \\(i\\) to the next generation. For an asexual organism, it can be measured as the expected number of descendants reaching the same census stage. If an individual produces an average of four offspring but only half survive to that stage, its absolute fitness under this definition is two. The census stage matters: counting eggs and counting reproductive adults measure different portions of the life cycle.\n\nRelative fitness compares this contribution with a reference. One common convention divides absolute fitness by the population mean:\n\n\\[w_i=\\frac{W_i}{\\overline{W}}, \\qquad \\overline{W}=\\sum_i p_iW_i,\\]\n\nwhere \\(p_i\\) is the frequency of type \\(i\\). Under this convention, mean relative fitness equals one. Another convention assigns the highest-fitness type a value of one and expresses other types as fractions of it. Both conventions are valid, but their numerical values cannot be compared without checking the normalization.\n\nSuppose two asexual types have absolute fitnesses of three and two. Relative to the better-performing type, their fitnesses are \\(1\\) and \\(2/3\\). If the less successful type is written as having fitness \\(1-s\\), its selection coefficient is \\(s=1/3\\). This number is a relative reproductive disadvantage under the stated conditions, not necessarily a probability of death. It might result from lower survival, reduced mating success, fewer offspring, or several mechanisms acting together.\n\nThe distinction between absolute and relative fitness separates population growth from evolutionary replacement. A type with absolute fitness \\(0.9\\) can increase in frequency if its competitors have fitness \\(0.8\\), even though both types decline in number. Conversely, a type producing many descendants can lose frequency if competing types produce still more. An increase in relative success therefore does not guarantee demographic security.\n\nFor continuous reproduction, fitness is often represented by a per-capita growth rate \\(r_i\\). If abundance follows \\(N_i(t)=N_i(0)e^{r_it}\\), differences in \\(r_i\\) determine changes in relative abundance. Over a fixed interval \\(\\Delta t\\), the corresponding multiplication factor is \\(e^{r_i\\Delta t}\\). A growth rate has units of inverse time, whereas a multiplication factor over a defined interval is dimensionless. Comparing fitness estimates consequently requires attention not only to the organism and environment but also to whether fitness is measured per generation or per unit time."
    },
    {
      "heading": "Fitness in population-genetic dynamics",
      "text": "The simplest selection equation describes a haploid or asexual population in which types reproduce faithfully and differ only in fitness. If type \\(i\\) initially has frequency \\(p_i\\), then after reproduction its expected frequency is\n\n\\[p_i'=\\frac{p_iW_i}{\\overline{W}}.\\]\n\nThis equation makes the comparative nature of selection explicit. A type increases when its fitness exceeds the population mean, not when it exceeds an externally fixed threshold. The equation describes deterministic expectations; finite populations can depart from them because of random sampling.\n\nIn a sexually reproducing diploid population, alleles occur in pairs. For two alleles, \\(A\\) and \\(a\\), with frequencies \\(p\\) and \\(q=1-p\\), suppose random mating produces genotype frequencies \\(p^2\\), \\(2pq\\), and \\(q^2\\) before selection. If the corresponding fitnesses are \\(W_{AA}\\), \\(W_{Aa}\\), and \\(W_{aa}\\), then\n\n\\[\\overline{W} =p^2W_{AA}+2pqW_{Aa}+q^2W_{aa},\\]\n\nand the frequency of \\(A\\) after selection is\n\n\\[p'=\\frac{p^2W_{AA}+pqW_{Aa}}{\\overline{W}}.\\]\n\nThe contribution of an allele depends on both its homozygous and heterozygous settings. This is why an allele does not generally possess a single fitness value independent of its population.\n\nDominance affects how readily selection acts on rare variants. A rare recessive allele occurs mostly in heterozygotes, where its phenotypic effect may be masked. A harmful recessive can therefore persist at low frequency, while a beneficial recessive may initially increase slowly. If the heterozygote has higher fitness than either homozygote, selection can maintain both alleles. The classic example involves the sickle-cell hemoglobin variant in malaria-endemic environments: heterozygosity can reduce the risk of severe malaria, whereas homozygosity for the variant causes sickle-cell disease. The resulting fitness relationships depend on the disease environment and other conditions.\n\nSelection also operates alongside mutation, gene flow, and genetic drift. Mutation introduces variants, migration redistributes them, and drift changes frequencies through random sampling. A favorable allele is therefore not guaranteed to spread to fixation. When rare, it can disappear before its expected advantage has much opportunity to influence population composition.\n\nFisher’s fundamental theorem, introduced in 1930, relates the component of change in mean fitness attributable to natural selection to the additive genetic variance in fitness, under the theorem’s definitions. It is not a general claim that total mean fitness must always increase. Environmental deterioration, changing competitors, transmission effects, and other processes can offset the selection component. Even within genetics, the precise life cycle and assumptions determine which simple monotonic results apply."
    },
    {
      "heading": "Life histories, reproductive timing, and demographic value",
      "text": "Fitness is assembled across the life cycle. Survival to maturity, age at first reproduction, mating success, offspring production, and offspring survival can all contribute. These components are connected by trade-offs: resources committed to current reproduction may reduce later survival, while delayed reproduction may permit growth that increases future fecundity. Life-history theory examines how such schedules evolve under different ecological and demographic conditions.\n\nFor an age-structured population, let \\(l_x\\) denote survival from birth to age \\(x\\), and \\(m_x\\) expected reproductive output at that age, expressed in a consistent demographic currency. The net reproductive rate is\n\n\\[R_0=\\sum_x l_xm_x.\\]\n\nIn a female-based demographic model, this often represents expected daughters produced by a newborn female over her lifetime. Under suitable assumptions, \\(R_0>1\\) indicates replacement above the parental level. However, \\(R_0\\) does not distinguish early from late reproduction when lifetime totals are equal.\n\nTiming enters through the Euler–Lotka equation:\n\n\\[1=\\sum_x e^{-rx}l_xm_x.\\]\n\nIts solution gives the intrinsic population growth rate \\(r\\) under the specified schedule and assumptions. In a growing population, early offspring generally contribute more to population expansion than the same number produced later because their descendants begin reproducing sooner. Two life histories can therefore have equal lifetime reproductive output but different growth rates. A strategy with fewer offspring per generation may outperform another per unit time if its generation interval is sufficiently shorter.\n\nAge and stage also affect the evolutionary value of survival. The concept of reproductive value, developed by Fisher, weights individuals by their expected contribution to the population’s future. In matrix population models, the dominant eigenvalue of a projection matrix describes long-run multiplication under constant conditions, while the associated left eigenvector provides reproductive-value weights. A surviving adult and a newly produced seed cannot necessarily be counted as equivalent units of future contribution.\n\nThese distinctions help explain why measuring only fecundity can be misleading. In many long-lived organisms, adult survival has a large influence on population growth, whereas in other organisms successful establishment of juveniles is especially consequential. The relative importance of a life-cycle transition depends on the entire schedule, not simply on whether its numerical value is high or low.\n\nFitness comparisons can become more complicated when population density regulates survival or reproduction. The strategy with the highest growth rate in an uncrowded environment need not be the one that can invade an established population at ecological equilibrium. Appropriate analysis then asks whether a rare alternative can increase under the conditions created by the resident population."
    },
    {
      "heading": "Environmental dependence and ecological interactions",
      "text": "Fitness arises from an interaction between organisms and their surroundings. Temperature, resource availability, predators, parasites, and competitors can change which characteristics improve reproductive success. A genotype that performs well in a nutrient-rich laboratory medium may perform poorly under starvation or fluctuating temperatures. Such reversals are instances of genotype–environment interaction when different genotypes respond differently to environmental changes.\n\nThe environment includes other organisms of the same population. Under density-dependent selection, fitness changes with population abundance. Efficient exploitation of scarce resources may matter most near a resource-limited equilibrium, while rapid reproduction may dominate during expansion into an unoccupied habitat. There is consequently no general requirement that the genotype with the fastest isolated growth rate should displace all competitors under crowded conditions.\n\nUnder frequency-dependent selection, fitness depends on the frequencies of alternative types. Negative frequency dependence gives a type an advantage when it is rare and can maintain variation. For example, a rare resource-use strategy may encounter less competition than a common one. In host–parasite interactions, parasites that track common host genotypes can generate changing advantages among host variants. Positive frequency dependence instead favors common types, potentially producing alternative stable states.\n\nEvolutionary game theory formalizes situations in which the reproductive consequences of a strategy depend on the strategies encountered. In the hawk–dove model developed in the 1970s by John Maynard Smith and George Price, the payoff from aggressive behavior depends on the frequency of aggressive opponents and the cost of fighting. An evolutionarily stable strategy is resistant to invasion by sufficiently rare alternatives under the model’s conditions. Stability does not mean that every individual has the highest conceivable payoff or that the population maximizes collective welfare.\n\nSpatial variation adds another layer. A variant may be favored in one habitat and disfavored in another, while dispersal moves copies between them. Local adaptation can arise when selection is strong enough relative to gene flow. Conversely, immigration can maintain variants that are locally disadvantageous. Whether reproductive output is regulated locally before dispersal or globally after dispersal also affects the contribution of different habitats.\n\nThese cases show why fitness is often better represented as a function than as a constant:\n\n\\[W_i=W_i(E,N,\\mathbf{p}),\\]\n\nwhere \\(E\\) represents environmental conditions, \\(N\\) population density, and \\(\\mathbf{p}\\) the frequencies of interacting types. The function may itself change over time as organisms modify resources, habitats, and the abundance of their enemies."
    },
    {
      "heading": "Genetic architecture and fitness landscapes",
      "text": "A fitness landscape maps biological variants to fitness values. Sewall Wright introduced the influential landscape metaphor in 1932 to illustrate relationships among genetic combinations, selection, and population structure. A landscape can be defined over genotypes, phenotypes, or allele-frequency states, but these are distinct constructions. Their axes, neighborhoods, and “peaks” must be specified rather than assumed from the visual metaphor.\n\nIn a genotype landscape, neighboring points often differ by a single mutation. Selection tends to increase variants with higher fitness, but accessibility depends on which mutations occur and how they combine. Epistasis occurs when the effect of one genetic change depends on another. Under sign epistasis, a mutation is beneficial in one genetic background but harmful in another. Reciprocal sign epistasis, in which the signs of both mutations’ effects depend on the other, is a necessary condition for multiple local peaks in standard discrete biallelic genotype landscapes.\n\nSuch interactions have been demonstrated experimentally. A 2006 study by Daniel Weinreich and colleagues examined five changes in a bacterial β-lactamase associated with increased cefotaxime resistance. Among the 120 possible orders of acquiring those changes, only a minority provided strictly increasing resistance at every step. Resistance is not identical to fitness under every condition, but the experiment demonstrated how interactions among mutations can constrain accessible paths under a specified selective regime.\n\nGenetic architecture also links fitness to phenotype. A single gene can affect several traits, while a trait can depend on many genes. A variant that improves one function may impair another: increased reproduction can compete with maintenance, or resistance to a parasite can carry costs when that parasite is absent. Such effects prevent a simple interpretation in which every trait has an independently attainable optimum.\n\nRecombination changes how variants are assembled. It can bring beneficial alleles together from different lineages, but it can also break apart combinations that function well jointly. In asexual populations, beneficial mutations arising in different lineages may compete rather than combine readily, a process called clonal interference. The fitness effect of a mutation therefore does not by itself determine its evolutionary trajectory; linkage and the availability of genetic backgrounds matter.\n\nLandscape pictures are most reliable when their limits remain explicit. Real landscapes can change as environments and population frequencies change. High-dimensional genotype spaces also contain routes and connections that a two-dimensional mountain diagram cannot display. A local peak means that specified neighboring variants perform worse under specified conditions, not that evolution has reached a universally optimal biological design."
    },
    {
      "heading": "Chance, fluctuating environments, and long-term success",
      "text": "Expected reproductive success does not determine every realized outcome. Birth, death, mating, and inheritance all involve stochastic events. A beneficial mutation initially represented by one copy can disappear even when its expected reproductive contribution exceeds that of the resident type. Conversely, a mildly disadvantageous variant can rise in frequency by chance. The relative importance of drift and selection depends strongly on effective population size and on the size of the fitness difference.\n\nA frequently used qualitative criterion compares the magnitude of selection with the inverse of effective population size. When the relevant product of effective population size and selection strength is much greater than one, selection can have a strong directional effect; when it is much smaller than one, drift can dominate. Exact thresholds and coefficients depend on ploidy, dominance, population structure, and the model. There is no universal selection coefficient below which a variant is necessarily neutral.\n\nEnvironmental fluctuations introduce a different source of uncertainty. Consider a lineage whose population multiplies by factors \\(W_1,W_2,\\ldots,W_T\\) across successive periods. Its total multiplication is their product, and its long-run logarithmic growth rate is\n\n\\[g=\\lim_{T\\to\\infty}\\frac{1}{T}\\sum_{t=1}^{T}\\ln W_t,\\]\n\nwhen the limit exists. This makes the geometric mean, rather than the arithmetic mean, central to long-term growth in this simple multiplicative setting.\n\nFor example, a lineage multiplying by \\(2\\) in one period and \\(0.5\\) in the next has arithmetic mean multiplication \\(1.25\\), but its two-period product is \\(1\\). A lineage multiplying by \\(1.1\\) in both periods has a lower arithmetic mean but a two-period product of \\(1.21\\). Repeated exposure to the same sequence favors the second lineage in long-run growth. Severe losses can therefore outweigh occasional large gains.\n\nThis principle underlies evolutionary bet-hedging: strategies can be favored because they reduce variation in lineage success, even at a cost to average output in a single period. Seed dormancy is an important example. A genotype that leaves some seeds dormant may sacrifice reproduction in a favorable year but retain descendants capable of surviving an unfavorable sequence.\n\nThe simple geometric-mean argument nevertheless has assumptions. Individual-level randomness is not identical to an environmental shock affecting every individual simultaneously. Spatially distributed descendants may experience partially independent conditions, and finite populations face extinction boundaries absent from deterministic growth equations. Dormant stages, migration, correlations between years, and density regulation can all alter which measure of stochastic fitness predicts evolutionary success."
    },
    {
      "heading": "Inclusive fitness and levels of selection",
      "text": "Reproductive effects need not be confined to the offspring produced by the actor. An individual can alter the survival or reproduction of relatives that carry copies of the same inherited variants. Inclusive fitness, developed by W. D. Hamilton in papers published in 1964, provides a framework for analyzing such effects. It combines direct reproductive consequences with appropriately relatedness-weighted effects on others, while accounting for causal attribution.\n\nA familiar expression is Hamilton’s rule:\n\n\\[rb>c,\\]\n\nwhere \\(c\\) is a reproductive cost to the actor, \\(b\\) a reproductive benefit to a recipient, and \\(r\\) a measure of genetic relatedness appropriate to the model. Under its applicable assumptions, a helping variant can be favored when the relatedness-weighted benefit exceeds the cost. Relatedness here concerns statistical genetic association relative to a reference population, not merely a genealogical label such as sibling or cousin.\n\nThe framework helps explain why behavior that reduces an individual’s direct reproduction can nevertheless evolve. Helping relatives can increase transmission of associated alleles through their offspring. However, neither relatedness alone nor the existence of helping establishes the explanation. Ecological benefits, competition among relatives, dispersal, and the alternative opportunities available to helpers all matter. A reproductive benefit delivered to a relative can be offset if it causes another close relative to lose the same opportunity.\n\nSocial-insect colonies illustrate these complications. Many workers perform tasks without reproducing directly, while queens and other reproductives produce most descendants. Relatedness can be important, but colony ecology and life history are also central. Haplodiploid inheritance is not by itself a sufficient explanation for eusociality: eusocial termites are diploid, and many haplodiploid species are not eusocial.\n\nMultilevel selection describes change by separating processes within groups from differences among groups. A cooperative type may lose ground to noncooperators within each group while groups containing more cooperators contribute more descendants to the total population. Under suitable formulations, kin-selection and multilevel-selection approaches can provide different mathematical partitions of the same underlying process. Their usefulness depends partly on which causal relationships and population structure the analysis emphasizes.\n\nConflicts also occur within organisms. A genetic element can gain transmission through meiosis at the expense of the organism’s reproductive output, and a cell lineage can proliferate at the expense of the multicellular body. These cases make the unit of fitness indispensable. Increased success for a gene, cell, organism, or group need not mean increased success at every other biological level."
    },
    {
      "heading": "Measuring fitness in experiments and natural populations",
      "text": "Fitness measurement begins by specifying the biological unit, comparison, environment, and interval. In microbial experiments, a common method places a test strain and a reference strain in direct competition and records their relative abundances before and after growth. If both experience the same observation interval, a difference in their average per-capita growth rates can be estimated as\n\n\\[m_i-m_j= \\frac{1}{t} \\ln\\left[ \\frac{N_i(t)/N_j(t)} {N_i(0)/N_j(0)} \\right].\\]\n\nThe interpretation assumes that abundance measurements identify the competitors reliably and that any frequency dependence is considered.\n\nCompetition assays capture effects that separate growth measurements may miss, including interference, cross-feeding, and differential use of shared resources. Nevertheless, the result is specific to the assay. A strain advantageous in a well-mixed flask may behave differently on a surface, in a structured host environment, or under nutrient limitation. Genetic markers used to distinguish strains also require controls because the marker itself can affect performance.\n\nThe Escherichia coli long-term evolution experiment, initiated by Richard Lenski in 1988, illustrates the value of standardized comparisons. Replicate populations have been propagated under a defined serial-transfer regime, with frozen samples permitting comparisons between evolved and ancestral bacteria. Such experiments can directly measure changes in competitive performance and examine historical contingency. They do not establish a context-free ranking: adaptation to the experimental environment may involve trade-offs elsewhere.\n\nIn natural populations, researchers often use survival, mating success, offspring number, genetic parentage, or recruitment of offspring into the breeding population. Long-term studies are especially valuable for species with overlapping generations. Counting offspring at birth can misrepresent evolutionary contribution if juvenile survival differs systematically, while observing only successful breeders excludes individuals whose reproductive success was zero. Unobserved dispersal can also make surviving animals appear dead or reproductively unsuccessful.\n\nStudies of phenotypic selection frequently relate relative reproductive success to measured traits. A selection differential describes the association between a trait and fitness, whereas a multivariate selection gradient estimates an association after accounting for specified correlated traits. Such analyses do not automatically establish causation. Nutritional condition, for example, may influence both a conspicuous trait and reproductive output.\n\nReliable inference therefore benefits from complementary evidence: experiments, pedigree information, repeated observations, demographic models, and explicit treatment of measurement uncertainty. Fitness estimates usually describe distributions rather than fixed outcomes for every individual. Reporting an average advantage without its conditions, sampling variation, or life-cycle coverage can obscure the biological process the measurement is intended to represent."
    },
    {
      "heading": "Adaptation, applications, and interpretive limits",
      "text": "Fitness provides a common language for studying adaptation, but it does not imply that every biological feature is adaptive. Traits can persist because they are selectively neutral, linked to selected variants, constrained by development, or inherited from ancestors under different conditions. Even when selection is demonstrated, the observed trait may be a correlated response rather than the direct target. An explanation requires evidence connecting heritable variation to reproductive consequences.\n\nThe evolution of antimicrobial resistance illustrates the distinction between a useful phenotype and total fitness. A resistance mutation can increase reproductive success in the presence of a drug while reducing growth or competitive ability without it. Such costs are variable rather than universal. Additional mutations can compensate for harmful side effects, and genetic linkage can influence persistence. Consequently, resistance level alone is not a complete measure of fitness, nor does removal of a selective agent guarantee immediate evolutionary reversal.\n\nIn conservation biology, fitness-related measurements help connect genetic variation with population persistence. Inbreeding can increase homozygosity for harmful recessive variants and reduce survival or reproduction. Yet population-level outcomes also depend on habitat, environmental change, demographic structure, and chance. Genetic diversity can contribute to adaptive potential without functioning as a single numerical substitute for present reproductive performance.\n\nEvolutionary thinking also applies to cancer, where somatic cell lineages differ in survival and proliferation within a tissue environment. A cellular change that promotes clonal expansion can harm the organism. This is not a contradiction: cellular and organismal fitness refer to different units and reproductive processes. Similar distinctions are essential when considering viruses, mobile genetic elements, or symbiotic associations.\n\nThe term is used by analogy in algorithms inspired by evolution. A computational fitness function assigns scores that guide reproduction or retention of candidate solutions. Unlike biological fitness, such a score is often explicitly designed by a programmer and may remain fixed throughout a run. Biological environments usually contain no corresponding external evaluator, and reproductive success can alter the conditions that determine future success.\n\nFinally, fitness is a descriptive concept, not a measure of moral worth, social value, or biological progress. It does not rank species on a universal ladder, and high reproductive success in one environment does not establish superiority in another. Its explanatory strength comes from a narrower task: relating particular inherited differences to expected reproductive contributions under identifiable conditions."
    }
  ],
  "links": [
    "natural-selection",
    "genetics",
    "evolution",
    "charles-darwin",
    "population-genetics",
    "genotype",
    "allele",
    "genetic-dominance",
    "mutation",
    "gene-flow",
    "genetic-drift",
    "matrix",
    "genotype-environment-interaction",
    "phenotype",
    "chromosomal-crossover",
    "meiosis",
    "causation",
    "measurement-uncertainty",
    "adaptation",
    "genetic-diversity",
    "cancer",
    "algorithm"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}