Codex @codex  0

Repeatable differences in behaviour among individuals, assessed after accounting for relevant contexts and states. Repeatability does not by itself establish heritability or an adaptive function.

A one-trait short-term response model $R=h^{2}S$, where $R$ is the inherited change in mean, $S$ the phenotypic selection differential and $h^2$ narrow-sense heritability. It assumes an appropriate additive inheritance model and does not turn environmental phenotype change into inherited response.

Heritability is the fraction of phenotypic variance in a specified biological population and environment attributed to genetic variation. It is a biological population parameter, not the genetic fraction of one individual's trait.

A model in which a binary diagnostic outcome occurs when an underlying continuous liability exceeds a threshold. Genetic and environmental components can influence liability. Estimating liability-scale heritability requires assumptions about prevalence, ascertainment and the liability distribution; crude diagnostic concordance is not a continuous-trait correlation.

If observed variation is true-score variation $V_T$ plus independent measurement error $V_{ϵ}$, and genetic variance $V_G$ is part of $V_T$, then $h_{\mathrm{obs}}^2=V_G/(V_T+V_{ϵ})$. Writing measurement reliability as $q=V_T/(V_T+V_{ϵ})$ gives the displayed identity. Removing independent error can raise the heritability estimate without changing the genetic contribution to the true score. The result requires this measurement model; it does not identify measurement reliability by dividing arbitrary estimates from different longitudinal models.

Take selection for darker coats as the premise of the hypothetical example, rather than an established prediction about polar bears. The course of an adaptive walk depends on the available variation, its inheritance, fitness effects and ecological context. Warming alone does not specify a pigmentation mechanism, the strength of selection or whether dark coloration would improve total reproductive success.

The initial response can use standing genetic variation. If darker phenotypes already segregate and have an additive heritable component, a response can begin without waiting for a new mutation. The breeder's equation $R=h^{2}S$ describes a short-term one-trait expectation under its assumptions. Here $S$ is the difference between the mean phenotype of successful parents and the population mean, while $h^2$ is narrow-sense heritability in the relevant environment. Environmental darkening without inherited differences does not supply the same response.

If useful variants are absent, arrival can limit adaptation. In a diploid population with census size $N$ and per-haploid-genome beneficial mutation rate $U_b$, the approximate number of new beneficial copies is $2N{U}_b$ per generation. Most rare variants are lost initially. In a suitable large-population, weak-selection model, an allele whose heterozygote has advantage $s$ has establishment probability of approximately $2s$; this is not a general formula for every dominance relation or bottleneck. The mutation supply depends on actual reproductive numbers, while effective population size controls much of the drift and inheritance history. Long generation times can make a short walk in generations slow in calendar years.

The genetic architecture matters. A change in pigment production, hair structure or the spatial regulation of a developmental pathway can have different pleiotropic consequences. A large-effect allele may be hidden when recessive, constrained by costs in another trait, or beneficial only after a change at another locus. Under sign epistasis, the sign of a step depends on the starting genotype; under diminishing returns, later favourable steps tend to have smaller benefits on an already adapted background. One-step-at-a-time fixation is only an approximation for a sexual, potentially polygenic population.

Recombination can assemble favourable variants inherited from different parents, but can also break favourable combinations. Gene flow can introduce useful alleles or oppose local adaptation if migrants come from a differently selected population. Population bottlenecks can remove rare variants and strengthen drift. Spatial differences in substrate, prey, mate choice or thermal conditions can maintain more than one coat phenotype instead of producing uniform dark fixation.

Finally, the fitness landscape can move while the population responds. Selection on survival, reproduction, energetic balance and correlated traits must be considered jointly. If demographic decline is faster than adaptive change, the existence of an accessible darker phenotype does not guarantee evolutionary rescue. No particular number of substitutions, waiting time or final shade follows from the premise alone. One would need measurements of inherited variation, population structure, generation time and the fitness of alternative phenotypes in the changing environment.

The strongest study would establish independent heritable size increases, map their causes within lineages, and then ask at what biological level those causes recur. Similar adult sizes alone do not establish independent adaptive substitutions, and a repeated change in the same gene need not be the same causal mutation.

Begin with dense sampling of island and mainland relatives across the clade, including small-bodied island lineages where available. Reconstruct a phylogenetic tree from many genomic regions, estimate colonization relationships, and compare body-size histories with uncertainty in ancestral states. A single large-bodied ancestor followed by dispersal to several islands is a different history from multiple independent increases. Discordant local trees require attention to incomplete lineage sorting and introgression. Record island area, resource availability, predators, competitors and colonization history, because these can explain why natural selection favored size, or why some apparent replicates share the same historical cause. The number of independently evolved shifts, rather than the number of animals sampled, determines replication for a claim about convergent evolution.

Define the phenotype carefully. Measure growth trajectories, adult skeletal dimensions, lean mass, fat mass and maturation time at known ages and separately by sex; body weight alone can conflate structural size, condition and pregnancy. Measure individuals blind to ancestry. Rear descendants from several independent families in a common garden experiment, randomizing cages and controlling diet, density, temperature and pathogen exposure. Include more than one rearing environment to test genotype-environment interaction. Reciprocal crosses and cross-fostering can help distinguish inherited genetic variants from maternal provisioning and early rearing effects. Persistence across controlled generations supports a heritable component, but does not by itself eliminate every persistent nongenetic effect. Field pedigrees and heritability estimates can complement, rather than substitute for, these controls.

Generate comparable genome data. High-quality assemblies from representative island and mainland animals help detect structural variants and gene duplication; population resequencing estimates allele frequencies and haplotypes. Mapping every lineage only to a mainland reference can preferentially miss island sequence, so use alternative references or a pangenome where feasible and compare mapping quality across groups. Sequence both sexes and retain sex chromosomes and mitochondrial genomes when technically possible. Design sampling and coverage using a statistical power analysis, with multiple unrelated individuals per lineage and independent cohorts for replication; one genome per island is inadequate for within-population inference.

Reconstruct demography using broadly distributed putatively neutral regions. Island founding can create a population bottleneck, increased inbreeding and long-range linkage disequilibrium. These processes can imitate aspects of a selective sweep. Compare candidate regions with null simulations that reproduce plausible colonization sizes, gene flow, recombination rates and sequencing ascertainment; propagate uncertainty in those quantities. High FST can reflect reduced within-population diversity rather than an exceptional accumulation of adaptive differences, so examine absolute divergence and diversity alongside it. A high ratio of nonsynonymous to synonymous changes can also reflect relaxed purifying selection, especially after a decline in effective population size. No single outlier statistic identifies a body-size mechanism.

Where fertile crosses are possible, perform quantitative trait locus mapping using island–mainland intercrosses, with enough recombinants to resolve several causal regions. Measure longitudinal growth and test sex-dependent effects, dominance and epistasis, while controlling family and shared rearing. Natural association mapping can help when large samples segregate relevant alleles; account for population stratification and relatedness with an appropriate Gaussian linear mixed model. Fixed island–mainland differences cannot be mapped within an island where they no longer vary, and a genome-wide association study that simply contrasts species confounds ancestry with body size. As an empirical precedent, Gough Island mouse intercrosses identified multiple regions affecting body weight and growth rate, illustrating that rapid extreme size evolution can have a distributed genetic basis.^[ref] These regions are starting points for resolution, not automatically individual causal genes.

Integrate mapping with measurements in relevant tissues and developmental windows. RNA sequencing of growth plate, muscle or endocrine tissue may identify candidate changes in gene expression, but match developmental stage, sex, diet and cellular composition so that a larger animal's different tissue mixture is not mistaken for a regulatory cause. allele-specific expression in an informative first-generation cross can distinguish some effects of cis-regulatory evolution from shared trans-acting influences, provided reference-mapping bias is controlled. Test accessible enhancers or coding changes suggested jointly by mapping, sequence and tissue evidence. Growth pathways are reasonable hypotheses, but discovering a familiar growth-related annotation is not causal validation and should not exclude unexpected genetic loci.

Resolve the level of convergence explicitly. Independent lineages might share the same new nucleotide substitution, different changes in the same gene, different genes in one pathway, or many changes distributed across a polygenic trait. Shared long haplotypes and local ancestry can reveal reuse through introgression or standing genetic variation. Distinguish these from recurrent mutations using local genealogies, surrounding sequence and age estimates. For pathway enrichment, control gene length, annotation density, mutation rate, genetic linkage and multiple testing; a shared gene annotation is not an independent replicate for every linked site. Failure to find repeated individual hits does not exclude convergence at a broader biological level.

Finally, replace candidate alleles or enhancers in controlled genetic backgrounds, ideally in both directions, and measure their effects on growth and its costs. Use unedited siblings, sham-edited lines, independent edited founders and a rescue or precise reversion to address manipulation and off-target effects. Test combinations if mapping indicates epistasis. Cell assays or another model species can establish biochemical activity, but need not reproduce the original mammal's whole-body effect. To call the change adaptive, also connect size and the candidate alleles to survival and reproductive success under relevant island conditions; mainland and island environments provide a useful reciprocal comparison. A persuasive result connects independent history, heritable size, specific causal variants and genotype fitness consequences, while allowing convergence to involve different genes or reuse of ancestral variation.

Converging family, twin, adoption and molecular evidence establishes genetic contributions to vulnerability to several mental disorders; it does not make onset inevitable or identify a single genetic cause for every diagnosis. The outcome must be specified. Genetic influences on lifetime diagnostic liability are not automatically the same as influences on the exact age of first onset, episode recurrence, symptom severity or treatment response.

Family aggregation provides a starting point. Relatives of affected individuals can have higher rates of the same disorder, with closer biological relatives generally showing greater risk. However, families share environments, social circumstances and diagnostic expectations as well as genes. A familial association alone therefore cannot distinguish genetic transmission from environmental transmission or their combination. Designs exploiting differences in genetic relatedness and rearing relationships address this ambiguity more directly.

In a twin study, monozygotic twins share essentially all inherited sequence variation, whereas dizygotic twins share about half of segregating additive variation on average. Greater resemblance among monozygotic pairs supports a genetic contribution if relevant environmental similarity and ascertainment assumptions are adequately addressed. Under a simple standardized ACE model, the twin correlations satisfy $r_{\mathrm{MZ}}=A+C$ and $r_{\mathrm{DZ}}=A/2+C$, giving $A=2(r_{\mathrm{MZ}}-r_{\mathrm{DZ}})$. For diagnoses, investigators commonly model an underlying liability-threshold model, rather than treating crude concordance percentages as continuous-trait correlations. Patrick F. Sullivan and colleagues' meta-analysis of 12 systematically ascertained schizophrenia twin studies estimated substantial heritability of liability, around $81\%$. That is a model-based population estimate with uncertainty, not an individual's probability of developing schizophrenia.^[ref]

Twin comparisons have limitations. Monozygotic twins may be treated more similarly, prenatal environments can differ systematically, and the equal-environments assumption concerns environments causally relevant to the disorder rather than identical treatment in every respect. Assortative mating alters expected resemblance, while nonadditive effects and measurement error complicate variance decomposition. Monozygotic concordance below one is incompatible with a simple fully penetrant deterministic account, but the difference cannot be assigned wholly to one specific environmental exposure: stochastic development, diagnostic error and other processes also contribute.

Adoption and extended-family data provide a partly independent test. Paul Lichtenstein and colleagues linked Swedish population and psychiatric inpatient registers, studying over two million nuclear families. They compared biological and adoptive relationships, full siblings and half-siblings for schizophrenia and bipolar disorder. Risk patterns followed biological relatedness, and adopted children of biological parents with bipolar disorder also showed elevated schizophrenia risk. The analysis supported partially shared genetic influences across the two diagnoses. This weakens an explanation based solely on living with an affected biological parent.^[ref]

It does not eliminate every environmental explanation. Selective placement, prenatal exposure, contact with biological relatives and nonrandom adoption can matter. Inpatient-register diagnoses preferentially identify people reaching those services, and changing diagnostic practice can affect estimates. Nevertheless, convergence across twins, extended families and adoption is more persuasive than repeating one family association, because the sources of confounding differ.

Molecular evidence measures inherited variation directly. In a genome-wide association study, genotype frequencies are compared between cases and controls across many markers, using stringent multiple-testing correction and replication. The Psychiatric Genomics Consortium's 2014 schizophrenia study analyzed up to 36,989 cases and 113,075 controls and identified 108 associated loci. The pattern supported a polygenic trait architecture involving many common variants, individually having small effects, rather than one common “schizophrenia gene.” Genotypes precede clinical onset, so the inherited sequence difference is not created by experiencing the disorder.^[ref]

However, an associated marker may tag a causal variant through linkage disequilibrium; a locus is neither necessarily one gene nor a fully identified biological mechanism. Population stratification, sample overlap and diagnostic heterogeneity can bias findings. A weighted genetic risk score can predict differences in risk in independent samples without providing accurate individual diagnosis or transferring equally across ancestry groups. Discovery of association is consequently stronger evidence for inherited liability than for the timing of a particular person's first episode.

The history of candidate gene research illustrates why replication matters. Richard Border and colleagues tested widely studied depression candidate-gene and gene-by-environment hypotheses across large samples with preregistered analyses. They found no convincing support for the historically proposed large effects. This does not erase genetic influences on depression or rule out every genotype-environment interaction; it challenges specific small-study claims and demonstrates why biological plausibility alone cannot validate an association.^[ref]

Finally, heritability is conditional on a population, environment and measurement system. A highly heritable vulnerability can still be preventable, and a low-heritability outcome can still involve essential biological mechanisms. Gene-environment correlation means inherited characteristics can influence the environments people encounter; genotype-environment interaction means their consequences can depend on those environments. Neither should be collapsed into an exclusive genes-versus-environment choice. The evidence strongly supports inherited differences in liability to several disorders, especially where family designs and replicated molecular findings converge; explaining actual onset additionally requires developmental and environmental evidence.

The common marginal trait variance isFor monozygotic pairs, shared genes and shared environment giveFor dizygotic pairs, the genetic covariance is halved while the common environment is unchanged:Subtracting cancels the common-environment variance and proves Falconer's formulathe heritability under the ACE model.

Individual characteristics contribute to adaptation, but their effects depend on adversity, relationships, opportunities and the outcome being measured. Psychological resilience is therefore not well explained by dividing people into inherently strong and weak types. A useful account asks which characteristic predicts which aspect of functioning under which conditions, and whether changing that characteristic would improve the outcome.

Cognitive capacities can help a person understand a problem, plan a response and use available support. Temperament and emotion regulation may influence how strongly threat is experienced and how readily attention can be redirected. Coping confidence may encourage engagement, while flexible strategies can help match responses to a controllable or uncontrollable situation. These are plausible mechanisms, not proof that a single score measures psychological resilience. Some apparently personal resources, such as social competence, also operate by eliciting or maintaining helpful relationships.

Longitudinal evidence supports contributions from both person and context. Emmy E. Werner's Kauai study followed a birth cohort from early life into adulthood, identifying protective patterns among people exposed to substantial developmental risks. Individual capacities, supportive caregivers and opportunities outside the immediate family all featured in the account. Because these resources were not randomly allocated, their predictive value does not isolate a pure individual effect from the environments that helped develop or express it.^[ref]

More specific evidence comes from Ann S. Masten and colleagues' ten-year follow-up of 205 children recruited during elementary school. Using multiple methods and informants, they examined academic, conduct and peer competence in relation to adversity. Intellectual functioning and parenting resources predicted favorable outcomes across domains; findings also suggested protective roles for IQ and parenting in relation to antisocial behavior. Resilient adolescents resembled competent low-adversity peers in several ordinary resources rather than exhibiting an entirely separate human type.^[ref]

The causal interpretation remains qualified. Better intellectual performance can reflect previous schooling, health and caregiving as well as inherited variation. Parenting can respond to child behavior, and both may share unmeasured antecedents. A variable measured before a later outcome is a stronger predictor than a simultaneous self-description, but prediction alone does not establish that raising its score will reproduce the observed benefit. An intervention needs to test the proposed process rather than assume the correlational ordering is causal.

Promotive and buffering effects should also be separated. For an adverse outcome $D$, adversity $S$ and resource $P$, consider an illustrative model,When ${\beta }_3=0$ and ${\beta }_2<0$, the resource has a constant favorable association across the modeled adversity range. A negative ${\beta }_3$ instead means that increasing the resource reduces the adversity slope, ${\beta }_1+{\beta }_{3}P$. This is a psychological stress-buffering pattern on this scale. Neither coefficient alone proves causation; interaction can change with the outcome scale, and a statistically nonsignificant interaction may reflect poor measurement or insufficient power rather than equal effects everywhere.^[ref]

Context can enable or suppress a capacity. Planning is less useful when there are no safe options, and willingness to seek support cannot ensure that responsive support exists. A coping response that helps during a brief challenge can be costly during prolonged insecurity. High observable achievement can coexist with internal distress, so defining success entirely by school performance can exaggerate the apparent protection conferred by conscientious behavior. Outcome selection changes which characteristics seem important.

Experimental evidence argues against treating context as a passive background. The supported foster-care offer in the Bucharest Early Intervention Project improved some later mental-health outcomes, with sex-specific uncertainty. Such a change cannot be explained solely by an immutable difference between supposedly resilient and nonresilient children. It demonstrates that modifying a developmental environment can change observed adaptation, while not establishing that every child benefits equally or through one unique mechanism.^[ref]

The relative importance of characteristics can change over time. A relationship or opportunity encountered later can alter a developmental course; stable adaptation after one event does not guarantee the same response to another. Prospective bereavement findings also show why baseline functioning matters: a favorable post-event score can represent stable functioning or improvement from earlier difficulty. A personality correlate measured only after the event may partly be a consequence of the adjustment being explained.^[ref]

Genetic findings do not restore a simple fixed-trait account. A heritable regression residual-based adaptation score still depends on exposure measurement, the surrounding environment and inherited influences on the component symptoms. Heritability partitions variation in a studied population; it does not say that an individual's coping capacity is fixed or that a social intervention cannot help. Equally, environmental influence does not make every individual difference transient or unimportant.

Individual characteristics are consequential resources within a developmental system, not sufficient explanations of who will adapt successfully. Good research compares personal and contextual predictors prospectively, tests interactions and mechanisms, and evaluates functioning across domains. Prevention can strengthen relevant skills while also protecting relationships and material conditions. This is a more useful consequence of individual-difference evidence than attributing continuing difficulty to an alleged lack of personal psychological resilience.

Inherited variation contributes to psychiatric vulnerability and to some measured differences in adaptation after psychological stress, but its importance is phenotype-, population- and environment-dependent. Genetic influence is not equivalent to inevitability, a single large-effect gene or a wholly separate set of protective genes. A satisfactory answer must distinguish genetic effects on exposure, on symptoms and on responses conditional on exposure.

Family resemblance provides an initial clue, but families transmit environments and social circumstances as well as DNA. Twin comparisons exploit differences in relatedness: monozygotic pairs share essentially all inherited sequence variation, whereas dizygotic pairs share about half their segregating additive variation on average. Under an ACE model, greater resemblance of monozygotic pairs can support a genetic component if the relevant environmental and ascertainment assumptions hold. Equal environments means comparable effects of relevant environmental similarity, not literally identical treatment of every twin.

Disorder-level evidence is substantial. Patrick F. Sullivan and colleagues' ascertainment-aware synthesis of schizophrenia twin studies estimated liability heritability at roughly $81\%$, with uncertainty. Paul Lichtenstein and colleagues' Swedish family-register study supported partially shared inherited liability to schizophrenia and bipolar disorder using biological, adoptive and sibling relationships. The convergence supports inherited vulnerability while showing that diagnostic categories do not necessarily partition genetic causes into entirely independent sets.^[ref][ref]

Molecular evidence provides a different test. The Psychiatric Genomics Consortium's 2014 schizophrenia study identified associations spanning 108 loci in a large multi-stage genome-wide analysis. Many common variants contribute small effects; an associated locus is not automatically one causal gene or a known mechanism. Population stratification, linkage disequilibrium, case ascertainment and sample overlap still require control. A genetic risk score can predict risk differences in independent data while remaining far from a deterministic diagnosis or a transportable measure of one person's psychological resilience.^[ref]

Psychological resilience needs additional definition because favorable outcomes without adversity do not establish psychological stress resistance. Ananda B. Amstadter and colleagues studied 7,500 adult twins at two assessments, modeling internalizing symptoms relative to cumulative stressful events. Their regression residual-based phenotype had estimated heritability around $31\%$ at each wave; longitudinal modeling that accounted for measurement error and occasion-specific influences gave a larger latent estimate, around $50\%$. This supports both inherited and environmental contributions to that operationalization, not a universal percentage for every meaning of psychological resilience.^[ref]

The shift in estimates is conceptually informative. Independent measurement error adds to observed variance without adding the corresponding genetic signal. In a simple measurement model, if $V_G$ is genetic variance, $V_T$ true-score variance and $V_{ϵ}$ independent error variance, thenRemoving noise can therefore raise estimated heritability without changing anyone's genes. This illustration is not a reconstruction of the study's full model: occasion-specific effects and longitudinal assumptions also matter, so $31/50$ should not be reported as an independently established measurement reliability coefficient.

A regression residual phenotype also inherits assumptions from its construction. If adversity is undermeasured or a prediction model is misspecified, the regression residual includes more than protective capacity. Genetic effects on depression or anxiety can appear as genetic effects on low symptom burden after psychological stress, even without a separate buffering mechanism. Demonstrating a psychological resilience-specific effect would require stronger evidence that inherited variation changes the response to comparable exposure, with adequate exposure measurement and explicit competing models.

Gene–environment processes further complicate simple partitioning. Gene-environment correlation occurs when inherited characteristics influence environments encountered or elicited, or when parental characteristics affect both inherited liability and the rearing environment. Genotype-environment interaction instead means that inherited differences have different consequences across environments. Neither can be inferred merely because both genes and psychological stress predict the outcome. Restricting analysis to people with a particular exposure can also create selection bias if exposure is influenced by several causes relevant to symptoms.

Historical candidate-gene findings illustrate the need for stringent evidence. Richard Border and colleagues' large, preregistered analyses did not support widely proposed depression candidate-gene or candidate-gene-by-environment effects. This challenges those specific claims and the power of small discovery studies; it does not demonstrate that all interactions are absent. Replication, correction for multiple testing, ancestry-aware analyses and precise phenotype definitions are essential before a plausible biological story becomes a reliable finding.^[ref]

Finally, high heritability does not mean an outcome cannot be changed. It describes sources of variation under a particular range of environments, not the response to an environment absent from the data. Genetic effects can act through modifiable developmental systems, and external support can benefit people across inherited-risk levels. Predictive accuracy and equity can also deteriorate when scores are transferred to populations unlike the discovery sample.

Genes contribute meaningfully to liability and to particular measured adaptation phenotypes, while evidence for specific protective mechanisms is narrower than evidence for heritability. The strongest account combines family and molecular results with explicit exposure and developmental models, rather than assigning a fixed genetic percentage to an individual or replacing environmental prevention with genetic classification.

Knowing the molecular basis of a behaviour is not a prerequisite for understanding its evolution or many of its evolutionary consequences. Evidence about transmission is essential for claims of inherited change, while detailed genetic architecture becomes essential for particular mechanistic predictions. The claim therefore depends on what is meant by a genetic basis and on the explanatory question being asked.

Tinbergen's four questions distinguish mechanism, development, adaptive significance and evolutionary history. Identifying a neural or molecular mechanism helps explain how a response is produced; it does not automatically show why a variant spread, whether it currently increases reproductive success, or whether the same mechanism generated the ancestral transition. Conversely, a well-supported evolutionary explanation can begin with experimentally demonstrated biological fitness consequences and transmission without a complete molecular account. These explanatory levels are complementary rather than substitutes. ^[ref]

The phrase genetic basis can refer to several different achievements: identifying inherited variation; estimating the number and effects of contributing genetic loci; finding a causal nucleotide change; or describing every gene required for development of the relevant sensory and motor systems. These are not equivalent. Many genes can be necessary for an animal to perform an action while none explains variation in that action within the studied population. Knocking out a broadly required neural gene and eliminating courtship does not identify an evolutionary substitution for a naturally varying mate preference.

quantitative genetics can predict a response without naming the causal molecular variants. Under the appropriate additive, generational and environmental assumptions, the multivariate breeder's equation gives:where $\mathbf{G}$ is the additive genetic covariance matrix and $\mathbf{\beta }$ the vector of phenotypic selection gradients. Behavioural measurements, pedigrees, repeated observations and reproductive success can therefore estimate whether and in what direction a response is expected. The covariances matter: selection favouring exploration can produce a correlated response in aggression or constrain the response when their biological fitness effects conflict. Naming an exploration-related gene would not replace this population-level information.

These estimates require strong controls. behavioural repeatability is consistency among observations of the same individual; it can reflect persistent environment as well as inherited differences and is not equivalent to heritability. Parent–offspring resemblance can also reflect teaching, maternal provisioning or shared habitat. cross-fostering, common garden experiments and appropriately structured breeding designs help separate these alternatives. heritability is conditional on the sampled population and environment, not a measure of immutability. A response measured in one environment may fail elsewhere because of genotype-environment interaction or altered selection. Thus ignorance of exact molecular variants is compatible with informative prediction, but ignorance of transmission and its confounders is not.

Molecular work adds power when it establishes a causal link to naturally occurring variation. In Drosophila melanogaster, rover and sitter food-search variants were connected to the foraging gene, which encodes a cGMP-dependent protein kinase. Mapping, enzyme activity and transgenic rescue linked the molecular difference to behavioural performance; comparable locomotion without food helped exclude a general activity explanation. ^[ref] This is much stronger than finding a statistical association between a gene and a behavioural score. Nevertheless, the experiment alone does not establish which ecological conditions historically maintained the polymorphism. That requires biological fitness comparisons across resources, densities and frequencies, with genetic-background controls.

genetic architecture is particularly informative when animal behaviour changes gene flow. In Heliconius melpomene and Heliconius cydno, mapping identified several major regions associated with male mating preferences; one lay close to the wing-pattern regulator Optix. Tight genetic linkage can help maintain associations between a mating cue and preference despite exchange between lineages. It does not establish pleiotropy, and courtship assays must distinguish attraction from the female's response and subsequent mating. ^[ref] Here molecular localization helps evaluate whether a proposed route to reproductive isolation is genetically feasible. A preference could still have been shown to reduce heterospecific mating before its contributing regions were mapped.

The influence of one individual's animal behaviour on another's development further complicates a simple own-genotype explanation. Maternal provisioning in Onthophagus taurus was studied using a breeding design that separated direct and maternal genetic contributions to offspring traits. ^[ref] An indirect genetic effect occurs when another individual's inherited properties alter the focal individual's phenotype. Such effects change the environment in which selection acts and can alter expected evolutionary responses. Identifying only the offspring's molecular variants would miss an important part of the causal system. Conversely, an observed maternal effect is not automatically a maternal genetic effect; provisioning can vary for environmental reasons.

Transmission need not follow genetic parentage. In wild great tits, experimenters trained demonstrators to use alternative puzzle-box techniques and followed the spread of the introduced variants. Persistent local traditions and preferential adoption of the common local technique supported social learning and conformist transmission. ^[ref] The experiment established a route of behavioural inheritance without identifying a molecular difference between the techniques' users. It did not prove that learning ability has no inherited basis, or that the new tradition changed allele frequencies. Rather, it shows why an account of behavioural change needs to identify the relevant transmission process instead of assuming it.

Even a learned response with no demonstrated inherited differences can influence subsequent evolution. A new resource use can alter selection on digestion; a socially learned mating signal can alter assortative mating; habitat choice can change contact with other populations. The immediate causal effect can be tested by manipulating the action or its social transmission and measuring biological fitness or exchange. To claim ensuing genetic adaptation, one must additionally show inherited change. To explain exactly which variants can respond, how quickly, and with what correlated effects, information about genetic architecture becomes increasingly valuable.

The most productive programme therefore combines behavioural intervention, transmission studies, biological fitness measurement and molecular analysis when it addresses an unresolved mechanism. Candidate variants should be tested across natural genetic backgrounds and environments, preferably with targeted perturbation and rescue; historical claims also need comparative or temporal evidence. Exact molecular knowledge is neither sufficient nor universally necessary: the necessary evidence is the evidence that connects the particular behavioural difference, its transmission and its measured evolutionary consequence.

Primary two-wave twin study with an exposure-conditioned symptom-regression residual phenotype. Wave-specific and latent longitudinal heritability estimates concern different modeled variance components.

A somatic mutation occurs in a body-cell lineage outside the germline mutation route. Its descendant biological cells can inherit it through DNA replication, even though it is not transmitted through gametes. Thus heritability within a cell lineage differs from inheritance between sexually reproducing organisms.