How Science Works | How the Same Genes Can Produce Different Traits in Different Environments

A DNA sequence does not specify a final height independently of everything around it. It contributes to molecular machinery that detects signals, regulates activity, builds tissues and allocates resources. Those processes occur in a physical and biological setting. Changing the setting can change the outcome without changing the inherited sequence.

Imagine a response chain: an environmental condition changes a signal; the signal alters regulatory activity; development or physiology changes; a measurable trait follows. This diagram identifies possible links to investigate. It does not claim that every environmental response uses the same sensor or regulatory pathway.

The title’s “same genes” is shorthand for holding genetic background sufficiently constant for the comparison. Sharing a species name is not enough. Two unrelated plants may possess different versions of many genes, and their different heights could reflect those differences as well as their environments.

Nor does identical DNA guarantee identical outcomes in identical surroundings. Developmental variation, measurement error, previous experience and unmeasured conditions remain. A reaction norm usually describes an expected value or distribution, not a rule that every individual must match exactly.

Environmentally altered gene expression also differs from a change in DNA sequence. Some responses involve persistent regulatory states, but persistence by itself does not demonstrate inheritance through multiple generations. Each additional claim needs its own evidence. The useful first question is simply which trait changes when which condition changes, with what genetic and developmental background held comparable.

Contents · Next chapter

Draw an environmental axis horizontally and a trait axis vertically. For genotype A, place its average trait value at each measured condition, then show the pattern connecting those observations. Draw genotype B separately. The two curves are two reaction norms for the same measured trait.

A horizontal line indicates no detected change in that trait across the tested conditions. It does not mean the organism lacks plasticity in every other trait, at every age or beyond that environmental range. A rising line indicates an increase in the measured trait, not automatically an improvement in performance.

Parallel separated lines describe a simple case: genotypes differ in average value but show the same absolute response over the range. Different slopes indicate that the response depends on genotype. This is a genotype-by-environment interaction, often abbreviated G × E. Lines need not cross for such an interaction to exist.

Crossing lines make the consequence particularly visible: the genotype with the larger trait in one environment has the smaller trait in another. A ranking made under one condition cannot simply be carried into the other.

Always inspect units and scales. An equal increase of four centimetres differs from an equal proportional increase of forty per cent. Transforming a trait can change the apparent interaction. Scientists therefore choose a biologically meaningful scale and state it before interpreting which genotype is “more plastic.” There is no single ranking of plasticity that is independent of the trait, environmental interval and measure used.

Contents · Previous chapter · Next chapter

Consider a fictional plant trait measured in millimetres. Let an environmental index E run from 0 to 1. It represents two endpoints of a controlled comparison, without claiming a particular real nutrient concentration or temperature. Suppose the expected responses within this interval are A(E) = 10 + 8E and B(E) = 14 + 2E.

At E = 0, A measures 10 mm and B measures 14 mm. At E = 1, A measures 18 mm and B measures 16 mm. A changes by 8 mm; B changes by 2 mm. On this absolute scale and over this interval, A has the larger response.

The genotype ranking reverses. To find the model’s crossing, set 10 + 8E equal to 14 + 2E. Subtracting gives 6E = 4, so E = 2/3. At that point both expected trait values are 46/3 mm, approximately 15.33 mm.

Diagram to sketch: draw A from the lower-left point (0, 10) to the upper-right point (1, 18), and B from (0, 14) to (1, 16). Label the crossing near two-thirds of the horizontal distance. Place vertical error bars beside hypothetical observations only if uncertainty values are actually supplied; do not invent them.

The straight lines are assumptions of this invented model. If an experiment measured only the endpoints, it would establish different endpoint responses but would not prove linearity or locate an actual crossing precisely. Intermediate measurements might reveal a threshold, plateau or bend. Extending either equation beyond E = 0 to 1 would add a prediction unsupported by the specified model domain.

Contents · Previous chapter · Next chapter

Now suppose the population contains equal numbers of A and B, and their proportions stay fixed. At E = 0, its mean trait is (10 + 14)/2 = 12 mm. At E = 1, the mean is (18 + 16)/2 = 17 mm. The population average changes by 5 mm entirely through the modelled environmental responses.

No genotype-frequency change is needed. Calling this five-millimetre shift genetic evolution would confuse a change in expressed traits with a change in inherited population composition.

Consider a different comparison at constant E = 0. If A’s fraction changes from one-half to three-quarters, the new mean becomes 0.75 × 10 + 0.25 × 14 = 11 mm. Here the one-millimetre decrease comes from changed genotype proportions, despite no change in either reaction norm.

If both things happen, the final mean at E = 1 with three-quarters A is 0.75 × 18 + 0.25 × 16 = 17.5 mm. Relative to the original 12 mm, the total increase is 5.5 mm. One bookkeeping route assigns 5 mm to changing environment at the original mixture, then 0.5 mm to changing the mixture in the final environment.

Reversing that bookkeeping order assigns −1 mm to changing the mixture first and +6.5 mm to the subsequent environmental change. Both sum to 5.5 mm. Because the genotypes respond differently, the separate numerical contributions depend on the chosen reference comparison. The total is unambiguous; a unique causal allocation requires an explicitly defined counterfactual, not just two population averages.

Contents · Previous chapter · Next chapter

The reaction-norm graph records a trait. Fitness concerns contribution to future generations in a specified setting. A taller plant is not automatically a more successful plant: height might aid access to light while increasing construction costs or exposure. The relevant outcome must be measured rather than inferred from the upward direction of a line.

In an original hypothetical extension, suppose A and B leave 20 and 30 surviving offspring respectively at E = 0, but 40 and 25 at E = 1. Their performance rankings reverse along with the stated environmental comparison. These invented values establish genotype-level differences in reproductive outcome within the model.

They do not prove that A’s larger height response caused its reproductive advantage. A and B may differ in other traits. A mechanistic claim about adaptive height plasticity needs evidence separating that response from correlated differences, ideally including informative manipulations or comparisons that address the relevant alternatives.

A real example comes from Pigliucci and Schlichting’s 1996 Arabidopsis study. The researchers compared responses to nutrient availability and their relationships with reproductive fitness. They found selection associated with means of several traits and with plasticity in some traits, including branching. The findings distinguish selection on a trait’s average expression from selection associated with its environmental response. They do not supply a universal rule that steeper reaction norms are better. Original study abstract.

A response may improve performance relative to an alternative yet still fail to prevent a population decline. “Helpful under these conditions” and “sufficient for persistence” are separate conclusions, especially when reproduction, survival and environmental change operate on different timescales.

Contents · Previous chapter · Next chapter

A common garden places different genetic backgrounds under comparable conditions. If a difference persists, immediate surroundings alone cannot explain it. However, one shared environment samples only one point on each reaction norm. It cannot reveal whether the backgrounds respond similarly elsewhere.

A stronger plasticity comparison crosses backgrounds with environments: each background is represented in each condition. Clonal material can help hold genotype constant where biologically appropriate. In sexual species, siblings are related but generally not genetically identical, so family designs and statistical models must reflect the actual inheritance structure.

Imagine that all A plants occupy one chamber and all B plants another. A chamber difference could masquerade as a genotype difference. Likewise, many plants sharing one treatment chamber are not automatically many independent environmental replicates. The treatment’s assignment and the unit of replication matter as much as the number of measurements.

Age, starting size, parental conditions and measurement timing also matter. A trait measured on the same calendar day can compare different developmental stages if conditions alter growth rate. Measuring at a common developmental stage answers a different question from measuring after equal elapsed time; neither choice should be hidden.

The 2025 seed-beetle study by Rêgo and colleagues illustrates the distinction between evolutionary history and assay environment. It compared experimentally evolved backgrounds using a common-garden design that included two assay temperatures and addressed non-genetic parental effects, alongside genomic evidence. Its broader results concerned the repeatability of thermal adaptation, rather than a universal beetle response curve. Original research.

Contents · Previous chapter · Next chapter

Plasticity is sometimes imagined as a switch that an adult can turn back whenever conditions change. Some physiological or behavioural responses are reversible on relevant timescales. Other responses alter development during a sensitive period and persist after the original cue disappears.

For that reason, “same current environment” does not necessarily mean “same relevant environmental history.” Two individuals can arrive in the same final surroundings after different developmental experiences. Their different traits may reflect those earlier experiences without requiring a new inherited DNA variant.

A more informative conceptual diagram adds time: early environment → developmental pathway → later phenotype, with the later environment shown separately. The arrow from early conditions does not vanish just because the organisms now share a location.

Suppose an observed response is slow. A brief exposure could suggest no plasticity even though a longer, age-matched comparison would reveal it. Conversely, an immediate change might be transient and disappear before reproduction. The duration of exposure, delay before measurement and biological timescale belong in the result.

Intergenerational effects add another layer. Parental conditions can influence offspring through provisioning and other mechanisms. Finding an offspring difference is therefore not enough to identify a DNA-sequence change, prove an indefinitely inherited epigenetic state or establish adaptive evolution. Common-garden generations, pedigrees, genomic evidence and mechanistic work address different parts of that problem.

The practical lesson for reading research is to ask what was kept constant across time, not only across space. A reaction norm is conditional on the organism’s developmental stage and history, even when a simple two-axis graph leaves those conditions in its caption.

Contents · Previous chapter · Next chapter

An organism often responds to an indicator of future conditions rather than to the eventual challenge itself. Developmental temperature, for example, may carry information about a seasonal environment. A response can be useful when that relationship is reliable and less useful when the relationship changes.

This creates a distinction between the cue and the environment that determines reproductive consequences. A graph against the cue describes the response. Testing whether that response matches later conditions requires additional observations. A strong response to an unreliable cue can produce a particularly large mismatch.

Oostra and colleagues studied the seasonally plastic butterfly Bicyclus anynana in 2018. Their factorial analysis found extensive seasonal differences in gene expression but relatively little within-population genetic variation in that transcriptional plasticity. The result separates possessing a strong response from possessing much genetic variation in the response itself. Their climate-change interpretation concerned the risk of changing cue reliability; it was not a demonstration that all plastic species must decline. Original research and abstract.

Return to the fictional A and B lines. A’s steeper slope makes its trait more responsive to E. If E is an informative cue, that responsiveness might help. If E ceases to predict the relevant future environment, the same slope could become disadvantageous. The line alone cannot tell us which situation applies.

That is why “more plastic” cannot substitute for an ecological explanation. The reliability of information, speed of response, costs of the resulting phenotype and consequences for reproduction all affect whether responsiveness helps in the circumstances being studied.

Contents · Previous chapter · Next chapter

A population can evolve different reaction norms when heritable differences in environmental responses influence reproductive outcomes. Evolution may alter average trait expression, the response slope, a threshold or the shape of a nonlinear curve. It need not always increase responsiveness.

In the original A/B example, a change in genotype frequencies changes which reaction norms are represented in the population. That is different from merely moving the unchanged population from E = 0 to E = 1. Over generations, both processes can occur together, so investigators need designs that separate them.

Gibbin and colleagues’ 2017 study examined marine polychaete worms under warming and acidification scenarios, using reciprocal transfers across generations to compare life-history responses. Most measured traits retained their plastic responses; juvenile developmental rate in the combined scenario was the exception highlighted in their results. The evidence shows why neither universal loss nor universal increase of plasticity should be assumed under persistent environmental change. Original research.

An evolutionary interpretation must also consider selection among lineages, parental carry-over and the conditions under which traits were measured. Merely finding a changed response after several generations does not uniquely identify its molecular cause. Complementary genetic and developmental evidence can strengthen that inference.

Selection can sometimes favour expression of a previously induced trait with less dependence on the original cue, a pattern called genetic assimilation when appropriately demonstrated. This does not mean an organism consciously copies an acquired improvement into its DNA. It describes an evolutionary change in how inherited developmental systems produce traits. It is one possible outcome, not the inevitable destination of plasticity.

Contents · Previous chapter · Next chapter

A useful paper worksheet begins with four entries: the trait and its units, the environmental variable and tested range, the genetic backgrounds being compared, and the age or developmental stage at measurement. If one is missing, identify how its absence limits interpretation.

Next distinguish three observations. A change within a background across environments supports an environmental response. A difference between backgrounds in one environment supports a background effect under that condition. A difference between their responses supports an interaction, subject to uncertainty and the chosen measurement scale.

For the original model, check your understanding: which genotype has the larger trait at E = 0? B. Which changes more between the endpoints? A. Does the crossing at E = 2/3 follow from two measurements alone? No; its exact location uses the stated straight-line model. Does the increase from a population mean of 12 to 17 mm require genetic evolution? No; the mixture stayed constant.

Then inspect the evidence for consequences. Was reproductive performance measured, or was the trait merely described as useful? Were evolutionary changes distinguished from current environmental responses? Could prior environments or non-independent replicates explain the pattern? Were claims about future conditions tested or extrapolated?

The central idea is precise: inherited systems can produce multiple outcomes, and their responses can differ. Understanding that flexibility requires measuring the response, its limits and its consequences separately. Continue through Evolutionary Biology for selection, inheritance and population change, or return to the How Science Works master guide to follow another mechanism.

Contents · Previous chapter · Continue to Evolutionary Biology