eduKateSG · How Science Works
Choose your reading route
Assign environmental measurements, variant comparisons and frequency changes to separate causal arrows in an evolutionary-feedback claim.
Separate causal arrows and environmental measurements · Chapters 1–3 · Start this route
Calculate frequencies and name the statistic · Chapters 4–6 · Start this route
Audit comparisons, feedback and environmental inheritance · Chapters 7–10 · Start this route
Full contents · Check your understanding · How Science Works master guide
Organisms can alter the conditions around them. A more demanding claim is that those alterations change which heritable variants leave descendants. To test that claim, scientists must connect an environmental measurement to a comparison of relative biological success and then to changing population composition.
One observation rarely establishes the whole chain. A visible structure may show environmental modification without demonstrating selection. A variant becoming common may show a population change without revealing why it happened. The task is to assign each piece of evidence to the specific causal arrow it can support.
This is an experiment-reading lesson within Ecology. The broader concept belongs to the existing Landscape Engineering and Niche Construction article. Here, a published oxygen-gradient case introduces an original paper-only worksheet about measurements, comparisons and inference.
The worked numbers below are invented and are not results from the cited research. They provide no culture, chemical-handling, variant-engineering or environmental-release procedure. Read the evidence and calculate on paper; do not grow unknown organisms or attempt to reproduce microbial evolution experiments.
Contents
Separate causal arrows and environmental measurements · Chapters 1–3
1. Separate the three arrows before reading the results
2. A published oxygen-gradient case supplies the evidence types
Calculate frequencies and name the statistic · Chapters 4–6
4. Calculate variant frequencies separately from total growth
5. Name the statistic you actually computed
6. Change the starting mixture to expose a hidden assumption
Audit comparisons, feedback and environmental inheritance · Chapters 7–10
7. Ask what X and Y actually changed
8. Complete a four-column evidence worksheet
CHAPTER 1 OF 10 · Separate causal arrows and environmental measurements
1. Separate the three arrows before reading the results
Draw three boxes: organism activity, environmental state and relative success of heritable variants. Draw an arrow from activity to the environment, then from the environment to relative success. Add a fourth box for the resulting population composition. This diagram gives the evidence a set of distinct jobs.
The first arrow asks whether organisms changed an environmental variable. The second asks whether that variable’s state changes the comparison between variants. The third asks whether the difference in relative success changes their representation in the population. A feedback loop arises when the resulting population also influences its surroundings.
A useful competing diagram begins with an external environmental change that affects both organism activity and variant success. In that version, the organisms did not generate the relevant change themselves. Another competing explanation is that organisms change the environment, but the measured change has no effect on selection.
These alternatives can produce superficially similar photographs or time series. To distinguish them, the study needs measurements and comparisons that do more than show two events occurring together.
Heritable variation is important because a reversible change in the same individuals is not automatically evolutionary change. A population can become more active, larger or differently shaped through a plastic response. The interpretation should specify whether it concerns within-organism adjustment, sorting among inherited variants or both. The familiar word “adaptation” should not be allowed to replace that distinction.
CHAPTER 2 OF 10 · Separate causal arrows and environmental measurements
2. A published oxygen-gradient case supplies the evidence types
Koza and colleagues studied Pseudomonas fluorescens and the biofilm-forming Wrinkly Spreader variant. Their paper appeared online in 2010 and in a 2011 journal issue. Oxygen profiles showed that early colonists generated gradients before substantial population growth. Initially sterile profiles and a respiration-inhibition comparison supported an organism-driven origin. Defined-variant competition showed that the biofilm-forming variant’s relative advantage changed with oxygen conditions: it was favoured under the higher-oxygen comparison and disadvantaged under the lower-oxygen comparison. The authors also followed the appearance of variants in diversifying populations. Together, these evidence types linked environmental modification to environment-dependent competitive outcomes. Original research.
The case is valuable because environmental profiles, competition and population observations answer different questions. The study’s logarithmic competitive-fitness statistic is not the simple multiplication-factor ratio used in our invented worksheet below.
From this point, treat the numerical records as a separate classroom construction. Their conditions X and Y have no specified culture recipe, oxygen concentration, exposure duration or organism. The comparison is deliberately abstract so that the reasoning can be checked without converting a research paper into a protocol.
The aim is to identify which additional evidence would be needed for a conclusion, rather than to declare that every line in a worksheet reproduces a completed laboratory test.
CHAPTER 3 OF 10 · Separate causal arrows and environmental measurements
3. Read an environmental profile as a measurement
Imagine an instrument reporting a resource index on an arbitrary scale. In an original fictional dataset, the initial shallow and deeper readings are both 100 units. At a later matched observation, a no-organism reference still reads 100 and 100, while an organism-containing system reads 100 and 30.
The vertical contrast in the latter system is 100 − 30 = 70 index units, compared with zero in the reference. This pattern supports an association between organism presence and development of a spatial difference under the compared conditions. It is more informative than simply seeing a final low reading without knowing the starting state.
It does not yet identify which biological process caused the change. Consumption, altered transport or another organism-associated effect could contribute. Nor does it establish that the measured index is the variable responsible for later variant success. The first arrow can gain support while the second remains untested.
Diagram to sketch: place position on the vertical axis and resource index on the horizontal axis. Draw the initial and reference profiles at 100 throughout the two sampled positions. Draw the later organism-associated profile connecting the shallow value of 100 to the deeper value of 30. Mark that there are only two measured positions; the connecting line is a visual aid, not proof of a smooth gradient between them.
Calibration, measurement timing and instrument disturbance also matter. A drifting instrument or mismatched observation could create an apparent change. The general question is whether the comparison distinguishes a changed environment from a changed way of measuring it. This remains a measurement problem before it becomes an evolutionary explanation.
CHAPTER 4 OF 10 · Calculate frequencies and name the statistic
4. Calculate variant frequencies separately from total growth
Now introduce hypothetical heritable variants A and B. In both conditions X and Y, each begins at 100 arbitrary count units. Assume counts represent the same biological quantity, observations cover the same comparison interval, and classifications are consistent.
In X, A finishes at 800 and B at 400. Total count is 1,200, and A’s final frequency is 800/1,200 = 2/3, approximately 66.7%. Its starting frequency was 100/200 = 1/2. A therefore rises by about 16.7 percentage points.
In Y, A finishes at 200 and B at 400. Total count is 600, and A’s final frequency is 200/600 = 1/3, approximately 33.3%. Its frequency falls by about 16.7 percentage points, even though its own count has doubled.
This is the important distinction: A can increase in absolute number while declining relative to B. A photograph showing more A at the end would miss the relative outcome that matters for the population’s composition.
The totals also differ between conditions. Both populations grew, but the change in A’s share went in opposite directions. A faster-growing whole population does not tell us which variant is favoured within it.
These records describe net outcomes over an interval. Counts alone need not separate reproduction from survival, movement or transitions between categories. For the simplified interpretation here, assume a closed comparison with faithfully inherited categories and no switching. If those assumptions fail, frequency change can have causes other than differential success of the two inherited variants.
CHAPTER 5 OF 10 · Calculate frequencies and name the statistic
5. Name the statistic you actually computed
A’s multiplication factor in X is 800/100 = 8; B’s is 400/100 = 4. The ratio of their multiplication factors is 8/4 = 2. In Y the corresponding ratio is (200/100)/(400/100) = 2/4 = 0.5.
These are finite-interval descriptive ratios. In X, A multiplied twice as much as B relative to their own starting counts. In Y, its multiplication factor was half B’s. The numbers are not universal properties of A and B across all environments, durations or starting mixtures.
The same calculation can be expressed through odds. Initially A:B is 1:1. It becomes 2:1 in X and 1:2 in Y. Multiplying starting odds by the relative multiplication factor gives the final odds for these records.
If one instead computes a ratio of logarithmic multiplication factors, the numerical values differ. For X, ln(8)/ln(4) = 1.5; for Y, ln(2)/ln(4) = 0.5. The equality of the Y values is incidental, not a reason to treat the two statistics as interchangeable. Here both logarithms in the denominator are nonzero because B multiplied fourfold.
A ratio, a logarithmic ratio, a per-generation selection coefficient and a change in frequency answer related but distinct questions. Before comparing two papers, inspect their definitions and time bases. A large-looking number can reflect a different denominator rather than a stronger biological effect.
The worksheet’s primary descriptive results remain frequencies of two-thirds and one-third, with multiplication-factor ratios of two and one-half. Their meaning is clear because the starting values, ending values and assumptions are visible.
CHAPTER 6 OF 10 · Calculate frequencies and name the statistic
6. Change the starting mixture to expose a hidden assumption
The first worksheet began with equal counts. To see why starting composition matters, use the same hypothetical multiplication factors in X but start with 25 units of A and 75 of B. After multiplying by eight and four respectively, the counts become 200 and 300.
A’s final share is now 200/500 = 0.4, or forty per cent. It has increased from twenty-five per cent, yet it remains the minority. Being favoured over an interval does not mean becoming the majority during that interval.
More generally, if A starts at frequency p and its multiplication factor relative to B is R, the resulting frequency is p next = pR/[pR + (1 − p)], provided those multiplication factors apply to that mixture. With p = 0.25 and R = 2, the expression gives 0.5/(0.5 + 0.75) = 0.4.
The qualification is essential. In real interacting populations, relative success may depend on starting frequency, density, resource availability and environmental change during the comparison. Reusing R from one mixture in another is a model assumption, not something the first dataset proves.
The environment may itself change as the mixture changes. A variant that becomes common could alter resource use or spatial structure, changing its own later advantage. A single endpoint contrast therefore cannot establish a fixed long-term trajectory, inevitable dominance or stable coexistence.
This is why competition at defined starting conditions complements observations of newly appearing variants. It can clarify relative performance without resolving every question about how variants originated or how their interactions evolve later.
CHAPTER 7 OF 10 · Audit comparisons, feedback and environmental inheritance
7. Ask what X and Y actually changed
The original count worksheet establishes a reversal in relative multiplication between X and Y under its assumptions. It does not identify the cause of that reversal. To connect it to the environmental-profile worksheet, X and Y must be characterized in terms of the relevant measured environmental state.
A convincing comparison would ask whether the same proposed resource difference is present, whether important alternatives also changed, and whether the variants encountered those conditions in comparable ways. A condition label is not a measurement.
For example, static and shaken settings differ physically in more ways than a single resource index. Mixing can alter transport, spatial arrangement and access to surfaces. A contrast between them cannot automatically be interpreted as an oxygen-only intervention. Similarly, a metabolic inhibitor can affect several biological processes; it is supporting evidence whose side effects must be considered, not a magic switch that isolates one arrow perfectly.
The appearance of a new variant creates another distinction. More frequent appearance could reflect more opportunities for variants to arise, better survival of new variants, or stronger subsequent expansion. If a condition changes population growth, those possibilities can change together.
None of this implies that organisms generate the exact useful mutation because the environment demands it. Evidence of selection among variants does not demonstrate directed mutation. The origin of variation and its subsequent differential success need separate tests.
The existing experiment-design guide develops randomization, replication and confounding in detail. Here their specific job is to prevent one environmental label from silently standing for several causal changes.
CHAPTER 8 OF 10 · Audit comparisons, feedback and environmental inheritance
8. Complete a four-column evidence worksheet
Use four headings on paper: observation, supported inference, useful comparison, remaining alternative. Fill them with claims narrow enough that another reader could disagree with a particular link rather than with the whole story at once.
Record one: the fictional resource profile changes in the organism-containing system but not the reference. Supported inference: organism presence is associated with the measured environmental alteration under the compared conditions. Useful comparison: a baseline that tracks abiotic change and measurement behaviour. Remaining alternative: some other difference between the systems caused the profile change.
Record two: A’s relative multiplication reverses between X and Y. Supported inference: relative net success depends on the conditions. Useful comparison: defined variants with recorded starting and ending counts under both conditions. Remaining alternative: X and Y differ in another relevant variable, or the count method detects the variants differently.
Record three: the resource state that organisms generate matches the state associated with A’s advantage. Supported inference: the two previously separate arrows can form a plausible connected explanation. Useful comparison: evidence that isolates the proposed environmental link from correlated changes. Remaining alternative: the matching resource measurements are only a proxy for the actual selective factor.
Record four: the changing population subsequently changes the environment again. Supported inference: a feedback loop is plausible. Useful comparison: time-resolved evidence linking population composition to later environmental states. Remaining alternative: an external trend drives both series.
These are original prompts, not a list of experiments claimed to have been performed. Their value lies in distinguishing what each record contributes. No single result is asked to prove the entire feedback loop.
CHAPTER 9 OF 10 · Audit comparisons, feedback and environmental inheritance
9. Environmental inheritance is an additional question
Inheritance of a genetic variant and persistence of a modified environment are different processes. Descendants may inherit alleles while also encountering a resource state left by earlier organisms. Demonstrating the second process requires evidence that the environmental modification lasts long enough to affect later recipients.
An original abstract memory model makes this distinction visible. Let L represent the magnitude of an environmental legacy on an arbitrary scale. After the original constructors are absent, suppose half the legacy remains after each equal interval: L next = 0.5L. Starting at eight units, it becomes four, two and one over three intervals.
This calculation describes persistence under an assumed decay rule. It says nothing about genetic transmission and does not show that the remaining legacy changes selection. A separate relationship between L and relative variant success would be needed to connect the memory model to evolution.
Laland, Odling-Smee and Feldman’s 1999 theoretical study coupled genetic variation to resource dynamics, including organism effects and independent renewal and depletion. Its two-locus model showed that such coupling can change evolutionary outcomes under specified assumptions. This was a theoretical analysis, not an environmental-inheritance experiment. Original paper abstract.
The oxygen-gradient case does not establish every form of ecological inheritance. Nor is this lesson simply a succession account: succession asks how community composition changes through time, whereas the present question asks whether organism-generated conditions alter selection among inherited variants. The processes can overlap without becoming the same scientific claim.
CHAPTER 10 OF 10 · Audit comparisons, feedback and environmental inheritance
10. State what closes the loop and what remains open
A strong evidence summary identifies the organism-associated environmental change, the condition-dependent comparison of heritable variants and the population consequence. It then names the alternatives addressed by the comparisons and the uncertainties that remain. The conclusion should be no broader than that chain.
Check the original arithmetic. Does A’s count increase in Y? Yes, from 100 to 200. Does its frequency increase? No, it falls from one-half to one-third. Is its multiplication-factor ratio in X equal to the ratio of logarithmic multiplication factors? No: the former is two and the latter is 1.5. Does starting at twenty-five per cent with R = 2 make A the majority after one model interval? No; it reaches forty per cent.
Finally, ask what is missing from the numbers alone. They do not show that organisms created X, that one measured resource caused the relative advantage, that variation arose in a directed way, or that a legacy persisted after its constructors disappeared. Those questions belong to additional evidence rather than more confident wording.
A changed microbial environment is also not automatically a communication signal. For the separate question of signal production, sensing and response, use the existing Quorum Sensing and Biofilm Evidence guide.
The transferable skill is to read a feedback claim one arrow at a time. Continue with the broader Niche Construction article, return to Ecology, or choose another mechanism in the How Science Works master guide.