Counting an animal population immediately before breeding and counting it immediately afterwards can give different totals without any disagreement about the animals. Newborns have entered one count but not the other. A projection must preserve that timing in its classes, survival terms and reproduction terms.
The short answer is to write the life cycle first, place the census on it, and compose the same biological events in the appropriate order. A consistent change of census timing changes the representation of our fixed age-based model. It does not create a new biological growth rate merely by moving the observation point.
This guide follows an original fictional population with a brief breeding event and two reproductive ages. Every survival probability, offspring expectation and starting count is invented. The calculations are expected-value accounts, not field estimates, capture instructions or animal-management advice. The Zoology subject guide supplies the wider connection between animal evidence, explanation and responsibility.
eduKateSG · Zoology
Find your next learning step
Follow one declared animal life cycle through births and survival before comparing census representations.
ROUTE 1 · CHAPTERS 1–5
Define animals and events
Label census timing, counted classes and the two event maps.
ROUTE 2 · CHAPTERS 6–10
Reconcile annual projections
Check successive histories and derive the positive annual factor.
ROUTE 3 · CHAPTERS 11–15
Inspect the conversion
Connect growth factors, compositions and common coefficient errors.
ROUTE 4 · CHAPTERS 16–20
Change biology and observation
Separate intermediate censuses, stage durations and changed survival.
ROUTE 5 · CHAPTERS 21–25
Check evidence and practise
Use limiting cases, transparent inputs and independent transfer.
Chapter contents
Define animals and events · Chapters 1–5
1. Put the breeding event on a timeline
2. Define ages and the counted population
3. Keep raw reproduction separate from census recruitment
Reconcile annual projections · Chapters 6–10
6. Project from one pre-breeding census to the next
7. Project from one post-breeding census to the next
8. Reconcile the two accounts at corresponding times
9. Follow another full year without changing definitions
10. Derive the long-run growth factor in the smaller account
Inspect the conversion · Chapters 11–15
11. Explain why the corresponding growth factor survives the conversion
12. Compare compositions without equating their class proportions
13. Do not demand an inverse where information was discarded
Change biology and observation · Chapters 16–20
16. Treat an observation date inside the year as a new modelling question
17. Keep stage models from becoming a careless relabelling
18. Change a biological input and identify the genuine effect
Check evidence and practise · Chapters 21–25
21. Check totals at each operation before trusting a projection
22. Use a no-birth limit to inspect the model’s memory
23. Build an evidence check that follows the animals
CHAPTER 1 OF 25 · Define animals and events
1. Put the breeding event on a timeline
Imagine a population with one idealised breeding pulse per year. The pulse is treated as instantaneous for this model. Between pulses, animals experience the specified survival process and advance in age. We will count just before a pulse or just after one, rather than at an unspecified date somewhere in the year.
The word just matters. There is no mortality between these two adjacent census points in the initial model; the breeding operation itself leaves the existing parents alive. Later annual survival determines whether they reach the next pulse. These choices make the two operations distinct and testable.
A continuous breeding season, deaths during reproduction or migration would require a changed account. They cannot be added silently while retaining formulas derived for our pulse model. This is why a life-cycle diagram is a reasoning tool rather than an ornament beside a matrix.
Write the repeating order in words: pre-breeding population, births, post-breeding population, annual survival and ageing, next pre-breeding population. Then place a marker at the chosen census. One projection step must return to the same kind of marker one year later. That loop defines what a per-year multiplier means.
CHAPTER 2 OF 25 · Define animals and events
2. Define ages and the counted population
Our account counts females only, and reproduction means expected female newborns per female parent at the relevant pulse. A mixed-sex offspring total would need an appropriate conversion before being inserted. We are not modelling mating limitation, sex-ratio dynamics or individual reproductive variability.
The youngest pre-breeding class is labelled age one. These animals are reaching their first breeding pulse. The second is age two. Immediately after a pulse, the account also includes age-zero newborns. Existing age-one and age-two parents remain in their respective classes at that instant.
Maximum age is two in this invented life cycle. Age-two animals can reproduce at the current pulse, but none survive to a third pulse. That is a model assumption about the interval after their final reproduction. It is not a claim about a particular species or a conclusion inferred from a missing sighting.
These definitions explain why the pre-breeding vector has two entries while the post-breeding vector has three. The extra newborn class is meaningful, not an administrative nuisance to remove by force. A conversion must respect the animals represented by each position rather than equating first entries simply because they occupy the same place on the page.
CHAPTER 3 OF 25 · Define animals and events
3. Keep raw reproduction separate from census recruitment
Let b₁ equal one and b₂ equal two. These are expected newborn females per age-one and age-two female at a breeding pulse. Let p₀ equal 0.5 for survival from birth to the next pulse, p₁ equal 0.8 from age one to age two, and p₂ equal zero after age two.
A raw offspring expectation of two is not automatically a coefficient of two in every projection. A pre-breeding census next year includes only newborns that survived their first annual interval. A post-breeding census next year includes births produced by parents that survived to that next pulse.
The coefficients therefore combine different events depending on the source and destination classes. Naming the multiplication matters more than memorising which row usually contains a particular product.
For a quick verbal test, take one age-two female immediately before breeding. She produces an expected two newborns now. At the next pre-breeding census, their expected contribution is one because newborn survival is one-half. She herself contributes nothing to the next adult classes because p₂ is zero. The complete contribution is derived from the timeline, not from copying the raw birth expectation into an unlabeled table.
CHAPTER 4 OF 25 · Define animals and events
4. Build the births operation from an actual count
Suppose the initial pre-breeding state is 40 age-one females and ten age-two females. The pulse produces 40 times one plus ten times two, or 60 expected newborn females. Existing parents remain, so the post-breeding state is 60 newborns, 40 age-one females and ten age-two females.
Call the pre-breeding vector y and the post-breeding vector x. The births operation B maps y to x. Written row by row, B has rows (1, 2), (1, 0) and (0, 1). Its first row calculates newborns; the remaining rows carry the existing parental classes across the pulse.
The retained parents must not be counted as newborns. Equally, omitting the parent rows would incorrectly erase adults at the pulse before their specified annual survival had acted. Each row has a biological interpretation that can be checked against the original 40-and-ten account.
This is a useful way to learn matrices without letting notation get ahead of understanding. Calculate one pulse with ordinary arithmetic, explain where all three resulting classes came from, then recognise the matrix as a compact record of those operations. The symbols should shorten a clear account rather than conceal an unclear one.
CHAPTER 5 OF 25 · Define animals and events
5. Build survival and ageing as a separate map
Starting from a post-breeding vector x, the next pre-breeding state has two classes. Half the newborns become age-one females, and four-fifths of age-one females become age-two females. Age-two females have no surviving class in this model.
Call this map P. Its rows are (0.5, 0, 0) and (0, 0.8, 0). Applying it to (60, 40, 10) gives (30, 32). Those are expected values at the next pre-breeding census.
Notice what the zero final column means. The ten age-two parents were present just after the earlier pulse, but they do not contribute directly through survival to the next census. Their offspring were already counted in the newborn component. Adding another contribution from those parents would double-count their reproduction.
The dimensions also check the meaning. P accepts a three-class post-breeding vector and returns a two-class pre-breeding vector. It is not interchangeable with B merely because both concern the same population. Write the input and output labels beside each map before composing them. That habit catches an incorrect order even when the numerical products look plausible.
CHAPTER 6 OF 25 · Reconcile annual projections
6. Project from one pre-breeding census to the next
A pre-breeding projection begins with y, applies births B, then annual survival and ageing P. Its complete map is A_pre = PB, with the rightmost operation acting first on a column vector.
Multiplying the declared maps gives rows (0.5, 1) and (0.8, 0). Thus next year’s age-one expectation is 0.5 times this year’s age-one count plus one times its age-two count. Next year’s age-two expectation is 0.8 times this year’s age-one count.
For y equal to (40, 10), the result is (30, 32). It matches the event-by-event account. The coefficient one in the first row’s second position is p₀b₂: two newborns per age-two parent, each with survival one-half to the next pre-breeding census.
That interpretation is the important check. Reporting the coefficient as a raw birth rate would be wrong, even though it occupies a reproduction-related row. It is recruitment to the particular destination census class. In a real study, the data definitions must establish which components were already included before multiplying by survival again.
CHAPTER 7 OF 25 · Reconcile annual projections
7. Project from one post-breeding census to the next
A post-breeding projection begins with x, applies annual survival and ageing P, then births B at the next pulse. Its complete map is A_post = BP. The biological operations are the same, but the census loop starts at a different point.
The resulting rows are (0.5, 1.6, 0), (0.5, 0, 0) and (0, 0.8, 0). From the initial post-breeding state (60, 40, 10), the next state is (94, 30, 32).
The 1.6 coefficient is p₁b₂. An age-one female present just after the first pulse survives with probability 0.8 to reproduce at age two during the next pulse, producing an expected two newborn females conditional on reaching that event. The product is 1.6 expected newborns at the next post-breeding census.
There is no extra p₀ multiplying those newborns in this particular coefficient. They have just been born at the destination pulse, so the model has not yet applied their first annual survival. Adding it here would move the destination to another point in the life cycle while still calling the result post-breeding.
CHAPTER 8 OF 25 · Reconcile annual projections
8. Reconcile the two accounts at corresponding times
The pre-breeding account gave (30, 32) at the next pulse. Applying births to that state gives 30 plus twice 32, or 94 newborns, with the existing 30 and 32 parents retained. This is exactly (94, 30, 32), the post-breeding projection’s result.
The two accounts agree because we compared corresponding states. We did not feed the same unlabeled numerical vector into differently defined models and demand the same output. We transformed the initial state and the destination state consistently.
| Position in the fictional annual history | Expected state | Counted classes |
|---|---|---|
| First pre-breeding census | 40, 10 | Age one, age two |
| First post-breeding census | 60, 40, 10 | Newborn, age one, age two |
| Next pre-breeding census | 30, 32 | Age one, age two |
| Next post-breeding census | 94, 30, 32 | Newborn, age one, age two |
The totals 50, 110, 62 and 156 answer different timing questions. A rise from 50 to 110 across the first pulse is not an annual growth estimate. Keeping the census label beside each total prevents a change in what was counted from masquerading as a demographic discovery.
CHAPTER 9 OF 25 · Reconcile annual projections
9. Follow another full year without changing definitions
Continue from the next pre-breeding state (30, 32). The pulse produces 94 newborns. Annual survival then gives 47 age-one females and 24 age-two females at the following pre-breeding census.
The subsequent pulse produces 47 plus twice 24, or 95 newborns. Its post-breeding state is therefore (95, 47, 24). Applying A_post directly to the previous post-breeding state (94, 30, 32) gives the same result: 0.5 times 94 plus 1.6 times 30 equals 95 for the newborn row.
This second cycle is more informative than checking one convenient starting state. It tests whether the relationship persists when the age composition has changed. The pre-breeding total moves from 62 to 71, while the post-breeding total moves from 156 to 166. Their one-year total ratios differ.
Different transient total ratios do not contradict equivalent life-cycle representations. They sum different classes at different points around the pulse. To compare a particular annual total ratio, retain the census convention; to compare long-run model growth, examine the corresponding asymptotic structure rather than selecting whichever short interval gives the preferred headline.
CHAPTER 10 OF 25 · Reconcile annual projections
10. Derive the long-run growth factor in the smaller account
Suppose a positive age composition is multiplied by the same factor λ at successive pre-breeding censuses. Let its age-one component be u and age-two component be v. The projection requires λu = 0.5u + v and λv = 0.8u.
Substituting v = 0.8u/λ into the first equation gives λ² = 0.5λ + 0.8. The positive solution is [0.5 + √3.45]/2, approximately 1.178709. This is the dominant annual factor for the declared fixed age-based account.
The calculation does not say every starting vector grows by exactly that ratio in its first year. It identifies a composition whose expected class counts scale together. Our earlier starting state (40, 10) was not that composition, which is why its first total ratio need not equal λ.
Check the equation forward: square the numerical factor and compare it with half the factor plus 0.8. This tests arithmetic independently of the story. Then return to the biological units: λ is a dimensionless multiplier per annual census interval, not an offspring count per parent or a probability of surviving one year.
CHAPTER 11 OF 25 · Inspect the conversion
11. Explain why the corresponding growth factor survives the conversion
If A_pre y = λy, take the corresponding post-breeding vector By. Applying its projection gives A_post(By) = BPBy = B(A_pre y) = λBy. The corresponding post-breeding composition therefore has the same nonzero growth factor.
This short derivation states the required relationship. The population vectors must be related by the declared births map. Merely calling two independently assembled matrices pre-breeding and post-breeding does not establish the equality.
For our positive composition, By is nonzero because B retains both parental classes. The argument consequently produces an actual post-breeding composition. It is not multiplying a disappearing vector and treating zero as evidence of population growth.
The demographic construction guide by Bruce Kendall develops the event-map approach for age-based projections. The example here uses different invented inputs and follows the states explicitly. Our proof’s strength is also its boundary: it applies to two representations constructed from the same fixed maps, not to every pair of animal models whose authors chose different census dates.
CHAPTER 12 OF 25 · Inspect the conversion
12. Compare compositions without equating their class proportions
Set the stable pre-breeding age-one component to one. The corresponding age-two component is 0.8/λ, approximately 0.678709. Its age-one share is therefore one divided by about 1.678709, approximately 59.57%.
After births, the corresponding post-breeding vector is approximately (2.357418, 1, 0.678709). Newborns now occupy a substantial share of the total. The age-one fraction of this larger three-class total is much smaller than its fraction in the pre-breeding account.
There is no disagreement about the population’s underlying annual process. The denominator changed because the census crossed a birth pulse. A class percentage must retain both its class definition and its census time.
This offers a useful interpretation check for reports. If two surveys describe different age compositions, ask whether they sampled comparable points in the breeding season and used compatible age categories. Timing is one possible explanation, not a conclusion to impose without evidence. The mathematical example shows how a difference can arise; field attribution requires the actual survey and biological records.
CHAPTER 13 OF 25 · Inspect the conversion
13. Do not demand an inverse where information was discarded
P maps three post-breeding classes into two pre-breeding classes. Its final column is zero. Changing the post-breeding age-two count while holding the other classes fixed does not change the next pre-breeding vector through P.
For example, post-breeding states (60, 40, 10) and (60, 40, 25) both map to (30, 32). The extra age-two animals in the second supplied state have no future survival in this model. P alone cannot recover their earlier number from the later state.
This is not a licence to invent any earlier count. A post-breeding state produced by B from a particular pre-breeding state has additional consistency constraints. In our model, its newborn component must equal its age-one component plus twice its age-two component. The second vector above does not satisfy that pulse-generated relationship with the same other components.
The lesson is to distinguish projection from reconstruction. A forward map can deliberately discard information irrelevant to later dynamics. An inverse claim needs enough additional evidence or constraints to recover it. Do not insert a matrix inverse simply because a census-conversion explanation sounds as though it should work backwards automatically.
CHAPTER 14 OF 25 · Inspect the conversion
14. Catch survival counted twice in recruitment
Suppose a report defines a quantity as expected surviving age-one recruits per age-two female at the previous pre-breeding pulse. Under our inputs, that quantity is already one. Multiplying it by p₀ again would reduce the coefficient to 0.5 and count newborn survival twice.
The word fertility does not settle this issue. A source may use it for raw births, surviving recruits or a matrix coefficient combining several events. Read the explicit definition and observation endpoints before choosing the multiplication.
A simple diagnostic asks the learner to reconstruct one parent’s expected contribution in words. Two newborns are expected at the pulse; half survive to the next pre-breeding census; the contribution is one. If the proposed coefficient differs, identify which event was added, omitted or moved.
This check is more reliable than insisting that every first-row term must have the same visual form across sources. Different census conventions legitimately produce different products. What must remain consistent is the path from a defined source animal through the specified events to a defined destination class.
CHAPTER 15 OF 25 · Inspect the conversion
15. Catch survival assigned to the wrong parent age
The post-breeding coefficient from current age-one animals to next year’s newborns is 0.8 times two, or 1.6. It uses survival from age one to age two followed by reproduction at age two. Using b₁ instead would attach reproduction to the animal’s earlier age rather than its age at the destination pulse.
A mistaken coefficient of 0.8 may therefore contain the right survival number and still represent the wrong life cycle. Correct-looking components do not guarantee a correct product if their timing is mismatched.
Follow a single labelled fictional female through the year. Immediately after the first pulse she is in age one. At the next pulse, if she survives, she is in age two. Her current class label is a source label, while the reproduction input belongs to the event she reaches.
This tracing habit transfers to many demographic models. Ask which age, stage or condition applies when the event happens. Do not choose a rate merely because its subscript matches the source column. A source-column coefficient can legitimately depend on a later state reached within the projection interval.
CHAPTER 16 OF 25 · Change biology and observation
16. Treat an observation date inside the year as a new modelling question
A census several months after breeding is not automatically equivalent to either endpoint census. Newborns have already experienced part of their first interval, while older animals have experienced part of theirs. The model needs survival and class definitions appropriate to that actual date.
Our annual probabilities do not determine every within-year survival history. Survival of one-half across a year could arise through several different time patterns. Assigning a fractional-year probability by taking a power introduces an additional constant-hazard assumption; it is not information contained in the annual total alone.
For a paper comparison, imagine all first-year losses occur early in one model and late in another, with the same annual survival one-half. A midyear census can differ even though endpoint survival agrees. This demonstrates the missing information without claiming either history describes a real animal population.
The right next step is to identify the observation date and the biological evidence available at that resolution. A more detailed projection may be justified, or the conclusion may need to remain at the annual scale. More decimals in an annual coefficient cannot manufacture an unobserved within-year schedule.
CHAPTER 17 OF 25 · Change biology and observation
17. Keep stage models from becoming a careless relabelling
Our example advances animals through fixed age classes. A developmental stage can last different amounts of time for different individuals. Replacing age labels with stage labels while keeping superficially similar matrices can therefore change the life history being represented.
Okuyama’s 2019 study of census timing and stage-duration distributions shows why common stage-based formulations need careful attention to those distributions. A nominal change between pre- and post-breeding conventions can introduce different minimum or variable stage durations if the construction is not biologically matched.
That warning limits the earlier growth-factor statement. It was proved for our explicitly shared B and P maps. It was not a claim that every conventional two-stage formula is equivalent merely after exchanging a survival term in its top row.
A learner extending the example should first ask whether animals advance deterministically with age or can remain in a stage. If residence time varies, define the transition process and its timing. The stage-structured construction chapter is a specialist route into that additional responsibility. Preserve the age example’s clarity while allowing the model itself to change when the biology changes.
CHAPTER 18 OF 25 · Change biology and observation
18. Change a biological input and identify the genuine effect
Hold the census construction fixed but reduce first-year survival from 0.5 to 0.25. This is a changed biological model, not a different observation convention. Raw reproduction remains one and two, and age-one survival remains 0.8.
The new pre-breeding rows are (0.25, 0.5) and (0.8, 0). Its positive growth factor solves λ² = 0.25λ + 0.4. The solution is approximately 0.769690, below one. A consistently rebuilt post-breeding model gives the same factor for the corresponding history.
Compare that with merely moving the census in the original life cycle, which preserved its factor of approximately 1.178709. The distinction is now concrete: changing survival alters an input; changing the representation preserves the supplied input and moves its place in the account.
Check every coefficient affected by the biological change. Altering only a headline survival value while leaving a composite recruitment term untouched creates an inconsistent model. A useful audit keeps raw inputs separate from derived coefficients so their dependencies can be rebuilt transparently.
CHAPTER 19 OF 25 · Change biology and observation
19. Explain what a total count can and cannot reveal
Suppose an observer reports a total of 110 females with no age composition or census date. Our first post-breeding vector happens to sum to 110, but that does not establish that this was the state observed. Many age compositions share the same total, and different dates represent different points in the cycle.
Even at a fixed pre-breeding census, 40 age-one and ten age-two animals have total 50, as do ten age-one and 40 age-two. Their expected newborn outputs are 60 and 90 respectively. The same total carries different reproductive implications under the declared inputs.
Animal detection introduces another layer. A recorded total may not equal the true population total. The existing guide on all-zero survey histories and imperfect detection addresses one specific observation problem. Do not substitute our exact expected-state exercise for a field-estimation method.
A useful report therefore distinguishes population state, observation process and projection model. Each contributes different evidence. A neat matrix cannot establish the unseen age composition, and an impressive survey count cannot by itself validate every demographic coefficient.
CHAPTER 20 OF 25 · Change biology and observation
20. Attempt a fresh pulse-and-survival account
Use a new invented population with b₁ equal to zero, b₂ equal to three, p₀ equal to 0.4 and p₁ equal to 0.5. Maximum age remains two, with no survival after age two. Begin immediately before breeding with (20, 10).
First calculate births: ten age-two females contribute 30 expected newborn females. The post-breeding state is (30, 20, 10). Annual survival gives the next pre-breeding state (12, 10). The following pulse produces 30 newborns again, leaving post-breeding state (30, 12, 10).
The new pre-breeding matrix has rows (0, 1.2) and (0.5, 0). The post-breeding matrix has rows (0, 1.5, 0), (0.4, 0, 0) and (0, 0.5, 0). Apply both to their correctly matched starting states and verify the same event history.
The positive growth factor solves λ² = 0.6, giving approximately 0.774597. Because this example has reproduction only at the later age, transient patterns can alternate with initial composition. Do not assume every adjacent total ratio equals the positive factor. The earlier reproduction-schedule article develops that separate timing-and-periodicity issue.
CHAPTER 21 OF 25 · Check evidence and practise
21. Check totals at each operation before trusting a projection
Each event offers a simple accounting check independent of the matrix multiplication. Across our births-only pulse, the post-breeding total must equal the pre-breeding total plus expected newborns. Starting with (40, 10), that is 50 plus 60, or 110. No parent disappears at this particular event.
Across annual survival and ageing, no new animals enter in our model. The next pre-breeding total cannot exceed the preceding post-breeding total because every supplied survival probability lies between zero and one. From (60, 40, 10), the result is 30 plus 32, or 62, which is below 110.
These checks do not establish that the exact class allocation is correct. A mistaken projection could preserve a plausible total while assigning survivors to the wrong ages. They are useful because they catch a different family of errors: births omitted, parents counted twice or survival creating animals from nothing.
Try a deliberate mistake. If someone adds newborns to a post-breeding vector again before applying annual survival, the second pulse has been inserted at the wrong place. Ask them to locate that extra event on the original timeline. The defect becomes visible in the sequence even if the final total happens to look biologically plausible.
CHAPTER 22 OF 25 · Check evidence and practise
22. Use a no-birth limit to inspect the model’s memory
Set both birth expectations to zero while leaving our maximum age and survival inputs unchanged. After one full year, an initially age-one animal may remain as age two. After another, no member of the original population remains because no animal survives beyond age two and no new cohort was born.
The corresponding pre-breeding projection has rows (0, 0) and (0.8, 0). Applying it twice gives the zero vector for any initial two-class state. This is a structural check on the declared finite lifespan, not a prediction about any actual species.
If a proposed no-birth projection retains a positive class forever, it contains an unintended survival loop or a different lifespan assumption. The limiting case helps reveal what the equations really say. It is often easier to inspect an extreme simplified input first, then restore reproduction, than to diagnose a full matrix solely from its dominant growth factor.
Restore the original birth inputs after completing the limiting-case check. The zero-birth result was a temporary test of structure, not a new estimate of reproduction. Keeping a separate copy of the declared inputs prevents a useful diagnostic simplification from accidentally becoming the final reported model. State which version supports each conclusion.
CHAPTER 23 OF 25 · Check evidence and practise
23. Build an evidence check that follows the animals
For a supplied projection, ask for the census date, class definitions, counted sex or population basis, birth-rate definition and survival interval. Mark which inputs are measured, inferred or assumed. Then trace one animal from each source class through every event until the destination census.
Check a full numerical state by two methods: event-by-event arithmetic and the composed projection. Convert the initial state as well as the matrix when changing census. Compare corresponding destinations, and distinguish a transient total ratio from an asymptotic factor.
For real animal evidence, appropriate specialist design and review remain necessary. The Zoologists guide connects research questions with ethical field studies and traceable observations. The branch on survival bounds with unresolved follow-up records handles a different limitation in supplied fate evidence.
A calculation check should end with a specific conclusion: these two supplied projections describe the same declared event cycle under this conversion. It should not end with an unsupported claim that the population was measured accurately, the model fits a species or a particular conservation action has been established as best.
CHAPTER 24 OF 25 · Check evidence and practise
24. Choose practice that reveals the missing connection
For repair, use two parental age classes and one birth pulse, with no matrix notation. Ask where the newborns came from and which animals survive to the next pulse. Introduce notation only after the learner can preserve the class labels independently.
For stabilisation, alternate pre- and post-breeding starting states. Include a coefficient that already counts survival, another using the wrong reproductive age and a total without its census date. Require a verbal correction as well as revised arithmetic.
For extension, compare matched representations, test noninvertible maps and identify what an intermediate census or variable stage duration would require. The task should demand a changed assumption and a new prediction rather than a longer version of the original calculation.
The Gold Standard education parent connects explanation with independent transfer. The How X Works hub offers broader systems routes. Carry forward the same discipline: define the counted state, preserve event order and ask whether a changed result reflects changed biology or changed representation.
CHAPTER 25 OF 25 · Check evidence and practise
25. Questions, limits and authoritative routes
Does a post-breeding count have to be larger?
In our instantaneous births-only pulse, newborns are added while parents remain, so its total is at least as large as the immediately preceding count. That conclusion depends on the supplied pulse definition. A different model including simultaneous losses needs a different account.
Can I use the same starting vector in both matrices?
Only if its entries really represent the same defined states, which they do not here. Convert the state through the relevant event map. Matching a few numerical entries is not a substitute for matching their biological meaning.
Does census timing never affect a modelling result?
A consistent conversion of our fixed age-based life cycle preserves its corresponding nonzero growth factor. Careless parameter conversion or a changed stage-duration model can change the life history itself. Keep that boundary explicit.
Does an expected fractional animal count cause a problem?
Expected values can be fractional even when realised animals are discrete. They describe a model mean, not a physical fraction of an animal or a guarantee about a small population’s future.
Primary methodological routes checked on 23 October 2026 are Kendall’s age-structured construction and stage-structured construction, alongside Okuyama’s 2019 original study. The examples and arithmetic in this article are original and use no species-specific field dataset. Return to Zoology when the next question concerns actual animal evidence or the appropriate specialist responsibility.