The Gold Standard of Zeta Diversity | Reconstructing Occupancy Frequencies from a Complete Zeta Spectrum

Assume N named sites with a binary species-by-site incidence table. A one means that a taxon is recorded at that site under the stated observation rule. A zero represents the corresponding recorded absence, not an automatically confirmed biological absence.

Each recorded species occupies some number r of those N sites. Let fr be the number of species recorded at exactly r sites, for r from one to N. The list of fr values is an occupancy-frequency histogram of the observed table.

The word occupancy here describes recorded incidence unless an independent detection model has justified a different interpretation. We are reconstructing a mathematical summary of a supplied finite table. We do not estimate a population of species outside the observation frame merely by applying the inverse. Defining that frame is the first requirement for interpreting the result.

Contents · Next chapter

For each order k, consider every unordered subset of k distinct sites. Count the taxa shared by all sites in each subset, then average those intersection counts. Call this unnormalised mean ζk.

There are C(N,k) such subsets, where C denotes the binomial combination count. Every subset has equal weight. The inverse developed here depends on this complete, uniformly weighted definition.

A spatially restricted analysis, a nearest-neighbour combination rule or a weighted subset sample can answer a different and useful question. Its values must not automatically be inserted into the complete-combination inverse. Before reconstructing any histogram, record the sites, the orders, whether all combinations were enumerated, and what each reported zeta value actually averages. Matching a symbol is not sufficient when its sampling rule differs.

Contents · Previous chapter · Next chapter

Suppose one species occurs in exactly r of the N sites. It appears in the intersection of a k-site subset only when all k chosen sites are among those r occupied sites.

The number of contributing subsets is C(r,k) when r≥k and zero when r<k. Its contribution to the mean ζk is therefore C(r,k)/C(N,k).

This counting step is the engine of the calculation. A species in every site contributes one to every zeta order. A species in only one site contributes to order one and to no higher order. Intermediate occupancy classes contribute according to their combination counts. Nothing in this step requires a fitted ecological model or a proposed mechanism of community assembly; it follows from the incidence definition and the chosen averaging rule.

Contents · Previous chapter · Next chapter

All fr species with the same recorded occupancy r make the same contribution to a complete-combination mean. Summing their contributions gives ζk=sum over r≥k of fr C(r,k)/C(N,k).

Multiply by C(N,k) and define Mk=C(N,k)ζk. Then Mk=sum over r≥k of fr C(r,k). Mk is the total number of species-sharing contributions across all k-site intersections.

Mk is not the richness of one intersection or the total distinct richness of the landscape. A single widespread species can be counted in several k-site intersections. Keeping that repeated-count meaning visible is what makes the forward formula and its inverse intelligible. The multiplication removes the average denominator; it does not remove the overlap among combinations.

Contents · Previous chapter · Next chapter

At the highest order, MN=fN because only taxa present in all N sites can contribute. One order lower, MN−1=fN−1+NfN. The highest occupancy class has already been identified, so its repeated contribution can be subtracted.

Continuing downward gives fr=Mr−sum over j>r of fj C(j,r). This recurrence reconstructs one occupancy class at a time, beginning with fN.

The system is triangular because taxa occupying fewer than k sites cannot contribute to order k. Its diagonal coefficients C(r,r) are all one, so the exact full system has a unique solution. That statement concerns the finite recorded-frequency problem with every order supplied. It does not establish that the same spectrum uniquely determines the original labelled incidence matrix or the underlying biological community.

Contents · Previous chapter · Next chapter

Take four fictional sites A, B, C and D. Include sixteen site-restricted species, four unique to each site. Include six pair-restricted species, one for each of the six site pairs. Include four triple-restricted species, one for each triple. Finally include two species present in all four sites.

The occupancy frequencies are f1=16, f2=6, f3=4 and f4=2. There are twenty-eight distinct recorded taxa in total. This is a constructed teaching table, not a field survey.

The construction is balanced: each site contains four unique species, three pair species, three triple species and the two ubiquitous species, for richness twelve. Giving the incidence construction as well as the histogram lets another reader check that the nonnegative integer frequencies describe an actual possible table, rather than just a convenient list of summary numbers.

Contents · Previous chapter · Next chapter

The total single-site contributions are M1=16+2×6+3×4+4×2=48. Dividing by four sites gives ζ1=12.

For pairs, M2=6+3×4+6×2=30. Dividing by C(4,2)=6 gives ζ2=5. For triples, M3=4+4×2=12, so ζ3=12/4=3. At order four, M4=f4=2 and ζ4=2.

The complete spectrum is therefore (12,5,3,2). Every pair in this balanced construction shares five taxa, and every triple shares three. Those individual checks agree with the combination totals. The zeta values are average shared counts per subset at different orders, while the M values are repeated-contribution totals. Neither list should be relabelled directly as the occupancy histogram.

Contents · Previous chapter · Next chapter

Pretend the original incidence table and histogram are hidden, but N=4 and the full spectrum (12,5,3,2) are known exactly. First rebuild M=(48,30,12,2).

The fourth-order equation gives f4=M4=2. The third-order equation is M3=f3+4f4, so f3=12−8=4. The factor four counts the four triples contained within the occupancy set of each ubiquitous taxon.

This subtraction is not an adjustment for detection or an estimate of environmental change. It removes the combinatorial contribution of taxa already counted in the highest class. A common mistake is to set f3 equal to ζ3 or M3. Both retain contributions from species occurring in all four sites, so neither isolates taxa occurring in exactly three.

Contents · Previous chapter · Next chapter

At order two, M2=f2+3f3+6f4. Therefore f2=30−12−12=6. At order one, M1=f1+2f2+3f3+4f4, giving f1=48−12−12−8=16.

The recovered list matches the constructed histogram exactly. Summing it gives twenty-eight recorded taxa. A forward recalculation using these recovered frequencies reproduces every supplied zeta value.

The forward check is important because it tests the whole account rather than only one recovered class. A sign error in an inverse formula can produce plausible positive numbers that fail the original intersection totals. Preserve both directions in a worked analysis: convert zeta means to Mk, reconstruct the histogram, and regenerate the means under the same combination rule before interpreting the result.

Contents · Previous chapter · Next chapter

For four sites, expanding the recurrence gives f4=ζ4, f3=4ζ3−4ζ4, f2=6ζ2−12ζ3+6ζ4 and f1=4ζ1−12ζ2+12ζ3−4ζ4.

These equations show that a lower occupancy class generally depends on several zeta orders, not just the matching order. In particular, the site-restricted class f1 requires cancellation of contributions from every higher class.

The coefficients also foreshadow sensitivity to errors. Large terms of opposite sign may subtract to a much smaller occupancy count. This does not invalidate the exact identity, but it means that rounding or estimating the zeta values can affect the reconstructed histogram. We will distinguish exact enumeration from uncertain estimates rather than treating every numerical spectrum as an error-free incidence table.

Contents · Previous chapter · Next chapter

For any N, the direct inverse is fr=sum over k from r to N of(−1)^(k−r) C(k,r) Mk. Substituting Mk=C(N,k)ζk gives the same formula entirely in terms of the zeta spectrum.

To check it, insert Mk=sum over j≥k of fj C(j,k). The coefficient multiplying a particular fj becomes the alternating sum of C(k,r)C(j,k) over k from r to j.

Use C(j,k)C(k,r)=C(j,r)C(j−r,k−r). The remaining alternating binomial sum is zero when j>r and one when j=r. All unwanted occupancy classes therefore cancel, leaving fr. This proof explains both the inverse and its alternating signs. It also identifies exactly where complete, equally weighted combination totals entered the argument.

Contents · Previous chapter · Next chapter

The total number of recorded taxa is S=sum of fr over r≥1. The same quantity can be written S=M1−M2+M3−M4+… with alternating signs through order N.

For the original four-site example, S=48−30+12−2=28. Equivalently S=4ζ1−6ζ2+4ζ3−ζ4.

This is the finite inclusion–exclusion account for the union of the site lists. A widespread species contributes several times at lower orders, and the alternating combination counts leave exactly one net contribution to the union. It is another independent check on the reconstructed histogram. The result still concerns taxa recorded somewhere in the supplied sites; it does not add species absent from every recorded list or estimate a regional total outside the defined sampling frame.

Contents · Previous chapter · Next chapter

Now supply only ζ1=12, ζ2=5 and ζ3=3 for four sites. Let the unknown fourth-order value be t=f4. The equations give f3=12−4t, f2=−6+6t and f1=24−4t.

Nonnegative integer counts allow t=1,2 or 3. These yield three different histograms, all reproducing the first three zeta orders exactly. Their total recorded richnesses are 29,28 and 27 respectively.

The ambiguity exists with exact lower-order values. It is not created by observational noise. A complete inverse needs the missing information or an additional justified constraint. Choosing one value of t because its histogram looks smoother or more plausible adds a preference; it is not a consequence of the three supplied averages alone.

Contents · Previous chapter · Next chapter

The t=1 alternative has f=(20,0,8,1). Make it balanced by placing five unique species at each site, two species on each triple, and one species everywhere. Each site then has richness twelve, each pair shares five taxa, and each triple shares three.

The t=3 alternative has f=(12,12,0,3). Place three unique species at each site, two species on each pair and three everywhere. It has the same single-site, pair and triple intersection counts as the original balanced construction.

Only the four-site intersection separates these examples: it contains one, two or three species. Thus even the individual lower-order counts, not merely their averages, can agree across these three fictional constructions. The missing highest order carries information that cannot be recovered simply by calculating the same lower-order combinations more precisely.

Contents · Previous chapter · Next chapter

A full zeta spectrum identifies the occupancy-frequency histogram under the stated conditions. It does not generally identify which species occupy which sites. Species and site labels are lost when shared counts are averaged across all subsets.

For a simple counterexample, consider four species each recorded in exactly one of four sites. They could all occur at A, leaving the other sites empty, or one species could occur at each site. Both tables have f1=4 and all higher frequencies zero.

Their complete spectra are identical: ζ1=1 and ζ2=ζ3=ζ4=0. Yet the site-level richness pattern differs strongly. This is why histogram recovery should not be described as reconstruction of an ecological map. Retain the original incidence matrix and site identities whenever spatial arrangement or the identity of restricted taxa matters to the question.

Contents · Previous chapter · Next chapter

Let f0 denote species that would belong to a larger biological frame but are recorded at none of the N sites. For every positive order k, C(0,k)=0. Adding any number of such all-zero rows leaves the entire observed zeta spectrum unchanged.

The inverse therefore cannot determine f0. A total that includes those species needs additional information and an appropriate sampling or detection model. The recorded union S and a true regional species inventory are different quantities.

This limitation is fundamental, even with every order known exactly and no arithmetic error. It also prevents a misleading interpretation of the word complete: the spectrum is complete across orders for the supplied incidence table, not necessarily complete in biological coverage. State which completeness is meant whenever the reconstruction is reported.

Contents · Previous chapter · Next chapter

A species recorded at some sites but missed at others can be assigned to the wrong occupancy class. A taxon truly occurring in four sites might appear in f3 if one occurrence is not detected, or in a still lower class if more are missed.

The inverse exactly reconstructs the recorded table’s frequencies when its conditions hold. It cannot distinguish a genuine restricted distribution from one produced by missed occurrences using the zeta values alone.

A detection-adjusted analysis requires its own observation model, independent repeat information where appropriate, and uncertainty. Do not correct selected sites informally while applying the exact combinatorial inverse to the rest. The mathematical account becomes more useful when its observed-incidence target remains explicit, so that a reader can separate data reconstruction from a later inference about biological occupancy.

Contents · Previous chapter · Next chapter

Every fr reconstructed from a genuine finite incidence table must be nonnegative. For exact complete enumeration, the fr values are also integers. These constraints provide stronger tests than merely checking that zeta declines with order.

An exact four-site fourth-order count must itself be an integer, while lower-order means have combination denominators. A curve can therefore fail an exact-table check even when all its values look smooth and plausible. Use the inverse to test the complete contribution relationships rather than rely on the appearance of the curve.

Contents · Previous chapter · Next chapter

Take N=4 and ζ=(12,5,3,0). The values are nonnegative and non-increasing. The inverse gives f4=0 and f3=12. Then f2=30−3×12=−6, which is impossible as a species count.

This example demonstrates that monotonicity is necessary but insufficient. The higher-order intersections must also be compatible with the lower-order repeated-contribution totals. Too much triple sharing without any four-site sharing requires more pair contributions than the supplied ζ2 allows.

Use the inverse as an exact feasibility check when the input claims to be a fully enumerated finite incidence spectrum. If it fails, inspect definitions, averaging weights, rounding and data processing before assigning ecological meaning to the resulting curve. A negative occupancy count is not evidence of a novel class of species; it is a conflict in the supplied account.

Contents · Previous chapter · Next chapter

If all combinations of a finite observed table are enumerated, each Mk is an integer contribution total. Its zeta mean is that integer divided by C(N,k). A reported decimal may be a rounded representation of such a rational number.

If only some combinations are sampled, or if a model estimates zeta values, the reported numbers need not satisfy the exact integer relationships. Applying the inverse can then produce fractional or even negative estimated classes because of uncertainty or incompatible estimates.

That does not mean one should round every class independently and declare a recovered table. Any reconstruction method must retain the constraints and uncertainty appropriate to the estimation procedure. Report whether the result is an exact histogram, an estimate of one, or a fitted constrained summary. Those descriptions support different conclusions and require different checking.

Contents · Previous chapter · Next chapter

For four sites, f1=4ζ1−12ζ2+12ζ3−4ζ4. If each supplied zeta value has an absolute error at most0.01, the triangle inequality bounds the absolute error in f1 by 0.32.

The corresponding simple bounds are 0.24 for f2,0.08 for f3 and 0.01 for f4. These are conservative bounds under the stated separate error limits, not confidence intervals or claims about an actual field instrument.

If errors are related, their joint structure can make the feasible uncertainty narrower or different from these separate bounds. The calculation shows why alternating inverses can magnify uncertainty in lower occupancy classes. Carrying many decimal places through an inverse does not establish that the input spectrum supports equally precise recovered frequencies. State the origin of the zeta uncertainty and propagate it through the actual transformation.

Contents · Previous chapter · Next chapter

A practical calculation should retain the exact combination totals when they are available. Replacing a rational mean by a short decimal and multiplying back can create noninteger Mk values, which then affect several recovered classes.

For example, an order-two total of fourteen across six pairs has ζ2=14/6. Storing only 2.33 reconstructs13.98 instead of fourteen. In a four-site inverse, that discrepancy contributes to both f2 and f1 with different signs and coefficients.

The problem is not that decimal arithmetic is forbidden. It is that an exact count is available and should not be replaced prematurely by a less precise representation. Preserve integer totals or rational values through the checking stage, then round the reader-facing result appropriately. This is especially useful when a claimed discrepancy is small compared with the error introduced by formatting the intermediate values.

Contents · Previous chapter · Next chapter

The forward identity counts all k-site subsets with equal weight. If an analyst selects only nearby combinations, a species’ contribution depends not just on how many sites it occupies but on where those sites lie.

Two species with the same occupancy r can then contribute differently because one occupies a spatial cluster and the other occupies separated sites. Grouping them only by fr no longer supplies enough information for the same formula.

The restricted-combination analysis may be scientifically appropriate, but it is a different statistic. A generic inverse copied from the complete-combination setting would be answering an unearned question. Record the selection or weighting rule, and derive a forward relationship compatible with it before attempting reconstruction. The sampling design belongs to the mathematical definition, not merely to the methods paragraph after the calculation.

Contents · Previous chapter · Next chapter

This article uses unnormalised shared-species counts. A Jaccard similarity divides an intersection by a union, and other normalisations introduce different denominators. Averaging those ratios is not the same operation as averaging raw intersection counts and later normalising.

Therefore a vector of normalised similarity values cannot be substituted for ζk in the inverse without a new derivation. The resulting numbers could look numerically tidy while lacking the contribution-total interpretation that made the binomial identity work.

Before using software output, identify the exact response column, normalisation option and combination scheme. One useful hand check is to calculate a shared count for a small subset directly from the incidence table and compare it with the documented statistic. That concrete reconstruction can reveal a definition mismatch earlier than inspecting an entire smooth curve.

Contents · Previous chapter · Next chapter

If comparable surveys produce different reconstructed recorded-occupancy histograms, the difference describes the supplied records. It does not by itself establish whether land use, dispersal, climate, taxonomy or observation changes caused it.

For instance, higher f1 can indicate more taxa recorded at only one site, but that pattern can have several ecological and observation explanations. The identity of those taxa and the spatial context may matter more than the count alone for a conservation decision.

The inverse is useful because it makes a summary interpretable and checkable. It does not replace study design or conservation judgement. A strong report states the observed transformation first, then the independent evidence that supports any explanation of change. Retaining the original incidence table makes those later questions possible rather than hiding them behind a reconstructed histogram.

Contents · Previous chapter · Next chapter

Hui and McGeoch’s 2014 work introduced zeta diversity as a framework connecting multi-site incidence patterns, including occupancy-frequency distributions. McGeoch’s author-maintained explanation of zeta diversity describes the order definitions and links to that original paper.

The complete-combination counting argument and the fictional examples here are presented explicitly so the reader can audit the forward and inverse calculations. They are not field measurements or a newly claimed ecological discovery. The author page was checked on 25 October 2026.

Read the source for the broader ecological framework and applications. Use this article for the narrower reconstruction task: identifying which occupancy-frequency information is encoded by an exact full spectrum and which information remains absent. Software versions or methods mentioned on a source page should be checked separately before reproducing a computational workflow.

Contents · Previous chapter · Next chapter

Consider three fictional sites with nine site-restricted taxa in total, six taxa each occurring at exactly two sites, and two taxa present at all three sites. Compute the complete zeta spectrum.

Then hide the histogram and recover it from your calculated spectrum. Use both the downward recurrence and the direct three-site formulas. Calculate total recorded richness through inclusion–exclusion and compare it with the sum of the recovered frequencies.

For a further transfer check, keep only the first two zeta orders and let the unknown third order be t. Derive the family of compatible frequencies and its nonnegative integer range. Explain what remains unidentified and what an additional all-three-site intersection count would contribute. The changed number of sites requires new combination coefficients rather than reuse of the four-site formulas.

Contents · Previous chapter · Next chapter

The histogram is f1=9, f2=6 and f3=2. Single-site contributions total M1=9+12+6=27, so ζ1=9. Pair contributions total M2=6+3×2=12, giving ζ2=4. The third-order value is ζ3=2.

Invert downward: f3=2, f2=12−3×2=6, and f1=27−2×6−3×2=9. Total recorded richness is 17. Inclusion–exclusion gives 27−12+2=17 as an independent check.

With only the first two orders, f3=t, f2=12−3t and f1=3+3t. Nonnegative integer counts allow t from zero through four. Their totals are 15+t. Supplying ζ3=2 selects the original histogram and total. The lower-order values alone do not establish that choice.

Contents · Previous chapter · Next chapter

Begin with the named sites, the taxonomic incidence rules and the precise combination average. Retain every supplied order and its combination count. Multiply means back to contribution totals, reconstruct from the highest order downward, and check nonnegative integer classes when exact enumeration is claimed.

Regenerate the full spectrum from the recovered histogram. Independently compare its sum with the inclusion–exclusion total. Retain the original incidence table because the inverse does not recover species identities or their spatial arrangement.

Finally state the inferential boundary: recorded-incidence frequencies are not automatically true occupancy frequencies, all-zero taxa are invisible, and estimated or restricted-combination spectra need additional treatment. A compact checklist is useful when it makes these concrete dependencies reproducible rather than merely certifying that a program produced numbers.

Contents · Previous chapter · Next chapter

For N=4, nonnegative reconstructed classes require ζ4≥0, ζ3≥ζ4, ζ2−2ζ3+ζ4≥0, and ζ1−3ζ2+3ζ3−ζ4≥0. The corresponding scaled expressions must be integers when they describe an exactly enumerated finite table.

These are not fitted ecological restrictions. They express the nonnegative occupancy classes after applying the inverse. They explain why checking only the direction of a plotted curve is weaker than checking the whole contribution account.

Conversely, if the inverse produces nonnegative integer fr values, some incidence table can be constructed: create fr distinct taxa for each class r and assign each to any selected subset of r sites. Its full complete-combination spectrum will be the supplied one by the forward formula. Additional constraints, such as independently known richness at each named site, could restrict which assignments are possible and would need separate checking.

This sufficiency argument concerns existence of a table, not uniqueness of its labelled arrangement. It is useful because it distinguishes an impossible summary from a summary that is feasible but spatially ambiguous. A reconstructed histogram can pass every count constraint while still leaving the ecological location and identity questions unresolved.

Contents · Previous chapter · Next chapter

In the original example, choosing uniformly among the twenty-eight recorded taxa selects a ubiquitous taxon with probability 2/28=1/14. Choosing uniformly among the forty-eight recorded species-site occurrences selects one of the ubiquitous taxa with probability 8/48=1/6.

The difference is a change in sampling unit. A species occupying four sites appears four times in the occurrence list, while a site-restricted species appears only once. The occurrence-weighted occupancy-class probabilities are r fr/M1 rather than fr/S.

The mean occupancy of a uniformly selected recorded species is M1/S=48/28=12/7 sites. That is different from a site richness or a mean conditioned on a species-site occurrence. Every number is valid when its sampling unit is stated.

This distinction helps prevent a reconstructed histogram from being given an inappropriate probability interpretation. A report about a randomly selected taxon, a randomly selected occurrence and a taxon selected because it was shared by a pair are conditioning on different events. Higher-order sharing increasingly favours widespread occupancy classes because they can contribute to more combinations, not because the method has assigned them a greater intrinsic ecological value.

Contents · Previous chapter · Next chapter

At order k, select a species-sharing contribution from the complete set of k-site intersections. The probability that it belongs to occupancy class r is wk(r)=fr C(r,k)/Mk. These weights sum to one when Mk is positive.

For a species recorded at r sites and already shared by k selected sites, a uniformly selected additional site retains it with fraction(r−k)/(N−k). Averaging that fraction with the contribution weights gives ζk+1/ζk.

In the original four-site example at k=2, occupancy classes two, three and four have weights6/30,12/30 and 12/30. Their additional-site retention fractions are zero, one-half and one. The weighted result is 0.2×0+0.4×0.5+0.4×1=0.6, equal to ζ3/ζ2=3/5.

This derivation supplies a precise interpretation for the ratio under the complete-combination scheme. It is not the same as selecting a species uniformly from the full recorded union. At k=3 the corresponding weights are one-third for three-site taxa and two-thirds for ubiquitous taxa, so retention is two-thirds. If Mk is zero, the ratio is undefined; a numerical convention should not silently turn that absence of a denominator into a biological probability.

Contents · Previous chapter · Next chapter

Consider deleting one uniformly chosen site from the original four-site dataset. An r-site taxon retains occupancy r when the removed site was unoccupied, with probability (N−r)/N. A taxon in class r+1 drops into class r when an occupied site is removed, with probability (r+1)/N.

Thus the expected new class frequency is fnew[r]=fold[r](N−r)/N+fold[r+1](r+1)/N, where fold and fnew denote the original and reduced-frame frequencies. The notation describes counts moving between adjacent occupancy classes, not a new species type.

For the balanced original construction, any site removal gives three-site frequencies(15,6,3). Four site-restricted taxa disappear from the recorded union. Twelve remaining singletons and three pair taxa that lose an occupied site make the new fifteen singletons. The resulting union has twenty-four taxa, while the first three zeta orders remain(12,5,3).

The unchanged lower-order zeta values do not imply an unchanged occupancy-frequency table, because the number of sites has changed. Both the inverse coefficients and the meaning of an occupancy class depend on N. This controlled deletion provides a practical transfer check: always retain the site count and frame when comparing reconstructed histograms. A change in data inclusion can alter the summary without a change in any species’ original observed distribution.

Contents · Previous chapter · Next chapter

If exactly one recorded species occurs at every site, all zeta orders equal one. The inverse must recover fN=1 and every lower class zero. The repeated combination contributions cancel exactly in each lower reconstructed class.

If instead each of N sites has one unique species and there are no shared taxa, ζ1=1 and every higher order is zero. The inverse recovers f1=N and all higher classes zero. These two spectra have the same average richness but different recorded union sizes and occupancy histograms.

They are useful independent checking cases because their incidence meaning is obvious before any matrix calculation. A reconstruction method that exchanges them, loses the factor N at order one or returns a nonzero intermediate class has a definition or arithmetic error. Passing these simple cases supports numerical consistency without establishing that a more complex survey has complete detection or a valid combination design.

Contents · Previous chapter · Next chapter

Does a complete zeta spectrum recover every species location?

No. It recovers occupancy frequencies under the complete-combination assumptions. Species identities, site-level arrangement and a detailed ecological map are not generally determined by those averaged counts.

Can the inverse discover species missed everywhere?

No. All-zero rows contribute nothing to any positive zeta order. Estimating those species requires additional sampling or model information.

Is a decreasing zeta curve always compatible with a finite incidence table?

No. The reconstructed frequency classes must also be nonnegative and, for exact finite enumeration, integer-valued. The checked example(12,5,3,0) fails despite being non-increasing.

Can sampled combinations be treated as the exact full spectrum?

Not automatically. Their estimates carry uncertainty, and spatial or other selection can change the statistic itself. Match the forward model and inverse to the actual combination scheme.

Contents · Previous chapter · Next chapter

Return to the Zeta Diversity subject guide for incidence definitions, sharing orders, sampling and ecological interpretation. This article develops the separate inverse calculation from a complete spectrum to recorded occupancy frequencies.

Connect the reasoning through The Gold Standard of Education and the How X Works hub. A useful final explanation identifies the averaging design, reconstructs the repeated-count ledger, verifies the inverse and states which biological and spatial questions remain unanswered.

Contents · Previous chapter · Return to Zeta Diversity