Wildlife data analysis · Alberta, Canada

What 650,000 harvested animals reveal about Alberta's big game

Three species — elk, white-tailed deer and moose — across ten hunting seasons. Modelling hunter pressure, species distribution, sex-selection bias, disease surveillance, and the gap between where animals are abundant and where hunters actually succeed.

Headline findings

Six results that are not obvious from reading the harvest tables directly.

Ten-season finding

Buck harvest is a constant; doe harvest is a dial

Male deer harvest sat between 22,000 and 30,000 every year for a decade with no trend. Meanwhile the antlerless share swung from 37% to 57% and back to 31%. Total harvest fell 40% from its 2023 peak almost entirely through that one lever.

Distribution

Elk and deer barely overlap

Across units holding both, harvest totals correlate weakly. WMU 350 produced 1,284 deer and 4 elk. WMU 522 produced 760 elk and 270 deer. Very few units are genuinely productive for both, so a unit's reputation for one species says little about the other.

Pressure

Big harvest ≠ good odds

Dividing harvest by success rate estimates hunter numbers. The highest-harvest units near Calgary are also the most crowded, and post below-median success. Volume measures crowds as much as animals.

Efficiency

Season length doesn't buy success

WMU 402 offers a 75-day general rifle season for bull elk and returns 7% success. WMU 312 offers 37 days and returns 26%. Access difficulty and terrain dominate opportunity length.

Same decline, opposite cause

Moose is the control case

Moose harvest fell 33% from its 2020 peak — but with both sexes down about 31% and a flat antlerless share all decade. Deer and elk fell because antlerless tags were cut; moose fell across the board. Sex composition is what tells the two apart.

Outliers

Two units break every rule

WMU 732 — CFB Suffield — posted 98% elk success on 320 animals, and WMU 728/730 — CFB Wainwright — posted 73% deer success. Both are controlled military-base draws: statistical artifacts of restricted access, not habitat quality.

Method and derived metrics

What the source data contains, what it does not, and how each derived figure is calculated.

Source structure

Each report gives, per Wildlife Management Unit: male, female and young counts, a total, and an estimated hunter success percentage. These are survey estimates, so small counts carry wide error bars.

Some rows are licence groupings rather than single units — 124/128/142/144/148/150, for example, where one special licence spans several WMUs. They count toward provincial totals but are excluded from per-unit joins, since they cannot be attributed to one geography.

Derived: estimated hunter count

hunters ≈ harvest ÷ (success% ÷ 100)

A first-order estimate that understates participation wherever one licence carries multiple tags.

Derived: species tilt

Elk share of combined harvest, elk ÷ (elk + deer), expressed as a percentage. Values above 50% indicate an elk-dominant unit; below 50%, deer-dominant. Computed only for units where both species recorded harvest.

What "antlerless" counts

Adult females plus young of either sex, not females alone. Alberta's antlerless licence authorises any animal without antlers of legal length, so a male calf qualifies. Since the young column is not split by sex, female-plus-young is the closest match to what the tag permits. Use the Female class for females alone.

Derived: sex ratio

Males per 100 females, (male ÷ female) × 100. A harvest ratio, not a population ratio — it reflects tag policy as much as herd composition. Units with zero females are undefined and excluded.

Statistics

Correlations are Pearson coefficients over units with non-zero harvest in both species. Medians use a stated harvest floor so near-empty units do not dominate. Everything is computed live in the browser from the embedded data.

Data coverage

Per unit: three species × three seasons (2023–2025), ~210,000 animals. Drives the dashboard, maps and every per-WMU figure. Provincial totals: three species × ten seasons (2016–2025), ~650,000 animals. Drives the trend section only — the older reports were parsed for totals, not per unit.

Every report shares one row shape (WMU, male, female, young, total, success%), so adding a season means dropping in one array; the controls, columns and charts rebuild from whatever is loaded. Prose figures outside the trends section are 2025. All nine datasets are reconciled against their published totals in the data integrity check.

A note on what success rate does and does not mean

The published rate blends general-licence and special-licence (draw) hunting. In units where cow elk draws are generous, overall elk success is pulled upward well above what a hunter carrying only a general antlered licence should expect. Comparisons in this analysis flag that distortion wherever the male share of harvest indicates it.

Province-wide picture

Three species, three entirely different harvest structures — and the sex composition is what separates them.

Harvest composition by sex
Share of total 2025 provincial harvest. Elk sits close to an even split, deer is strongly antlered-biased, and moose sits between the two.
Male Female Young

The composition gap is the single most informative feature of these datasets. Elk harvest arrives at a near-even sex split because wildlife managers engineer it that way: antlerless elk almost always requires a draw, and those draws are issued in the volumes needed to hit population targets. Deer harvest in 2025 was left far more to hunter preference, and hunter preference is overwhelmingly antlered. Moose sits in between, at roughly one antlerless animal in three, because antlerless moose has been tightly drawn for the entire decade.

The young-animal share reinforces this. Calves make up a meaningfully larger fraction of elk harvest than fawns do of deer harvest, which is what you expect when whole family groups are being culled under antlerless authorizations rather than individual animals being selected for antlers.

One caution on reading this section: it is a snapshot of 2025 alone. The ten-season view above shows that deer's antlered bias was far weaker two seasons earlier, so treat the deer bar as the current state of tag policy rather than a fixed characteristic of the species.

2026 outlook

A prediction model for the coming season, built to be checked by hand rather than trusted.

The model, in four steps

No 2026 harvest data exists yet — the season opens in September 2026 and the survey publishes around March 2027 — so this is a forecast, built on three seasons of per-unit data:

  1. Recency-weighted baseline — 2025 weighted 3, 2024 weighted 2, 2023 weighted 1.
  2. Fitted slope — ordinary least squares across the three seasons.
  3. Volatility — mean absolute year-over-year change as a share of the mean.
  4. Volatility-weighted shrinkage — the projection blends the trend with the most recent actual season, weighted by 1 − volatility. Stable units keep their trend; noisy units get pulled back toward what actually happened.

Step four earns its place. WMU 357 elk ran 636 → 1,375 → 625; trend alone projected 871, above every year but the spike. Shrinkage brings it to 663. The reported range is the projection ± half the volatility, so unstable units carry visibly wide intervals rather than a falsely precise number.

Species
Region
Confidence
Projected 2026 harvest — top 20 units
The hollow outlined bar is the actual 2025 harvest, for reference. The solid bar is the 2026 projection, coloured by confidence. Click any bar for that unit's season-by-season detail.
actual 2025 projected — high confidence medium low — treat as noise

Projected success is the recency-weighted average success rate, which assumes licence structure holds. That assumption is the model's biggest weakness — a single change to a unit's antlerless draw can move its harvest by more than this projection ever would.

What would actually change these numbers

The ten-season record shows antlerless allocation moving 20 percentage points inside three years. That single lever swamps everything in this model, so the draw booklet tells you more about 2026 than any trend line does. Trust the high-confidence rows — large, stable harvests on unchanged licence structure — and treat the rest as a starting point.

Species distribution: elk country and deer country

Testing whether a good elk unit is also a good deer unit. Mostly, it is not.

Elk harvest vs white-tailed deer harvest by WMU
Logarithmic on both axes. Points above the diagonal are elk-dominant units, points below are deer-dominant.
WMU WMU 312 equal harvest of both species
Species tilt — which animal actually comes out of each unit
Each bar shows how far a unit leans from an even split. Bars to the right are elk country; bars to the left are deer country. A bar reaching the far edge means essentially all of one species and none of the other. Units with at least 150 combined animals; hover for real counts.
elk-dominant → ← deer-dominant centre line = 50 / 50 · gold labels are the units named in the summary below

The Peace region and the Grande Prairie units form a contiguous block of elk-dominant country. The Rocky Mountain House belt is the mirror image — enormous deer harvests with almost no elk taken at all, despite sharing similar latitude and forest cover.

The practical implication for anyone planning a hunt: a unit's reputation for one species carries almost no information about the other. Units that are genuinely strong for both are rare enough to be identified individually rather than assumed.

Sex-ratio analysis

Males harvested per 100 females — a direct readout of tag policy and hunter selection.

Males per 100 females harvested, selected units
Logarithmic scale. 100 is an even split. Elk ratios swing wildly by unit because draws dictate them; deer ratios cluster high almost everywhere.
Elk (bulls per 100 cows) Deer (bucks per 100 does) Moose (bulls per 100 cows)

The management signal hidden in the ratios

Where elk ratios fall below 100, cow draws are doing heavy lifting to suppress growth. Where they climb into the hundreds, the unit is effectively antlered-only and the herd is being allowed to grow. Read the ratio and you can usually infer the tag structure without opening the draw booklet.

Deer ratios swing just as widely — 36 to 2,600 across the province in 2025 — but for a different reason. Elk spread reflects where cow draws exist; deer spread reflects how many antlerless tags were issued that year, a number the ten-season view shows moving sharply. Both are policy signals, but elk's is structural and deer's is a dial being turned.

Hunter pressure model

Inverting success rate to estimate participation, then testing whether crowded units perform.

Estimated hunters vs success rate
Bubble area is total harvest. If harvest volume simply tracked animal abundance, the busiest units would also be the most successful. They are not.
Distribution of success rates
Units with harvest above 50 animals, binned in 5-point intervals. Moose sits highest, deer in the middle, elk lowest — a draw-restricted species with few tags produces high success per hunter. The lone elk unit above 95% is a controlled military-base draw.
Elk Deer Moose

Regional breakdown

Alberta's WMU numbering encodes ecoregion: 100s prairie, 200s parkland, 300s foothills, 400s mountain, 500s boreal.

Total harvest by region and species
2025 season, single-unit rows only; multi-unit licence groupings excluded.
Elk Deer Moose
Aggregate success rate by region
Harvest-weighted: total regional harvest divided by total estimated regional hunters, rather than a mean of unit percentages.
Elk Deer Moose

Case study: the Calgary catchment

Applying the analysis to a practical question — which units within roughly two and a half hours of a major city actually reward the drive.

Deer success vs elk success, units near Calgary
Bubble area is combined harvest. The upper-right quadrant is what a multi-species hunter is looking for, and it is nearly empty.
Habitat and movement schematic — WMU 312
Illustrative geography of the highest-performing nearby unit. Not to survey scale; concentration areas are inferred from terrain, drainage and forest-edge structure rather than from telemetry.
Schematic habitat map of WMU 312 Forest reserve on the western edge, three inferred ungulate concentration areas along the forest fringe, and the road and river network around Sundre and Water Valley. Forest reserve — public land, bedding cover Bearberry valley Fallentimber corridor West of Water Valley Red Deer River Fallentimber Creek Hwy 22 Hwy 27 Hwy 584 Hwy 579 SundreWater Valley CremonaBergenBearberry dawn / dusk feeding movement → inferred concentration area forest reserve
The productive band in this unit is the interface between public forest cover to the west and agricultural feed to the east — animals bed in timber and move onto hay meadows and cutlines at first and last light. Harvest concentrates along that seam rather than deep in either land type.

Three-unit comparison

A foothills unit, a mountain unit and an urban-fringe unit — chosen because they fail and succeed for completely different reasons.

Best combined odds

WMU 312 — Sundre / Water Valley

Elk harvest292 @ 26%
Deer harvest648 @ 40%
Moose harvest21 @ 33%
Elk male share60%
General rifle, elk37 days
Land tenurelease + private
Limiting factoraccess permission
Unrestricted access

WMU 402 — Livingstone

Elk harvest56 @ 7%
Deer harvest94 @ 16%
Moose harvest5 @ 50%
Elk male share95%
General rifle, elk75 days
Land tenurepublic
Limiting factorterrain, elevation
Archery only

WMU 212 — Calgary fringe

Elk harvest135 @ 20%
Deer harvest107 @ 22%
Moose harvest57 @ 20%
Elk male share49%
General rifle, elknone
Land tenureprivate
Limiting factorpermission density
Normalised comparison
Success rates, male share of harvest, and estimated participation, side by side.

WMU 402's 7% is not comparable to WMU 312's 26%. Ninety-five percent of the elk taken in 402 were bulls, because antlerless there is draw-restricted — so that figure approximates a pure bull-hunting success rate. In 312, bulls were only 60% of harvest, meaning cow draws inflate the headline number. Adjusting roughly for that composition narrows the gap from nearly four-fold to closer to two-fold.

WMU 212 is the structural outlier. It is archery-only, which suppresses participation, yet returns 20% elk success — high for bowhunting. Its 49% male share confirms that antlerless elk are being taken freely on general licences rather than draws. The constraint is not animals or regulation but land: it is the ring of private property around a million-person city.

Season length is a weak predictor of outcome. 402 offers double the general rifle window of 312 and produces a fifth of the elk. Terrain, elevation and animal dispersion outweigh calendar access, which is a useful corrective to the common assumption that a longer season means a better hunt.

Regulatory context

Harvest data cannot be interpreted without knowing which tags were available. Selected 2026 season structure for the units discussed above.

WMUSpecies / class ArcheryGeneral Licence
302–314Elk, antlered 3-point+Sep 1 – Oct 24Oct 25 – Nov 30General
312, 314Elk, antlerlessSep 1 – Oct 24Oct 25 – Feb 28Draw
310–348White-tailed deer, antlered + antlerlessSep 1 – Oct 31Nov 1 – Nov 30General
400, 402Elk, antlered 3-point+Sep 1 – Sep 16Sep 17 – Nov 30General
400, 402Elk, antlerlessSep 1 – Sep 16Sep 17 – Nov 30Draw
400, 402White-tailed deer, antlerlessSep 1 – Sep 23Nov 1 – Nov 15General
212Elk, antlered + antlerlessSep 1 – Nov 30General
212White-tailed deer, antlered + antlerlessSep 1 – Nov 30General
214Elk, antlered 3-point+Sep 1 – Oct 24Oct 25 – Nov 30General
216, 221, 224Elk, antlered 3-point+Sep 1 – Oct 31Nov 1 – Nov 30General
404–408Elk, antlered 6-point+Sep 9 – Sep 16Sep 17 – Nov 30Draw

Why the 6-point rule matters to the data

Mountain units immediately west of the foothills require a six-point minimum on bull elk rather than three. That single regulatory difference explains a large share of the success-rate collapse between adjacent 300-series and 400-series units — the animals are present, but far fewer of them are legal.

Scout the units (WMUs)

Every unit, species and loaded season, filterable on every dimension. All figures, charts and the table recalculate live. Click any row or bar for a unit's season-by-season detail. Copy link encodes the entire view — filters, sort, open unit — into a URL you can send to someone; Export CSV downloads exactly what is on screen.

beta Ask in plain English — a question gets parsed into the filters below and answered from the data

Prototype: a rules-based parser, running entirely in your browser with no server. In production the parser swaps for a language model behind a small serverless endpoint — the data layer stays exactly as it is here.

Season
unit-level data: 2023–2025
Species
Class
Region
Top units
Click any bar to open that unit's multi-season detail. With several species selected, each bar is coloured by whichever species dominates that unit; red marks controlled-access military-base draws.
elk-dominant deer-dominant moose-dominant
Harvest vs success
Bubble area is estimated hunter count. Upper-left is the efficient quadrant — good odds without heavy participation.
Composition of selection
Male, female and young split across every unit currently matching.
Click a column to sort · click a row for unit detail

Hunter counts are derived as harvest ÷ success rate and rounded. Blank cells indicate no recorded harvest or an undefined ratio. Multi-unit licence groupings are hidden by default.

Zone maps — licence type and disease status

Alberta has no published geospatial layer for Wildlife Management Units on any open CDN, so rather than invent coordinates, the province is rendered here two ways: as a regional tile map covering every unit, and as a hand-built geographic schematic for the Calgary catchment.

Colour the tile map by

Every Wildlife Management Unit, grouped by ecoregion
Tiles are grouped by WMU series and sorted numerically — this is a schematic index, not a geographic projection. Hover any tile for detail; click one to filter the data explorer to that unit.
Geographic schematic — the Calgary catchment
Approximate relative positions of units within roughly three hours of Calgary, coloured by licence type for the selected species and labelled with that species' 2025 harvest and success. Dashed arcs are rough drive-time bands. Not to survey scale.
The pattern is structural: general-licence bull elk hunting forms a continuous north–south band through the foothills, flanked to the west by draw-only six-point mountain units and to the east by parkland units where the rifle season requires a special licence. A hunter unwilling to enter a draw is effectively confined to that middle band.

Reading the licence categories

  • General — buy over the counter, rifle season available.
  • General, archery only — no draw required, but no rifle season exists. WMUs 212, 247 and 248 are the notable elk examples.
  • Archery general, rifle by draw — the dominant parkland pattern. You may hunt with a bow on a general licence, but the November rifle season requires a special licence.
  • Draw only — no general opportunity at all; priority points required.

Why this matters to the harvest numbers

Licence structure explains most of the variance in the harvest data. Units where antlerless animals are draw-restricted show male-dominated harvests and lower headline success. Units with liberal antlerless general tags show balanced sex ratios and inflated success. Any comparison of raw success rates that ignores licence type is comparing different games.

Chronic wasting disease

A prion disease of deer, elk and moose, always fatal to the animal, spreading westward across Alberta since the first wild case in 2005. It reshapes both the harvest data and the practical decisions a hunter makes with a carcass.

Prevalence by species, 2025–26 surveillance season
Share of tested animals returning a positive result, with sample size. Mule deer carry the disease at roughly seven times the rate of white-tailed deer.
Prevalence by sex — deer species only
Males test positive at markedly higher rates in both deer species: consistent with larger home ranges, rut-season contact and longer average time at risk. This chart covers mule deer and white-tailed deer only, because Alberta publishes sex-specific tested counts for those two but not for elk or moose — and a prevalence rate needs a denominator. The elk and moose sex data that does exist is in the next chart.
Male Female
Sex split of positive cases — all four species
What can be shown for every species: how the confirmed positives divide by sex. Both deer species are strongly male-skewed, as expected. Elk and moose invert it — and for elk that is exactly what you would predict when the harvest feeding those samples is itself overwhelmingly female. Note the moose sample: all 164 moose tested province-wide came from CFB Wainwright, where sampling is mandatory on every harvested animal, and 7 were positive. That is a census of one military base, not a picture of Alberta's moose.
Male Female
Two seasons compared
Provincial prevalence fell from 16.2% to 9.7% between 2024–25 and 2025–26 — a sampling artifact, not recovery. Note that elk moves the opposite way, from 3.9% to 9.7%.

The most important caveat in the CWD dataset

Prevalence fell six points in one year, and that is not the disease receding. Testing more than doubled — 2,963 animals to 6,819 — and the added samples were overwhelmingly white-tailed deer from the western frontier: the species and geography least likely to test positive. The province states that prevalence continues to rise within units. The denominator changed, not the disease.

Where the frontier moved

Seven units recorded first cases in 2025–26: 222, 300, 316, 320, 322, 503 and 504. The previous season added 221, 247, 306, 505 and 508. Together those trace a front moving west and north — 316, 320 and 322 are foothills units, and 320 and 322 border 312 and 314, the best general-licence country in the province.

Species risk ranking

Alberta's weighted occurrence figures: male mule deer 1.00, female mule deer 0.40, male whitetail 0.30, female whitetail 0.10. A doe whitetail is the lowest-risk animal in the province; a buck mule deer the highest by an order of magnitude.

What this means in the field

  • Submission is mandatory in 48 units, including nearly the whole foothills band — 302 through 316, plus 320, 322, 334, 402 and 404. Hunting those units means planning for a head submission, not deciding afterward.
  • Do not shoot a deer in the head if you intend to submit it. The lymph nodes and obex are the test tissues and a head shot can destroy both.
  • Record precise kill location — GPS or section/township/range, not just the unit. The submission is rejected as surveillance data without it.
  • Results arrive by email through an AlbertaRELM account. Heads submitted after mid-November are generally not tested until January or February, which is a long time to leave meat in a freezer undecided.
  • Freezers are seasonal, roughly mid-October to mid-December, and were removed after the 2025 season in some areas — call the local office ahead rather than assuming.
  • Carcass disposal matters. Debone at the kill site where practical and leave the remainder there; if you transport it, the remains should be burned, buried or landfilled rather than scattered in a new area.

Human health

There are no known human cases of CWD. Health authorities nonetheless recommend not consuming meat from an animal that tests positive, and both Alberta and Health Canada advise waiting for a result before eating an animal taken in a surveillance area.

Mandatory deer head submission units
All 48 units where submission was mandatory in the most recent published programme. Note this list governs deer specifically — elk and moose heads are accepted from anywhere in the province but are not covered by the mandatory WMU list, which is why WMU 732 does not appear here despite Suffield elk being systematically sampled through the base's own controlled hunt. Switch the tile map above to "CWD status" to see these in regional context.

CFB Suffield — WMU 732 — and the elk signal worth watching

Suffield is a controlled-access base hunted by draw on fixed dates. In 2025 it produced 320 elk at 98% success, from 18 bulls against 239 cows and 63 calves — a population reduction executed through hunters, not recreational hunting.

The disease trend there is the part that stands out. Alberta describes the elk tested in both of the last two seasons as coming primarily from CFB Suffield, and across that sample elk prevalence went from 3.9% of 232 elk in 2024–25 to 9.7% of 308 elk in 2025–26 — nine positives to thirty. Elk prevalence roughly doubled in a single year while the overall provincial figure was falling, which is the reverse of the headline.

Twenty-two of the thirty positives were cows, matching a harvest that is itself overwhelmingly female, and the same season recorded the first positive elk ever detected at CFB Wainwright. Alberta does not publish those 30 positives broken out by unit, so the exact Suffield figure is not public — but the direction is not ambiguous.

A caution that applies to both bases: WMUs 728 and 730 (west and east CFB Wainwright) require CWD submission on every harvested animal, and Suffield's controlled draw achieves something close to it. So these units are near-censuses while the rest of the province is a voluntary sample. They will always look disproportionately represented in the positives, and that is a sampling property, not necessarily a disease property. What the two-year elk comparison shows is a rate rising within that same near-census — which is harder to explain away.

Where the harvest and disease data meet

Bucks were 69% of the 2025 deer harvest, and male whitetails test positive at roughly three times the female rate. So hunter preference happens to remove the highest-risk animals.

The ten-season view complicates it: antlerless deer were 57% of the harvest in 2022–23 before allocation was cut. The buck-heavy 2025 harvest is better targeted for disease control, but 2023 removed far more animals overall. Which effect dominates is not something the public data can settle.

Limitations

What this analysis cannot support.

  • Harvest is not abundance. These are counts of animals killed, mediated by tag allocation, access, terrain and hunter effort. A unit with low harvest may hold many animals that are simply hard to reach or largely illegal to shoot.
  • Success rates blend licence types. General and special-licence hunting are pooled, so units with generous antlerless draws show inflated overall success relative to what a general-licence hunter experiences.
  • Small samples are noisy. Several units report success rates above 30% on fewer than ten animals. These are survey estimates with wide confidence intervals and should not be ranked against units with hundreds of animals.
  • Derived hunter counts are approximations. The inversion assumes one animal per successful hunter, which understates participation wherever multi-tag licences are issued.
  • Controlled-access units distort comparisons. Military base draws and similar restricted hunts produce success rates that reflect exclusivity rather than habitat, and should be excluded from any ranking intended to guide unit selection.
  • Spatial inference is coarse. WMUs are large and internally heterogeneous. The habitat schematic in this analysis is interpretive: no sub-unit animal location data is published, so concentration areas are inferred from terrain and land cover rather than observed.
  • CWD prevalence is sampling-dependent. Provincial rates move with testing volume and species mix as much as with the disease itself, as the 2024–25 to 2025–26 comparison shows. Unit-level prevalence is not published in a form that can be joined to this harvest dataset.
  • Licence classification is simplified. Each unit is assigned one category per species based on the primary antlered season. Real regulations include sub-area boundaries, weapon-specific windows and multi-unit licences that a single label cannot capture. Always verify against the current guide.
  • Published sex columns do not always sum to the published total. Male, female and young are each estimated independently by the survey, so they can miss the reported total by a few dozen animals — elk 2025 is 36 short, moose 2025 is 10 short, deer 2025 is 5 over. Any percentage on this page built from the sex columns therefore carries a fraction of a point of imprecision. Unit row sums do reconcile to the published totals, which is what the integrity check verifies.
  • Some published success rates exceed 100%. Several multi-unit elk licence rows and the CFB Suffield draw report rates above 100% — 131% and 134% in the 2024 and 2023 elk reports. These are survey-estimation artifacts: the modelled harvest exceeds the modelled number of licence holders, which happens with small samples and multi-tag authorizations. The figures are reproduced exactly as published rather than silently capped, but they make the derived hunter counts for those rows meaningless.
  • The long series is provincial, not per-unit. Ten seasons of totals support the trend analysis, but unit-level detail exists here for 2023–2025 only. So statements about Alberta as a whole rest on ten years; statements about an individual WMU rest on three.
  • The 2026 outlook is a forecast built on three points per unit. Its central assumption — that licence structure holds — is exactly what the ten-season record shows changing by 20 percentage points inside three years. Treat low-confidence rows as noise.

Data integrity check

Every dataset here was transcribed from a published PDF, so each one is verified in the browser at load time rather than asked to be trusted.

For each species and season, the sum of all unit rows is compared against the "estimated total" line printed in the source report. This table is generated live from the embedded data — if a figure were mistyped, it would surface here rather than sitting silently inside a chart.