Wildlife metrics

CensusIO computes the standard wildlife metrics from your survey data: herd composition from animal classification counts, a harvest summary from hunter and harvest records, sex and age composition from individual animals, mortality by cause from found-dead records, detection-corrected occupancy and abundance from surveys that revisit the same sites, and survival (nest success and known-fate) corrected for how long each subject was watched.

Herd composition (sex and age ratios)

A classification survey records, for each group seen, how many animals were males, females, and juveniles. CensusIO pools those counts for each species and reports:

• Males per 100 females: the sex ratio, for example bucks per 100 does or bulls per 100 cows.
• Juveniles per 100 females: the recruitment ratio, for example fawns per 100 does, a productivity index for the year.
• Percent juvenile: young of the year as a share of all classified animals, another read on recruitment.

The ratios pool every classified group recorded for the species, so more groups give a steadier estimate. A ratio is left blank when no females were classified, since it would otherwise divide by zero.

What CensusIO needs

A survey type with a species field and three count fields it can read as males, females, and juveniles. The seeded Aerial Ungulate Survey has these already (males classified, females classified, juveniles classified). On a custom survey type, name the count fields so the words are clear, for example “Males”, “Does”, and “Fawns”: CensusIO recognizes buck, bull, doe, cow, fawn, and calf alongside male, female, and juvenile.

Harvest summary

When a survey records hunter effort, or checks in harvested animals, CensusIO summarizes the season:

• Hunter effort and success: from a per-hunter survey (days hunted and animals harvested), the success rate (the share of hunters who took at least one animal), the mean animals per hunter, animals per hunter-day, and the share who saw game.
• Harvest by species, unit, and method: from per-animal check records, how many animals were taken of each species (with the mean field-dressed weight), from each management unit, and by each method.

A plain fish length-weight survey is not mistaken for a harvest record: a harvest record is recognized by a dressed weight, a management unit, or a harvest method.

Sex and age composition

Where herd composition reads group classification counts, sex and age composition reads individual records, one animal at a time, from harvest-check and mortality surveys (each animal has a recorded sex and age class). For each species it reports:

• The sex ratio (males per 100 females) and recruitment ratio (juveniles per 100 females), built from the individual animals. A fawn or calf counts as a juvenile whatever its sex, and an animal of unknown sex is left out of the sex ratio.
• The age-class distribution: how many animals fell in each age class (for example fawn, yearling, adult) and what share of the sample that is.

Because it pools harvested and dead animals, this is a different sample than a live classification survey, so read the two together rather than expecting them to match.

Mortality by cause

A mortality survey documents animals found dead. CensusIO summarizes them:

• Cause of death: how many animals died of each cause (predation, vehicle collision, disease, winter kill, and so on) and what share of the total that is.
• Cause by species: the same breakdown split by species, so you can see what each species is dying of.
• Carcass condition: the freshness of the carcasses (fresh, moderate, old, scavenged), a rough guide to time since death.

Occupancy and abundance, corrected for detection

A survey never sees every animal. A species can sit at a site you visited and go undetected, and a count is almost always lower than the true number present. So the plain share of sites where a species turned up (naive occupancy) and the plain counts both read low. When the same sites are visited more than once, those repeat visits let CensusIO separate two things and correct for the gap:

• Occupancy (detection-corrected), from detection / non-detection across repeat visits (MacKenzie et al. 2002). It estimates occupancy (psi), the share of sites the species really occupies, and detection (p), the chance of detecting it on a single visit where it is present. Naive occupancy is corrected upward by how often visits miss a species that is there.
• Abundance (detection-corrected), from repeated counts (the Royle 2004 N-mixture model). It estimates abundance per site (the mean number of animals at a site), the total across the surveyed sites, and the per-visit detection (p). The naive index (the average of the highest count at each site) is lifted to account for the animals each visit missed.

Both report a separate estimate for each species where the survey records one.

What CensusIO needs

A survey type with three things:

• a site field (a station, plot, transect, point, or camera) that names the place revisited,
• an occasion that tells the repeat visits apart (a visit, pass, or round number, or failing that the visit date), and
• a presence flag (for occupancy) or a count (for occupancy or abundance) recorded on each visit.

There must be at least two sites, and at least one site visited more than once.

Record every visit, including the empty ones

The correction only works if a row is recorded for every visit, including visits where the target was not seen (presence false, or count zero). If empty visits are simply left out, the data has no information about misses, and the estimate just mirrors the detections. The models also assume a site does not change occupancy across its repeat visits (it stays closed) and that detections are independent between visits.

Across seasons (colonization and extinction)

When the same sites are surveyed over several seasons (years, breeding seasons), each with its own repeat visits, the multi-season occupancy report estimates how occupancy changes (MacKenzie et al. 2003):

• Colonization: the chance an empty site becomes occupied by the next season.
• Extinction: the chance an occupied site goes empty by the next season.
• plus the initial occupancy and the per-visit detection, and the implied occupancy in each season.

It needs the same fields as single-season occupancy plus a season field (a year or season number) that tells the seasons apart from the within-season visits, with at least two seasons and two sites.

Survival

Whether a nest or an animal survives is, like detection, biased by how long it was watched. CensusIO has two standard corrections:

• Nest success (Mayfield): from each nest’s exposure (the nest-days it was under observation) and its fate, the daily survival rate divides the failures by the total nest-days, then raises that to the nesting-period length for whole-period survival. This corrects the apparent success rate (successful nests over all nests), which overcounts because nests found late had already survived their riskiest days. Record a nest-days field and a fate (an outcome such as fledged or depredated, or a “succeeded” flag); a nesting-period field adds the period survival.
• Known-fate survival (Kaplan-Meier): from interval counts of how many tracked (for example radio-tagged) animals were at risk and how many died, the cumulative survival is the product of each interval’s survival, with a Greenwood standard error. Record the number at risk and the number of deaths per interval, with an interval or date field to order them.

Where these show up

Herd composition, the harvest summary, sex and age composition, mortality by cause, occupancy, abundance, nest success, and known-fate survival are built-in reports in the reports hub, so they export to CSV, Excel, and PDF with the rest.

Related

• Fisheries metrics
• The estimators
• The reports hub