Indexing principles and a widely applicable paradigm for indexing animal populations

1. What they did

Wildlife managers often need to know whether a population is going up or down, or whether one area holds more animals than another. Counting every animal, or estimating true density by capture and recapture or line transects, is often costly and depends on assumptions that are hard to meet. A population index, such as the number of tracks on a plot, is a cheaper alternative that reflects how many animals there are without estimating the actual number.

The author set out what makes an index most useful, and then presented one general way to collect and analyse index data that many different methods can fit into:

• Stations. Observations are made at stations spread through the area of interest. A station can be a plot of soil for tracks and other signs, a tracking tile, a chew card, a point for counting animals, a bait station, a camera, a trap line or a stretch of road where scats are counted.
• Repeated days. Each station is checked on several occasions (usually several consecutive days) in each indexing session, with the same time between checks.
• Measurements. Each check produces a count or an amount (for example, the number of track intrusions by each species into a plot), not just “present” or “absent”.
• Analysis. The General Index (GI) is the average across stations for each day, then averaged across days. The author derived a formula for its variance that allows for stations near each other, and days close together, being related.

A worked example used dingo track intrusions on 50 tracking plots on dirt roads, checked over four consecutive days on a cattle station in south-west Queensland, Australia.

2. What they found

What makes a good index

• Practical: easy to use and understand, with little chance of confusing species, and little inconvenience to landowners.
• Sensitive: it must change when the population changes. Muskrat structures last a long time, so counting them is a poor way to detect short-term change, whereas well-designed visual counts may do better.
• Precise, with its own variance: for an index, precision matters more than accuracy. A method that is easy in the field encourages more observations and so more precision.
• Robust: it relies on as few assumptions as possible.
• Useful extras: it can monitor several species at once and describe where activity is concentrated.

Designing the stations

• Animals rarely spread out at random. Stations can be placed where animals are likely to travel; many canids, for example, use dirt roads, so tracking plots along roads can sample them well, but only if the roads run through the whole area.
• Use the same stations in repeated surveys, and stratify by habitat where habitats differ.
• Stations should be the same size and prepared the same way, and checked after the same length of time (for example, every 24 hours after the plot was prepared).
• Counts or areas carry more information than presence or absence, which can lead to wrong conclusions. Activity on tracking tiles or chew cards can be measured as area, for example by counting squares on a grid.
• Stations should not fill up completely (a tile totally covered in tracks), so that more activity can still be detected.
• Recording each station’s location allows an index of pervasiveness (IP), which shows whether active stations are clumped, random or evenly spread.

The worked example

The mean number of dingo track intrusions per plot was 0.94, 0.82, 1.30 and 0.82 on days 1 to 4, giving a GI of 0.97, with a standard error of 0.122 and a coefficient of variation of 0.126.

3. What they concluded

• Many observation methods, for mammals, birds, invertebrates and even leaf or seed fall, can fit this general format and gain a sound index with its own measure of precision.
• An index is not an estimate of numbers. Relating an index to another density estimate only shows that the two methods agree; turning an index into true numbers needs populations of known size.
• If a true population estimate is essential, it is better to put the resources into proper density estimation from the start.
• The way to test an index is by experiment: index a population, change it, and index it again, ideally with an untreated control area.
• An index must rise and fall with the population, but it is not safe to assume a straight-line relationship.
• Index values should not be compared between species, because different species have different home ranges, road use and travel rates, so they leave tracks at different rates even at the same population size.
• A changed method (for example, a chew card with a different bait smell) is a different index and cannot be compared with the old one.
• The variance parts (station to station, day to day) can guide how many stations and days to use in future surveys.

4. How tracking was or was not used

Track counts are a central example. Tracking plots on dirt roads, where the number of times each species’ tracks enter the plot is counted each day, and tracking tiles or plates, where the area covered by each species’ tracks is measured, are among the main station types. The worked example counted dingo track intrusions into 50 plots over four days. Scat counts on road segments for canids are listed as another station type. No trailing was involved.

5. Tracker or observer certification

Not reported. The paper does not say who read the tracks in the worked example. The index depends on correctly telling each species’ tracks apart; the author notes that a good method should leave little chance of confusing species.

6. Other technologies, skills or innovations

• A general index with a variance formula that does not assume stations or days are independent, using variance components calculated with SAS PROC VARCOMP.
• An index of pervasiveness, using distances between active stations, to describe the spatial pattern of activity. Station locations can come from global positioning devices, maps or relative measurements.
• Measuring the area of activity on tiles or chew cards with computer software or a grid.

7. Implications for conservation (from the paper)

• Indices can monitor populations efficiently where density estimation is impractical, for purposes such as harvest management, biodiversity assessment, threatened species, disease surveillance and wildlife in conflict with people.
• The same tracking plots used to index predators of marine turtle nests could also have indexed crabs, which also prey on the nests.