# Using Sign Counts to Quantify Animal Abundance

> How do you turn a count of droppings or tracks into a real number of animals? This 1991 paper sets out a statistical method: mark some animals so their sign can be recognised, then count marked and unmarked sign on sample plots. A worked jack rabbit example shows how much marking and searching a precise estimate needs, and why every sign must be read correctly.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/1991-skalski-sign-counts-quantify-animal/

## Citation and access

- **Authors:** John R. Skalski
- **Year published:** 1991
- **Journal or source:** The Journal of Wildlife Management
- **Full citation:** Skalski, J. R. 1991. Using sign counts to quantify animal abundance. The Journal of Wildlife Management 55(4): 705–715.
- **DOI:** 10.2307/3809523
- **Access status:** Publisher link only
- **Where to find it:** https://doi.org/10.2307/3809523

## Tags

- **Species:** Mammals > Black-tailed jackrabbit
- **Tracking:** Individual identification; Scat; Sign surveys; Track counts; Tracking
- **Methods and technology:** Capture-mark-recapture; Chi-square test; Population estimation
- **Data type:** Modelling or theoretical
- **Observer qualification:** Not needed
- **Discipline:** Wildlife tracking
- **Year published:** 1991

## Detailed keywords

- **Species:** Worked example (hypothetical, not real field data): black-tailed jack rabbit (Lepus californicus), counting their pellets. Earlier sign-marking and sign-count studies mentioned (no scientific names given): ungulates and deer (pellet-group counts), mule deer, raccoon (radioisotope tagging), coyote (scat marking), varying hares (toe clipping for track identification), black bears and other carnivores, small mammals (dyes).
- **Location and habitat:** No study area. The example survey is a made-up study site divided into 160 subplots; a second figure shows a site split into four strata on either side of a roadway. Ravines, roadways and the banks of waterways are given as examples of areas animals may favour. The author was at the Center for Quantitative Science, University of Washington, Seattle, Washington, United States.
- **Topics:** animal sign as an index of abundance (tracks, excavations, scats); calibrating sign counts with sign production rates; pellet-group counts; sign marking (marking animals so their sign can be recognised): radioisotope tracers, scat-marking implants or injections, internal dyes, toe clipping for track identification, radio telemetry; two-phase survey design (mark animals, then search plots for sign); clearing old sign, then recounting sign left during the survey; closed populations; abundance and density estimation; survey design and sampling effort; sample size planning; uniform mixing versus clumping of marked sign; why sign-marking studies should not be analysed as mark-recapture studies; choosing which sign to count when comparing sites or treatments.
- **Statistics or models:** Finite sampling theory (nonparametric, no assumptions about the distribution of counts); an abundance estimator similar in form to the Lincoln Index, but with different assumptions; variance estimators for simple random sampling and stratified random sampling of plots, including the variation that comes from which animals happen to be marked; Taylor series approximations; sample size formulas based on the normal distribution; chi-square test of homogeneity for uniform mixing of marked and unmarked sign; density estimator (the “naive” density estimator of White et al. 1982). Inconsistency: the abstract says the method is for closed populations, while the introduction says it permits estimation “in closed and open populations”; the assumptions section requires a closed population, and notes that for density estimation only the marked animals need to be closed.
- **Data type:** Modelling or theoretical. A statistical method illustrated with a hypothetical (“mock”) survey; no new field data. Received 11 April 1987, accepted 12 December 1989, published October 1991.

## Synopsis

### 1. What they did

Animal sign, such as tracks, digging and scats, is often the only practical way to measure how many animals live in an area, especially for elusive species. Counts of sign give an index, which can be turned into a real number of animals only if you know how much sign each animal leaves. The author sets out a statistical framework for doing this with “sign marking”: catching some animals and marking them so that the sign they leave afterwards can be told apart from the sign of unmarked animals.

The survey he describes has two phases:

- Marking. A random sample of animals is caught, marked so that the mark passes into their sign (for example, a tracer that shows up in their scat), and released. To work out how reliable the estimate is, each marked animal’s sign must be identifiable to that individual.

- Searching plots. A random sample of plots is chosen from the study area. The plots are cleared of old sign at the start, then searched again at the end to count all the new sign, marked and unmarked.

He developed formulas for the number of animals, and for how uncertain that number is, for two ways of choosing plots: simple random sampling and stratified random sampling (for example, separate strata for different habitats or terrain). He also gave formulas for planning how many animals to mark and how many plots to search to reach a chosen level of precision, and showed how the method also gives density.

To illustrate, he used a made-up example: 10 black-tailed jack rabbits injected with unique radioisotope tracers, and 16 of 160 subplots cleared of pellets and searched again a week later.

### 2. What they found

The worked example

- The 16 subplots held 910 pellets, of which 298 came from the 10 marked jack rabbits.

- The estimate was about 31 jack rabbits (30.54), with a standard error of 13.87.

- The 90% confidence interval ran from 10 (the number known to be marked) to 53 animals.

- The coefficient of variation was 45.4%, too imprecise for a reliable estimate. More marking or more plot searching would be needed.

Planning sample sizes (with 10 marked animals)

- To be within 50% of the true number 80% of the time: 24 subplots.

- Within 40%: 44 subplots. Within 35%: 70 subplots.

- Within 25%: 947 subplots, far more than the 160 that exist. Even searching the whole site would not be enough; marking 15 animals and searching 57 subplots would reach this precision.

Clumped sign

- The marked pellets were not spread evenly across the site but clustered (a chi-square test rejected even mixing). In that case the simpler planning formula, which assumes even mixing, underestimates the effort needed, and the full formula must be used.

Assumptions

The method assumes that: (1) no animals move in or out during the survey (deaths are allowed if they are not caused by marking); (2) every animal is equally likely to be caught; (3) marking does not change how much sign an animal leaves; (4) the plot sample gives unbiased totals of all sign and marked sign; (5) the mark lasts and keeps showing in the sign throughout the survey; and (6) all sign is correctly identified as marked or unmarked, and as left during the survey.

### 3. What they concluded

- Earlier sign-marking studies were often analysed as if they were mark-recapture studies, or left out important sources of error. Their confidence intervals ignored the variation that comes from which animals happened to be marked, and so understated the true uncertainty.

- Because the method rests on sampling theory rather than assumptions about how the counts are distributed, it should hold up under varying environmental and population conditions.

- Both the number of animals marked and the number of plots searched strongly affect how precise a sign-marking survey is.

- Where the animals are closed to movement in and out, estimating density and estimating numbers are essentially the same thing in sign-marking studies. For density, only the marked animals need to stay on the site.

- When comparing sites or treatments, a change in conditions may change animal behaviour as well as numbers. For example, changing cover might change movement and diet, so track counts might change while pellet counts may or may not. A careful choice of which sign to count reduces the risk of biased comparisons.

- Without tests of assumptions and valid estimates of error, management decisions based on sign counts could be wrong.

### 4. How tracking was or was not used

No tracking was done: the paper is a statistical method with a hypothetical example. The method is built on sign surveys, searching plots and counting sign. The worked example counts pellets (scat). The paper also describes track counts that identify marked animals from the tracks of toe-clipped individuals, and marking scent posts or dwellings with dyes. A key assumption is that all sign is correctly identified as marked or unmarked.

### 5. Tracker or observer certification

Not reported. Because the example is hypothetical, no observers took part. The method assumes that every sign is correctly identified, but the paper does not discuss who searches the plots or how their skill would be checked.

### 6. Other technologies, skills or innovations

- Sign-marking techniques: radioisotope tracers injected into animals so their pellets are marked, scat-marking implants or injections, internal dyes that pass through the digestive system, toe clipping so individual tracks can be recognised, and radio telemetry.

- Recording sign in a table of subplots by marked individual, which gives the counts needed for the variance formulas.

### 7. Implications for conservation (from the paper)

- Calibrated sign counts can give real numbers or densities of animals, not just an index.

- By modifying index techniques, improving the reliability of sign counts, and including error estimates when weighing risk, management decisions made under uncertainty can be improved.

### 8. Implications for tracking (Original Wisdom's observations)

_These are Original Wisdom's observations, not the authors'._

- Every estimate from this method rests on assumption 6: that each sign is correctly identified and correctly dated as left during the survey. That is a tracks and signs identification task, so the skill of whoever searches the plots directly affects the numbers.

- The worked example shows how much effort a precise count can take: for some levels of precision, searching every plot is still not enough unless more animals are marked.

## Cite this summary

Lawrence, K. 2026. Summary of: Using Sign Counts to Quantify Animal Abundance (Skalski, 1991). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/1991-skalski-sign-counts-quantify-animal/