# A footprint technique to identify white rhino Ceratotherium simum at individual and species levels

> Can a rhino be recognised by their footprints alone? Working with 40 known white rhino in Namibia, researchers photographed 1276 footprints and turned each into 113 measurements. Their Footprint Identification Technique matched tracks to the right animal more than 90% of the time and told white from black rhino with over 98% accuracy, offering a cheap, non-invasive way to count and monitor endangered animals.

Summary by Kersey Lawrence, Original Wisdom. Updated 9 November 2026. Page: https://www.originalwisdom.com/tracking-literature/2008-alibhai-jewell-footprint-technique-identify-white/

## Citation and access

- **Authors:** Sky K. Alibhai, Zoe C. Jewell and Peter R. Law
- **Year published:** 2008
- **Journal or source:** Endangered Species Research
- **Full citation:** Alibhai, S. K., Jewell, Z. C. and Law, P. R. 2008. A footprint technique to identify white rhino Ceratotherium simum at individual and species levels. Endangered Species Research 4: 205–218.
- **DOI:** 10.3354/esr00067
- **Access status:** Hosted (freely available elsewhere)
- **PDF:** https://www.originalwisdom.com/wp-content/uploads/2026/11/2008_Alibhai_and_Jewell_etal_a-footprint-technique-to-identify-white-rhino-at-individual-and-species-levels.pdf

## Tags

- **Species:** Mammals > Black rhino; Mammals > White rhino
- **Region:** Africa > Namibia; Africa > Zimbabwe
- **Tracking:** Individual identification; Spoor age estimation; Track measurements; Track photographs; Tracking; Tracks; Tracks and signs identification; Trailing
- **Methods and technology:** Direct observation; Footprint Identification Technology (FIT); Global Positioning System (GPS); Population estimation
- **Data type:** Existing data; New field data
- **Observer qualification:** Not reported
- **Discipline:** Wildlife tracking
- **Year published:** 2008

## Detailed keywords

- **Species:** Studied: white rhino (southern white rhino, Ceratotherium simum simum), 40 known individuals (20 males, 20 females, aged from about 3 months to over 10 years). Comparison species: black rhino (Diceros bicornis), 18 individuals from an earlier study. Other species discussed from the literature: tiger (Bengal tiger), lion, cheetah, leopard, mountain lion, snow leopard, fisher (Martes pennanti), lowland and Baird’s tapirs, whales and sharks.
- **Location and habitat:** Otjiwa Game Ranch, central Namibia: a fenced area of about 100 km², mean annual rainfall around 500 mm, with sandy substrate in open terrain and game trails through acacia woodland; 28 waterholes. Black rhino footprints came from the Sinamatella Intensive Protection Zone, Hwange National Park, Zimbabwe. The Namutoni region of Etosha National Park, Namibia, is mentioned as an example of sampling design.
- **Topics:** Footprint Identification Technique (FIT); individual identification from footprints; telling white from black rhino by their footprints; censusing (estimating the number of animals in an unknown population) and monitoring (assigning new tracks to known animals); non-invasive alternatives to radio collars and immobilisation; geometric profile of a footprint from 13 landmark points, 13 derived points and 113 measurements; footprint library of known animals; track size (footprints per track); waterhole and road sampling; aging footprints for freshness; backtracking; effects of rain and wind; footprints changing over time; the tiger “pugmark” census debate in India; local trackers and anti-poaching units.
- **Statistics or models:** Discriminant analysis with forward stepwise variable selection in JMP (SAS) software; canonical variates (Fisher’s approach) with 95% confidence ellipses; Canonical Pairwise Comparison Technique (CPCT) for censusing, using 12 measurements; Canonical Ellipse Reduction Technique (CERT) for monitoring, using 30 measurements; linear discriminant analysis (LDA) with 30 measurements for species discrimination; leave-one-out (jackknife) and 50% training versus 50% testing validation. Inconsistencies in the paper: the abstract gives monitoring accuracies of 97 and 99%, but the Results give 97% and 93%. The reported accuracies appear to count only misclassified tracks, not unclassified (tied) ones; for example, CERT misclassified 5 and left 1 unclassified of 159 tracks and is reported as 97%. For species discrimination, 1 misclassified track out of 118 is reported as 99.5% (it works out to about 99.2%). The journal header gives “Published online January 18, 2007”, but the volume is 2008 and the paper was submitted in August 2007, so 2007 looks like a typing error.
- **Data type:** New field data: white rhino footprints collected at Otjiwa in 1999, 2002 and 2003 (1276 footprints, 159 tracks, 40 animals). Existing data: 360 footprints in 59 tracks from 18 black rhino, from the authors’ earlier study in Zimbabwe (Jewell et al. 2001).

## Synopsis

### 1. What they did

Many kinds of wildlife research need individual animals to be recognised. Radio collars and immobilisation have costs for the animals (earlier work on black rhino linked repeated immobilisation to lower female fertility, and many collars failed), and natural markings need the animal to be seen again. Tracks are easier to find than the animals themselves. Traditional trackers are known for reading information from tracks, and the authors set out to capture some of that information in an objective, repeatable way.

They refined their earlier method for black rhino and called it the Footprint Identification Technique (FIT). It has two stages:

- Turning a footprint into data. Fresh, undistorted left hind footprints were photographed with a scale. In Adobe Photoshop each image was rotated to a standard position and 13 landmark points were placed on the outline of the three toes and the heel. Thirteen more points were constructed from these, and the JMP software produced 113 measurements (lengths and angles), the footprint’s “geometric profile”.

- Analysing the data. A stepwise procedure picked the measurements that best told footprints apart, and canonical (discriminant) analysis compared tracks. They worked with whole tracks (a series of footprints from one animal), not single footprints, because one footprint can easily be an outlier.

They tested FIT on 40 white rhino of known identity on Otjiwa Game Ranch, a fenced ranch of about 100 km² in central Namibia. The animals were also identified by sight, independently of FIT, mainly by a small anti-poaching unit. In 1999, 2002 and 2003 each animal was tracked and its footprints photographed. To find footprints, the team checked all 28 waterholes (at first at least every 3 days) and drove every road at least once a week. Fresh tracks were followed to the rhino, or backtracked from a point where an animal had been seen stepping, so that only the target animal’s prints were collected. Each track’s location was recorded with the Global Positioning System (GPS), and each footprint was wiped out after it was photographed.

They tested three uses:

- Censusing (working out how many animals are in an unknown population) with a method called the Canonical Pairwise Comparison Technique (CPCT): every pair of tracks is compared, and two tracks are judged to be from the same rhino if their 95% confidence ellipses overlap.

- Monitoring (matching a new track to one of the known animals) with a new method, the Canonical Ellipse Reduction Technique (CERT): animals whose ellipses do not overlap the test track are removed step by step until only one remains.

- Telling white from black rhino, using 360 black rhino footprints from their earlier study in Zimbabwe.

Each use was tested in two or three ways, either leaving out the test track and building the analysis from the rest, or building it from half of each animal’s footprints and testing on the other half.

### 2. What they found

The footprint library

- 1276 footprints in 159 tracks from 40 white rhino (20 males and 20 females), with 3 to 12 footprints per track.

- Black and white rhino footprints share the same landmark points, so the same 113 measurements could be used for both species.

Censusing (CPCT)

- Test 1: of 11,329 pairwise comparisons, 104 were wrong (99.1% correct). Grouping the matches gave an estimate of 42 rhino for the 40 in the library (census accuracy 95%).

- Test 2: of 2080 comparisons from 64 tracks of 35 rhino, 15 were wrong (99.3% correct). The estimate was 32 rhino for 35 (census accuracy 91%).

- Accuracy depended on track size. Tracks of 6 to 8 footprints gave 95% or better, and 7 footprints was best. Short tracks gave large ellipses that overlapped too often (which would undercount animals), and long tracks gave small ellipses that overlapped too little (which would overcount). Tracks were split or joined to give 152 tracks of 6 to 8 footprints.

- Using 12 measurements gave the best results.

Monitoring (CERT)

- With each of the 159 tracks tested in turn, 5 were misclassified and 1 left unclassified (97% accuracy reported).

- With half of each animal’s footprints used to build the analysis, 3 of 69 test tracks were misclassified and 2 left unclassified (93%).

- CERT was less sensitive to track size than CPCT. Accuracy rose from 85% with about 15 measurements to 92% with 30.

White versus black rhino

- Of 218 tracks (159 white rhino, 59 black rhino), 1 was misclassified and 3 unclassified (99.5% reported).

- When all of the test animal’s other tracks were also left out, 4 were misclassified and 4 unclassified (98.2%).

- With a half-and-half split, 1 of 118 test tracks was misclassified (99.5% reported).

How many animals and footprints are needed

- About 15 to 20 known animals, and 15 to 20 footprints per animal, seem enough to build the analysis. Accuracy for single footprints levelled off at 20 footprints per animal.

### 3. What they concluded

- Footprints can identify individual white rhino very accurately, both for counting an unknown population and for monitoring known animals, and can tell white from black rhino with over 98% accuracy.

- With 40 animals, 159 tracks and 1276 footprints, this is a larger and therefore more robust test than most earlier footprint studies.

- The tiger “pugmark” census in India failed partly because the method lacked objectivity and used poor sampling, not because footprints carry no useful information. Footprints extracted objectively can identify individuals accurately.

- Strengths of FIT: it is cheap, needs no expensive equipment or vets, uses the skills of local trackers (often already employed as scouts or anti-poaching patrols), is arguably less subjective than methods that rely on visual judgement, is not limited by costly equipment, and is non-invasive.

- Limitations: heavy rain delayed footprint collection by 2 to 3 days; footprints change over time (young animals grow, and adult footprints seem to change too), so monitoring is best done as a continuous routine; and FIT needs a suitable substrate and a species with a complex enough foot. It is not suitable for every species or environment.

- The authors are adapting FIT for the Bengal tiger, lowland and Baird’s tapirs, and for harder substrates such as snow and mud.

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

Tracking was central to the study. Each rhino was tracked to collect its footprints. Fresh tracks found at waterholes and on roads were either recognised as a known animal’s or followed until the rhino was found. Where several rhino were together, the team backtracked from a point where the target animal had been seen to place its foot, using only unbroken tracks. Footprints were aged so that only fresh ones were collected. The paper says that collecting data for FIT is intuitive for skilled trackers, so local expertise can be used.

### 5. Tracker or observer certification

Not reported. The paper does not say who did the tracking and footprint photography, or describe their skills. Monitoring of the population (identifying the animals by sight) was done mainly by a small anti-poaching unit, helped by visiting students and research teams.

### 6. Other technologies, skills or innovations

- Digital photographs of footprints with a scale, processed in Adobe Photoshop.

- Landmark points placed by hand with cross-hair guides; derived points and 113 measurements generated by a script in JMP Statistical Discovery Software (SAS).

- Two new or refined analysis methods: CPCT for censusing and CERT for monitoring.

- GPS to record the location of every track.

- Sampling designed around rhino drinking: the authors assumed each rhino would visit a waterhole at least every other day before the rainy season.

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

- FIT is a cost-effective, non-invasive tool for censusing and monitoring endangered species, especially nocturnal or elusive ones, and avoids the costs to animals of capture and collaring.

- Because footprint images can be collected during routine work such as anti-poaching patrols, FIT could be a first-line monitoring tool or an extra objective tool alongside methods such as mark-recapture and camera trapping.

- It may give more realistic coverage of an animal’s home range than standard (non-GPS) telemetry or camera trapping.

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

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

- The method depends on good tracking at every step: finding fresh tracks, aging them, and following or backtracking to be sure every footprint belongs to the right animal. The statistics only work if the tracker’s identifications in the field are sound.

- The paper does not describe who did the tracking or test their skills, so readers cannot judge how much the results owe to the trackers’ experience.

## Cite this summary

Lawrence, K. 2026. Summary of: A footprint technique to identify white rhino Ceratotherium simum at individual and species levels (Alibhai and Jewell et al., 2008). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/2008-alibhai-jewell-footprint-technique-identify-white/