Using Patterns in Track-Plate Footprints to Identify Individual Fishers

1. What they did

Counting rare forest mammals is hard. Live-trapping is invasive, and genetic samples from hair or scat can be expensive and error prone. Track plates (sooted aluminium plates inside a covered box, with sticky contact paper at the baited end) have long been used to show which species are present. The authors asked whether the prints of fishers on track plates are detailed enough to tell individual animals apart, the way human fingerprints are.

Fisher prints on track plates show rows of tiny dots, made by papillae (small bumps) on the metacarpal pad of the foot. The authors:

• Worked out which foot made each print, using the position on the plate, the angle of the toes, and the shape and dot pattern of the pad. Only prints from the same foot (for example, left front with left front) were compared. Rear feet were identified but not used.
• Built a matching method. Prints were scanned at 2,400 dots per inch (or photographed) and compared on screen in ImageJ software by switching quickly between two aligned images and looking for the same dot patterns. A slight rotation between images helped matching dots stand out.
• Measured how unique the patterns are. They made ink prints from the front feet of museum specimens (14 feet from 10 fishers) and measured 1,400 spaces between neighbouring dots. From this they built a model of the chance that a print from a different foot would match by chance alone.
• Checked how consistent prints from the same foot are, using 24 dot pairs on several prints from 3 feet of 2 wild fishers.
• Tested people. Nine biologists, 4 with previous practice, took a test of 8 sets of three field prints (4 good quality, 4 fair quality), each with two prints known to match and one known not to.
• Applied the method in the field. At each of 10 sites in the Adirondacks, 6 baited, covered track plates were set 500 m apart along a trail or unpaved road for 12 days and checked every 3 days, in the summers of 2000 to 2002.

2. What they found

How unique the dot patterns are

• About 35% of dot spaces fell in the most common size range (0.22 to 0.28 mm). So the chance of a chance match is at most 0.35 multiplied by itself once for each dot pair matched.
• Matching 5 dot pairs gives at most a 0.005 chance of a false match; matching 10 pairs gives about 0.00003.
• Spacing of one dot pair barely changed between prints of the same foot (a standard deviation of 0.01 to 0.03 mm), roughly ten times less than the variation between different dot pairs.
• Spacings of neighbouring dots were only weakly related, so the model’s assumption of independence held well enough.

How well people matched prints

• Good-quality prints: 34 of 36 answers correct (94%).
• Fair-quality prints: 24 of 36 correct (67%).
• The 4 examiners with more experience missed, at most, only the most difficult print.
• Prints were harder to match when distorted, most often when the animal put less weight on the foot, so the dots were squeezed together along the length of the print.

Field trials

• The 10 transects gave 62 fisher detections, and 53 (85%) had at least one print good enough to match.
• Only 1 or 2 fishers were responsible for all visits at each site, even where there were 14 or 15 detections. Only 2 of the 10 transects had a second fisher, and in both the second one visited only once.
• With practice, confirming that two prints did not match usually took under 5 minutes.
• Automated human fingerprint systems could rule out the most different prints but could not link the dots, because they are built to match ridge endings and branches, and fisher feet have no true ridges.

3. What they concluded

• Dot patterns in fisher footprints are unique to individuals and are reproduced consistently on track plates, so individual fishers can be recognized from track-plate prints.
• Partial prints, common in field work, are still useful, because the chance of a false match drops quickly as more dots are matched. Given fishers’ territories and home ranges, the authors expect about 10 fishers could leave prints at nearby stations, so matching even 10 dots should be enough to conclude that one fisher made both prints.
• Detections from nearby track plates are not independent: one animal made most of the visits along a transect. Traditional short track-plate surveys therefore cannot measure abundance, but new designs using print matching could give noninvasive mark-recapture density estimates.
• Accuracy depends on print quality and on the examiner. Some people have more aptitude for pattern matching than others, so the authors say examiners need training and an evaluation of their matching skill. They offer an online tutorial.
• Dry conditions seemed to give better prints. Longer surveys and scent lures would probably detect more individuals.
• Other species with similar pad structures may also be identifiable this way: weasels and American marten have similar patterns but very small feet; striped skunk, Virginia opossum, raccoon and porcupine have larger, heavier feet that leave darker prints.
• Prints should be stored carefully so the paper does not wrinkle or bubble, and ideally photographed in the field at a fixed scale first.

4. How tracking was or was not used

The whole study is about reading footprints. Examiners identified fisher prints on track plates, decided which foot made each print (left or right, front or rear) from toe angle, pad shape and dot patterns, judged print quality, and matched prints to individual animals from the fine dot patterns of the front foot pad. No trailing was involved.

5. Tracker or observer certification

Reported, and partly tested within the study. No tracker certification is mentioned. The matching was done by biologists. Nine biologists took a matching test (4 had previous practice; most had no experience beyond the online tutorial), and 3 of the authors were among those who took it. The authors stress that experience with prints of known origin helps judge non-matches, and that technicians should be trained and have their matching aptitude evaluated before using the method.

6. Other technologies, skills or innovations

• Covered, baited track plates: aluminium plates sooted with a kerosene lantern made from a paint can, with contact paper at the baited end to collect prints.
• Desktop scanning at 2,400 dots per inch (measuring to 0.01 mm), or a 4-megapixel digital camera; grayscale images were adequate.
• Free ImageJ software for aligning, magnifying and “flicking” between two images to find matching dots.
• A probabilistic uniqueness model borrowed from poroscopy in human fingerprint forensics.
• Tests of automated fingerprint identification systems, including those of the New York State Division of Criminal Justice Services; the authors suggest software designed for dots rather than ridges might help narrow the search.

7. Implications for conservation (from the paper)

• The method offers a new tool for estimating fisher density, and perhaps that of other species, over large areas with less labour.
• At a minimum, the number of different prints gives the minimum number of animals; repeated collections could feed mark-recapture estimates.
• Advantages: pad patterns are relatively permanent (unlike ear tags), the method causes little stress to the animals, track boxes do not need daily checks, and the plates also record other carnivores and small mammals.