Factors affecting the (in)accuracy of mammalian mesocarnivore scat identification in South-western Europe

1. What they did

Much of what we know about carnivores comes from their scats, and scats are usually identified by eye. In south-western Europe, the red fox, stone marten and European wildcat are similar in size, live side by side, and can leave similar-looking scats. The researchers set out to measure how often scats identified by their appearance were wrong, and what made mistakes more likely.

• Collecting scats. In two protected areas, one in Portugal and one in Spain, they laid out ten 3 km transects on unpaved roads and trails in each area. Trained field technicians walked each transect twice per season, about 20 days apart, in summer/autumn and in winter/spring. They collected all carnivore scats within 2 m on each side of the line.
• Identifying by eye. The technicians identified each scat by its location, shape, size, colour and smell, using field guides. Scats judged to be over a month old, or whose species was doubtful, were left out.
• Checking with DNA. A laboratory team, who did not know the field identifications, identified the species from DNA in each scat, using two genetic markers. The DNA result was taken as the correct answer.
• Looking for causes. They modelled the chance that a field identification was right, using season, study area, the species the scat was called in the field, and how common each carnivore and its prey were, measured with camera traps.

They also reviewed 35 scat-based studies of European mesocarnivores (medium-sized carnivores) from 13 countries to see how many had checked their identifications.

2. What they found

DNA results

• 320 scats called red fox, stone marten or wildcat in the field were sent for DNA testing. 251 (78.4%) gave a species identification.
• 244 of these were from one of the three target species, though not always the one named in the field. Seven came from other species: two polecats and five dogs.

How often the field identification was right

• Called red fox: 86.3% correct (101 of 117). The wrong ones were stone marten (9), dog (5) and wildcat (2).
• Called stone marten: 77.8% correct (84 of 108). The wrong ones were red fox (22) and polecat (2).
• Called wildcat: only 11.5% correct. Most (84.6%) were red fox.
• Error rates therefore ranged from about 14% for scats called red fox to about 88% for scats called wildcat.

What affected accuracy

• Accuracy was higher in summer/autumn than in winter/spring, and higher in the Portuguese park than in the Spanish one.
• Scats called red fox were the most likely to be right; scats called wildcat were the least likely.
• Accuracy was higher where red foxes were more common on camera traps, and lower where wildcats were more common.
• How even the carnivore community was, and how common the prey were, made very little difference.
• The models explained only about 25% of the variation, so other important factors were not measured.

The literature review

Of the 35 European scat studies, only 8.5% had used DNA to check their identifications, and only about a third had taken precautions such as collecting only at well-known sites used by the target species.

3. What they concluded

• Errors in identifying mesocarnivore scats by eye are common, and depend on biological, environmental and human factors.
• Mistakes are worse for scarce species, and where species with similar scats are about equally common. The authors suggest this may be why wildcat scats were so often misidentified: where rabbits are plentiful, wildcats are more common and their diet overlaps more with the red fox’s.
• Seasonal diet may play a part: in winter, food is less varied, diets overlap more, and scats may look more alike. Differences in food between the two parks (rabbits are plentiful in the Portuguese park and nearly absent in the Spanish one) may also matter, but local diet data were not available to test this.
• The technicians’ prior knowledge of which carnivores live in each area might unconsciously bias their judgement.
• Error rates are specific to each situation and should not be transferred to other studies. Studies based on scats should measure their own error rates, with DNA or other methods, at least on a subsample, and take those rates into account in their analyses. Studies already based on scat appearance alone should be reviewed for possible bias.
• The same problems likely apply to other groups of carnivores with similar scats.

4. How tracking was or was not used

Tracks and signs identification of scats was the subject of the study. Field technicians searched transects for carnivore signs on foot and identified scats by location, shape, size, colour and smell, with the help of field guides (Bang, Dahlstrom and Mears 2007; Iglesias and España 2010). No trailing was reported.

5. Tracker or observer certification

The scats were identified by “trained field technicians”. No certification, training details or experience levels are given, but their identifications were tested against DNA in this study, and that test is the paper’s main result. The four people thanked for help with fieldwork are named, but the paper does not say which of them identified the scats.

6. Other technologies, skills or innovations

• DNA extracted from scats and identified with two markers: a nuclear gene fragment (interphotoreceptor retinoid-binding protein, IRBP) and a mitochondrial one (the control region). The mitochondrial marker worked far more often (97.6%) than the nuclear one (34.3%); both worked together in about a third (31.1%) of samples.
• Blind testing: the laboratory staff did not know the field identifications.
• Camera traps to measure how common each carnivore and prey species was.

7. Implications for conservation (from the paper)

• Errors in scat identification can bias ecological studies. Because the errors are worse for rare species, they can give unrealistic estimates of the status and distribution of threatened species, with serious consequences for their management.
• Scat surveys are often the only affordable way to survey carnivores over large areas, so they should not be discarded, but their biases should be acknowledged and accounted for.
• Restricting scat collection to well-known sites used only by the target species is one way to reduce bias.