The accuracy of scat identification in distribution surveys: American mink, Neovison vison, in the northern highlands of Scotland

1. What they did

American mink were brought to Britain for fur farming in the 1930s, and escaped mink now live wild across most of the country. The far north of Scotland seemed to be an exception, but nobody knew whether mink were truly absent there or whether nobody had looked properly. Mink prey on seabirds and water voles, so the answer matters for conservation.

To find out, the team surveyed 147 sites (71 coastal and 76 inland) across the northern highlands of Scotland in spring 2008:

• Sign surveys. Three experienced surveyors searched a 2 km stretch of waterway at each site for mink scats and footprints. Because they expected few mink, they took a cautious approach and recorded every sign similar in size and shape to mink, noting how confident they were. Each site was classed as positive, “possible”, or negative for mink.
• Tracking rafts. As an independent check, they set 55 floating rafts with a tracking plate under a tunnel on six rivers and at nine coastal sites, and checked them for footprints twice.
• DNA testing. Every scat thought to be from mink was collected and sent to a laboratory, where the species was identified from its DNA.

They also searched five major conservation and wildlife journals for mammal survey papers from the previous 5 years, to see how often sign surveys were checked with DNA.

2. What they found

Field identification

• The surveyors found signs they were confident belonged to mink at 7 sites (4.8%), and “possible mink” signs at another 50 sites (34%).
• All but one of these signs were scats; only one footprint was found.

Tracking rafts

• No raft had any tracks at the first check at the end of May.
• In June, one raft had a single incomplete track that looked like a mustelid of mink size, but it was too incomplete to identify with confidence.

DNA results

• 75 scats from 46 of the 57 positive or possible sites were tested. DNA worked for 45 scats (60%), from 31 sites.
• None of the 45 scats came from mink. Most were from pine marten (47%) or fox (41%); the rest were otter (6%), polecat (3%) and stoat (3%).
• The same team had earlier surveyed mink in southern England, where pine martens are absent. There, DNA confirmed 100% of a random sample of 61 “mink” scats as mink. The authors think the difference is at least partly due to pine martens.
• They could not explain why fox scats, normally much larger than mink scats, were mistaken for mink in Scotland but not in England. They suggest foxes may use riverbanks differently there, diet may change how scats look, or the spring scats may have come from young foxes.
• They put the lower DNA success rate (60%) down most likely to old or weathered scats.

Other studies

• Of 50 recent mammal survey papers, 34 (68%) were based on sign surveys, but only 11 included DNA checks of species identity. (The paper gives 11 as 22%, which is 22% of all 50 studies, not of the 34 sign surveys.)
• Other studies have found similar errors: up to 30% of “pine marten” scats wrongly identified at normal densities and 97% of 30 in a low-density area, and up to 54% of “snow leopard” scats actually from red fox. Some studies found high accuracy, for example coyote scats identified alongside tracks in snow, and Indigenous people in Paraguay identifying felid and canid scats (though not the three small spotted cats).

Costs

• DNA testing cost 15 to 50 British pounds (GBP) per sample with sequencing. Real-time PCR could cut this to 2 to 3 GBP per sample for large studies, with cheaper equipment, but needs careful design and checking.
• The survey itself took about 70 person-days, with an average of 100 km of driving per day.

3. What they concluded

• There are very few, if any, American mink in the northern highlands of Scotland. If any are there, they are most likely lone individuals passing through, not a resident population.
• Without DNA, the survey would have suggested mink at at least 7 sites (4.8%), and possibly at up to 50 more (34%), spread across the region at low density. That could have led to costly, time-consuming and fruitless eradication programmes. Instead, efforts can focus on preventing mink spreading from the south.
• Even experienced surveyors may not identify mink scats reliably where other mustelids and small carnivores live.
• Where several similar species live together, scats can easily be misidentified and may not reliably show which species is present. The authors recommend that studies relying on scats test identifications with DNA, at least during a pilot stage or on a subsample, and then decide case by case how much more DNA work is needed.
• Other ways to check identifications include bile acids in scats (which wash out in rain), scent chemistry of scats, groomed hairs in scats, trained scat detection dogs (useful for screening scats to cut DNA costs), and camera traps.

4. How tracking was or was not used

Tracks and signs identification was the core of the study, and what it tested. Surveyors searched 2 km stretches of riverbank, lake edge and coast for scats and footprints. Footprints were identified as mustelid by their shape and track pattern and told apart from otter by size. Scats were identified as mink by size, shape and smell. On the tracking rafts, mink footprints were told apart from otter by size and from polecat or ferret with a published method. No trailing was reported.

5. Tracker or observer certification

Other experienced trackers. The three surveyors (one of them co-author Andrew Harrington) had several years’ experience of mink and otter surveys in the UK, Estonia and Patagonia. No certification is reported. Their scat identifications were tested by DNA in this study: none of the 45 scats that gave DNA was from mink. Two of them had earlier identified mink scats in southern England with 100% accuracy (61 scats checked by DNA). A trainee went along but did not identify signs.

6. Other technologies, skills or innovations

• Floating tracking rafts with a tracking plate beneath a tunnel at the water’s edge, which record footprints of mink passing through.
• DNA extraction from scats stored in alcohol, with one set of PCR primers that amplifies a section of the mitochondrial cytochrome b gene from stoat, weasel, polecat, mink, pine marten, otter and fox, followed by sequencing. One test can identify all the target species.
• Cost comparison of sequencing and real-time PCR.

7. Implications for conservation (from the paper)

• Management based on wrong field sign data can waste time and money and misdirect conservation work, for example eradication campaigns against animals that are not there.
• DNA verification is an underused, powerful tool for surveys that currently rely only on surveyor experience. It may not be needed where no similar species live in the area.
• The cost of verifying should be weighed against the cost of the survey itself, and the risk that the whole survey is wasted if identifications are wrong.