Using mink tracks

Reading the lines - foretelling the mink cull

Key findings

• Mink tracks recorded on the GWCT Mink Raft are perfectly detailed, suggesting the possibility of discerning males from females, or even individuals.
• Sadly, we were unable to do this with any reliability.

Skilled trackers in hunting cultures, as every schoolchild knows, can not only distinguish between individual big game animals by their footprints, but also assess the condition of the animal. To do this, they make use of any unique characters (eg. injuries or deformities), but often have to recognise a subtle combination of shape, size and other features, while mentally filtering out the effects of weather and substrate condition on the track itself. Wildlife biologists obliged to work on species without decades of apprenticeship in the field have occasionally tried to achieve the same prowess by detailed measurement of tracks, followed by statistical analysis.

Although mink tracks are small compared with lion tracks, those gathered on mink rafts are superbly detailed. It would be extremely valuable to know from initial raft checks how many mink were available to trap at a series of rafts, or at least which were males and which females. Natalie Fisher, an MSc student from Bangor University, undertook to try this for us.

We took two approaches. First, captured mink were made to run across a track pan. Numerous measurements were taken from digital images of the tracks, and the combinations of these that best distinguished between the tracks of different individuals were derived by statistical analysis. Second, all mink tracks made at rafts before trapping were passed through computer software designed to spot natural groupings and to classify data accordingly. This is described as 'unsupervised learning' because no examples or prior knowledge are used to guide classification: the software doesn't need to know how many groupings there are or what they look like.

Although the tracks of known individuals differed in an identifiable way, there was considerable overlap, so tracks could be ascribed to a particular individual with only low confidence. Even though male mink are generally much larger than females, there was sufficient overlap in foot size and shape that no track could confidently be ascribed to either sex. The unsupervised learning software threw up a number of groupings that matched the number of individuals caught, but on close inspection this turned out to be coincidence. Both approaches may improve as we acquire more material for analysis, but the whole process was extremely laborious. At best, it may prove a useful research tool, but it clearly isn't suited to general use in mink trapping practice.

Figure 1. Mink tracks are small, but when they are this clear measurements are easy to take. Potentially they might allow us to estimate the number of mink present before trapping.