Second paper published on whisker spot patterns!

We showed you some promising results last year after we tested if whisker spot patterns could be used to identify individual Australian sea lions in captivity. We were able to show that the sea lions appeared to have different whisker spot patterns. When we matched the photographs of captive individuals with our software, 90% were actually correctly matched when the photographs were taken at 90° angle (side shot). This success rate, however, decreased considerably when the angle was changed, and also when the photographs were taken at during different sessions, i.e. weeks or months apart. We were suspecting that the problem with taking photographs during different sessions was actually angle problems and, with rearranging the sea lion and the photographer, the photographs were just not taken from the same angle anymore.

img_6440

In this second paper (which is available online), we tested the use of whisker spot patterns as an identification tool in wild Australian sea lions. We used photographs from different individuals from Carnac and Seal Islands, haul-out islands in the Perth metropolitan area, and from the Abrolhos Islands, islands off Geraldton also used for breeding, to increase the sample size. In total, we had good-quality photographs of 20 wild Australian sea lions to create a sketch of the pattern generally occurring amongst these individuals. Australian sea lions have on average 39 whisker spots, also called mystacial spots, (ranging from 30 to 46 in these 20 individuals), which were arranged in seven rows and 9 to 10 columns. Some individuals also had 1 to 2 whiskers above their eye, also called superciliary whiskers.

sl_paper

We were also able to show the locations where whisker spots most likely occurred in wild individuals and at which locations it was often quite uncertain to determine if a whisker spot occurred or not. Unfortunately, the locations where most whisker spots occurred within a pattern were also those which had the highest probability of occurring (i.e. most sea lions had these particular spots). This was not only shown by the main author of the paper, but also when five other observers tried to mark all the whisker spots in some preselected photographs. The different observers varied considerably in selecting the different spots although they were given the same photographs. We first compared these differences using the previously used technique from our first paper, the Chamfer distance transform algorithm, which had successfully been used in polar bears. The variation between observers was so high that for most photographs, the software outcome would suggest it to be a different individual, although they were the same, identical photographs.

sealion_whiskers

In lions, whisker spot patterns have successfully been used to identify individuals in many studies, using a slightly different technique. The technique of comparing the whisker spots patterns of different lions follows a row-column method (which is described on this webpage), which we then tested as well. One row with a relatively stable number of whisker spots occurring serves as a reference row and the whisker spots from a different row, with variable numbers of spots occurring, can then be located according to the reference row. The whisker spots from the more variable row can be drawn into a grid in relation to the reference row which can then be used to compare whisker spot patterns. The problem with Australian sea lions was, however, that we could not determine such a variable row in which whisker spots were also consistently selected by the observers.

The colour difference, i.e. the levels of red, green and blue colours, between the whisker spot and the surrounding fur impacted the probability of a whisker spot being marked significantly. The higher the contrast between the whisker spot and surrounding fur colour, more observers marked the spot and vice versa. This means, unfortunately, that we were not able to use the sea lions’ whisker spots to identify them individually successfully. We are hoping though that we can find another way to tell them apart in the future.

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s