RAUK - Archived Forum - Sand lizard markings

Do the different Sand lizard colonies have slightly different markings ?

I only have a small sample set at the moment, but I can see similarities between the Hankley 'Sandies' markings and a different trend with the Frensham ones.

Or am I being hoplessley subjective ?

Right, now have a pretty decent selection of pics of Surrey Heath Sandies, and have even knocked up a format for recording the markings to distinguish individuals.

What I now need is a system for characterising the markings, which saves me from checking each old photo against a new pic. Obviously location can be a fliter, but I need something which represents the markings, maybe numerically.

This MUST have been done with such things as fingerprints before pattern-matching on modern PCs. Before I re-invent the wheel and create my own system, does anybody know of an extant system in use ? would make far more sense for me to adopt a system than create a new one !

Any help out there ??

Thanks for that David,

What I am thinking of is developing a system which cues on characteristics of the marking pattern, not exact representation. (Or adopting one if one exists).

One example may be; on the back patternation, where there are elements of black banding which connect unbroken between both dorsal stripes, this could be represented as a boolean 1 or 0.

There MAY be a way of codifying some aspects of patternation. I shall rely heavily on the expertise of this forum to suggest possible characteristics.

Once we have a set of ideas, I can then just apply to the pictures I have to determine feasibility. If it looks hopeful, I'll liaise with Alan here in Surrey, and then maybe expand to see if such a system works with Dorset specimens.

Very interested to hear how the Newt characterisation worked out.

Thanks for the interest !

Cheers David,

My German is passable, or failing that I'm still in contact with my Austrian Ex. who happens to be a technical translator :P either way I'm sure it would be useful. The email link on my signature should work.

I'll keep this thread updated on any progress.

Got a half-decent pic of what I would consider a typical Surrey male sandy, how do the markings compare with Dorset specimens ?

Further to Chris GO's post please see my reply to the (now nearly a year old) topic re races. Although it sets out general principles re the differences between races (and they are pronounced enough so that most of us with experience can identify the race by appearance alone about 70% of the time) I have to agree with Chris re the difficulty in applying a universal rule. It can be no more than guidance. As I said in the post under the topic mentioned:-

"Having said this - every variation can be seen in every location. it would appear that the variations occur naturally and a combination of trace minerals in the habitat and natural selection (i.e. which pattern provides the best camouflage) are the determinants as much as the racial genes."

I was chatting with Emma Hutchings (sp?) from Marwell PT the other day, she mentioned they have somebody looking at the automated recognition of an individual L.a from a scanned pic, they are adapting the method used for Zebras (stripes). Maybe Chris G-O's student knows more?.

Assuming they are using a neural-net based system (that's a guess by me), the weightings may give some indication of racial differences. (just a thought).

When captive L.as are reintroduced...is stock from the appropriate race used ?

That does sound interesting. Certainly Sand Lizard markings seem to be as individual as fingerprints and effective as an identification at this level. I am not so sure at race level because all variations can be seen in all races at some time. I doubt if it could improve much on the "expert's" identification.

Regarding re-introductions of La, yes, where it is clear what race would have been found in a given area then only the appropriate race is used. For example, Welsh re-intros being thus far restricted to the North & West use the Merseyside race as there is historical contiguity of suitable largely dune habitat. Further South there are areas where it is difficult to decide whether Surrey or Dorset animals are more appropriate but these two races are far more similar to each other anyway.

Steve,

Did you ever come up with a numerical system for keeping up with individuals in reference to your first post in this thread?

It seems the concept is much along the lines of the methods used by Benson for adder, see

http://www.herpetofauna.co.uk/forum/forum_posts.asp?TID=1365 &PN=1&TPN=1

The idea being that using various features such as head marking, scalation and stripe variation you can code an individual adder. It should then be easier to check previous records to see if this animal has been encountered before, using partial or complete information, rather than checking potentially hundreds of images.

I've got a working example

BW140206_001_M_MB_P_5152142_FZZMMWTO

which translates to

site/date first seen_arbitrary field number_sex_base colour (mid brown)_scalation (P = parietal, due to a striking feature)_5152142, relates to scale group counts_FZZMMWTO is after Benson and is a way to describe the stripe, in this case it translates to fore-body zigzag mid-body mirror wave tail ovals

underscores are to keep it windows friendly,

In theory you could maybe get the FZZMMWTO bit from a fleeting glance, then get a high degree of confidence in the identification from a photographic image scale count.

Heh, yeah I got some way forward in assigning marking 'types', such as number of dorsal white dot clusters (3x2, 4x1 etc), continuous dark banding etc.

But, then had a chat with the folks at Marwell. They are looking at adapting their zebra automated recognition software to sand lizards.

Basically their system allows you to submit a pic of a zebra, and through various transforms, you get a set of unique values, which can be translated into an individual ID. (As I understand it). Tho I'm unsure how the lizard is isolated from the context.

When they get this working for Sandies, it will be far more effective than my approach, so I've stopped. Next time I see them, I'll chase progress.

Currently, I use photo-fit cards, like the one I've posted under individual adder section.