ID of European Pyrgus (grizzled skippers)

[Roger Gibbons]

I am not sure how many Pyrgus enthusiasts frequent these forums, but for any that are, here is my first shot at the identification of Pyrgus species (http://www.butterfliesoffrance.com/Pyrg ... cation.htm), which I hope is reasonably navigable. This is the first draft, as I thought it better to get input before going further, mainly because I may well have made errors or missed vital clues. Any input would be very welcome - could you let me have any comments by mid-March please. I then plan to update these web pages and spend some time studying the Natural History Museum archives where I can view some 50 or so of each species to see if this corroborates the ID features outlined here.

Roger

[Padfield]

Brilliant! I think the research you're doing for this is fantastic, Roger. We all have our own little tricks, which we think work with in our own regions, or regions we regularly visit, but I for one have never even thought about undertaking a comprehensive survey like this, visiting museum collections. I will learn a lot when I get down to studying your charts.

I'd be interested to know how you did the calculations for the scoring. It shouldn't be too difficult to put this on a strictly mathematical footing (using Bayes' theorem) if you wanted that.

The requirements for that would be that you looked at a sufficient number of specimens of each species (say 100) and for each feature noted the percentage observed with and without that feature. I imagine you intended to do that anyway.

Applying conditional probability to determine the likelihood of each species, given the combination of features noted, would be complicated to do manually, but could easily be written into an Excel program - or even programmed into an application. It would be statistically more significant than a simple summing of scores (I believe).

I don't know how good your Excel skills are but I could easily help you do this, if that is the way you wanted to go. I imagine a tick-box system of features and options (including the 'don't know' option - essential if reliable figures are to be produced). When you've filled it all in a percentage probability automatically appears next to the name of each species.

I thought I'd put these thoughts on the forum, rather than send them privately, as someone else on the forums might have the skills to do something like programme an iPhone app. for use in the field.

Guy

[Roger Gibbons]

The requirements for that would be that you looked at a sufficient number of specimens of each species (say 100) and for each feature noted the percentage observed with and without that feature. I imagine you intended to do that anyway.

That's the plan, the first stage being to define/describe the number of options for each unh mark for each species. In some cases there seemed to be only one option, some had several, and some were so variable as to be of no use. I had hoped that the BC trip to the NHM on 11 Feb would have given that option, but 20 or so people went and the social aspects rather overtook the study considerations. I did get to spend 10 minutes of so on each species and this did appear to confirm the variants that I had been working on, so the next question is to undertake the counts on a statistically valid number of each species. It's going to be a fun day!

The one thing I did note on 11 Feb was the variation in Foulquier's Grizzled Skipper (Pyrgus foulquieri/bellieri) which leads me to wonder if there were not some mis-identifications in that tray.

Roger