Dear Anton:

Quickly is not easy. But to give you an orientation on how to start, you should ask what type of environment you work in (marine or continental), what type of samples do you see (phytoplankton, phytobenthos, etc.), which groups dominate (centric or pennate), which geographical area you are working. Just there I could see if I have any course of action to follow. All the best.

Eugenia

The diatoms collected from a litoral of some Islands in Atlantic ocean. Benthic pennate forms are prevail.

But I'm mainly interested in common approach usefull for any bunch of diatoms.

Dear Anton:

Unfortunately, I don't know a manual with which to begin to address the determination of benthic pennate diatoms. You will have to resort to a less general bibliography. Round et al. (1990) could be used as a first step in recognized the genera.

All the best.

Eugenia

Dear Eugenia, thank you for the recomendations.

Dear Anton,

I have same problem ::)))

Dear Anton Lyakh,

As my colleagues mentioned, it is not easy to diagnose thousands of images that may belong to hundreds of species. Yes, in the beginning, you need to determine the type of environment from which the samples were collected, then choose the references that are concerned with the same intended environment.

To facilitate the task as much as possible .. Sort the pictures by genus and then by species, and I prefer that you diagnose each genus separately so that you have all the diagnostic information about before going to another genus and this ensures that you are not confused

The American Diatom site and the "A Beginner's Guide to Diatoms" book by Jacob John (2015), are important references for you now

My best wishes for success in your mission

Dear Maitham Al-Shaheen,

Thank you for the response. Unfortunatedly, sometimes it is very difficult to sort the pictures by genera. For doing that, it is necessary to know the diagnoses of all genera your see. So you need to preliminary study characters of all diatoms genera. This is not an easy task.

Dear Anton,

Similar to what others have suggested, like it or not, you have to learn how to determine the genus, or at higher taxonomy level like family or order, to sort the images for later in-depth and more detailed analysis. I'm not that familiar with benthic diatoms, but books like "Identifying Marine Phytoplankton by C R Tomas" contains a detailed morphological and morphometrical description of some common diatoms and dichotomy keys to their genus/species.

Another way is to use a computer program like automatic image analysis/identification, but it required a lot of computation power, specialized program, and good AI training. One of them is like in this article: https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-14-115

I also interested in that method, but haven't really go into that direction yet.

Hope it helps.

Cheers,

Arief R.

Dear Arief.

I think that the both ways are good. The second one is better that the first, because it can help to sort diatom images - to find similar images of diatoms and to put them into one group. But at present only human can finally identify them.

May be some one know existing software (like died Google Pcasa) which can help to group photos by similarity?

what About something like this: Article Quantitative comparison of taxa and taxon concepts in the di...

For the details you may ask Bánk Beszteri or Michael Kloster.

Hello there, I recommend you to get in touch with a taxonomist. Diatom ID is not easy. If you hurry up you will probably be wrong.

I don't think the strategy of taking thousands of photos is the most appropriate. If you only have general aspects, it is possible that your photos are insufficient to make reliable determinations. When one goes to the MEB with samples, one must be aware of the characters to consider in each genre that the sample contains. To do this, you should consult a taxonomist in the group to be able to carry out a taxa analysis plan. Interdisciplinary work seems to be the most appropriate mechanism to face the problem.

Diatoms of North America (www.diatoms.org) is a great place to start in addition to the Round et al., text. Also a good list of people to get in touch with for confirmation.

Dear Anton,

If the subject is diatom taxonomy; "speed is a disaster!" :)

All best

Fatih

Dear Anton,

As already suggested, it is either years of studying diatom literature or consulting diatom taxonomists. Chances are they will give you some more detailed advice and recommend the most knowledgable experts for the problematic groups. If you work on material from poorly explored habitat or geographic region, it is likely you will find some (many) yet undescribed taxa. If the diatom diversity in your samples is high, no software will deal with that (not yet).

Start with information on the environment, location, and age. Then consult the literature. I personally find it difficult to identify diatoms solely on SEM images.

Hi Anton Lyakh, Please check the 'Project ADIAC: Automatic Diatom Identification And Classification' (https://websites.rbge.org.uk/ADIAC/adiac.html). You may also search database by genera and species (https://websites.rbge.org.uk/ADIAC/db/Adiacgen.htm). Hope it helps.

Hi Anton, I tried Ankur's links and they seem promising if you know what taxa etc but I think you are after a reverse lookup - I had a look at Google Lens...it can already identify up to 90% of tree species it says...and its web page says it identified a dog breed (pretty specific it was too! A spitz or something, sounded exotic). I'll try to download it, but in the meantime, sympathies as I approached a diatom expert with photos, and they told me I needed to send images of frustules. The beautiful book that person published had some 10,000 or so categories and that was just for starters. So I went instead to my local professional imaging lab where we can image thousands of slides in very short timeframes, as you must be doing there. The lab head told me that the more images we have, the more we can potentially identify the same species from several angles/presentations/with overlaps and parts obscured or missing etc. So it seems all you are doing is on the right track from my novice viewpoint! I finally found a helpful resource on youtube, a diatom video created by experts in microscopy and diatom taxa with PhDs. For some reason, I cannot now pull up the resource again, but going to youtube and looking for diatom videos will surely bring up some contacts to some of these experts. :D though...to be frank i still have a number of unidentified species in my collection... ha :)