Hi. Are you using blast for the identification? Maybe the species you have are not represented in databases, in that case you will always have the same problem.

What is the length of the sequenced region? Often short regions do not allow identifications at species level.

A few years ago, we tested a set of primers for plants, maybe of some use to your work:

https://link.springer.com/article/10.1007/s12686-016-0605-0

Perhaps your taxonomic database was not specific enough for species classification. I would suggest that you try with the PLANiTS database for ITS2.

Article PLANiTS: a curated sequence reference dataset for plant ITS ...

Dear Mélodie,

I strongly recommend you take a look at the polymorphism in your group of taxa.

People have used, besides rbcl and ITS, the matK and trnH-psbA (Galimbert et al. 2014) genes as potential species-specific makers, and often more than one together. The trnH-psbA seems to be promising. However, that is a little bit different from animals where the COI has some clock-like passe of mutation across many taxa (it still challenging in cnidarians, though).

You might find out more about plant barcodes in Hollingswort et al. (2011) and Li et al. (2015). Especially for pollen identification, I've found some interesting reports and opinions in Bell et al. (2016), Galimbert et al. (2014) and Bruni et al. (2015).

My advice is to think about what criteria you are going to use for identification. Some of them are base on genetic distances (like K2P thresholds; ABGD), other on trees like the coalescent models (GMYC) and Poisson tree process (PTP) models, and some other on diagnostic characters (see review in Dellicour and Flot, 2018). The point is that all methods bring some bias and a consense of methods from different natures would lead to a more accurate delimitation.

If you are going to use a multi-locus approach, I would recommend the PHRAPL framework (See Jackson et al. 2017).

I hope you find this helpful.

Cheers

References:

Bell et al. 2016: Article Review and future prospects for DNA barcoding methods in for...

Bruni et al. (2015)

Article A DNA barcoding approach to identify plant species in multif...

Dellicour and Float 2018:Article The hitchhiker's guide to single‐locus species delimitation

Galimbert et al. (2014)

Article A DNA Barcoding Approach to Characterize Pollen Collected by Honeybees

Hollingsworth et al. (2011)

Article Choosing and using a Plant DNA Barcode

Li et al. (2015)

Article Plant DNA barcoding: From gene to genome

Jackson et al. (2017)

Article Species Delimitation with Gene Flow

Thanks a lot for all these references!

The markers used will be different if we want to detect plant consumed (short fragment, about 200-250 bp), but I presume if we target pollen we can look for longer fragments as DNA will not be digested.

We have to check database completness you're right, but my previous experience with a quiet full database was wrong identifications (sometimes matching species not present in the geographical area sampled), or lots of confusions and errors beyond family level.

No problems Mélodie Ollivier , but sorry, I completely missed the gut content part... If I got it right this time and you plan to use short fragments from partially degraded samples, perhaps it could be more interesting to use markers evolving faster than the regular genes, or maybe combining both. In animals (my homeland) we have the D-loop control region, which evolves faster than genes, still maternal lineaged, and it's somehow easy to use.

However, in your case, that I suppose you will use a high throughput sequencing, combining fragments might become very expensive. At the same time, to get a "silver bullet" that identifies plants at the species level with ~200 bp is difficult when you have several families to identify. You might also try to take a look at the Ion PGM system from ThermoFisher (~400bp), longer reads better chances.

There is a paper from Zhang et al. (2021; Plant-Herbivorous insect network) that had succeeded (apparently). I got not that deep into it but seems quite like what you are trying to.

This is a heck-challenging task, but I hope you get there.

You could find some help on these papers:

Article A DNA mini-barcode for land plants

Article Plant-herbivorous insect networks: who is eating what reveal...

Article DNA barcoding of fogged caterpillars in Peru: A novel approa...

Cheers