Senecio stapeliformis seeds...

[MatDz]

Colin, we've all already "paid" for the Dec issue of the journal, you're not jeopardizing the sales numbers by sharing a little "teaser" of the content (although I must say I'm already "bought" by your mention of the Euphorbia story in it!).

Are you covering the Madagascar ones as well?

[ralphrmartin]

.... What I mean briefly by this is that all the components of Euphorbia have a common origin.

If you go back far enough, all plants have a common origin, but that does not mean they should all be in one genus.

Surely the concept of a genus should
- be based on a rather close relationship
- include few enough plants that it is a useful concept.

What I mean by the last point is something like this:
- we could put all species of cacti into one genus, but that would not be very helpful
- we could put each species of cactus into a different genus, but that would not be very helpful either

There has to be some optimum average relatedness to determine a genus, leading to say roughly as many genera as species, for the concept of genus to be maximally informative. (I'm not saying all genera should be the same size, just that there is a most useful average size).

Even if Euphorbia are all closely related, putting 2000+ species into one genus does not seem like a very informative thing to do.

[Colin Walker]

Colin, we've all already "paid" for the Dec issue of the journal, you're not jeopardizing the sales numbers by sharing a little "teaser" of the content (although I must say I'm already "bought" by your mention of the Euphorbia story in it!).

Are you covering the Madagascar ones as well?

Mat, what I'm doing is reviewing the latest research on Euphorbia and one aspect is a discussion of the concept of a unified view of the genus based on state of the art technology, i.e. whole genome sequencing. So in answer to your question, yes, Madagascan spp. are covered, but so are the other 1,800 spp. in the genus.

[Colin Walker]

.... What I mean briefly by this is that all the components of Euphorbia have a common origin.

If you go back far enough, all plants have a common origin, but that does not mean they should all be in one genus.

Surely the concept of a genus should
- be based on a rather close relationship
- include few enough plants that it is a useful concept.

What I mean by the last point is something like this:
- we could put all species of cacti into one genus, but that would not be very helpful
- we could put each species of cactus into a different genus, but that would not be very helpful either

There has to be some optimum average relatedness to determine a genus, leading to say roughly as many genera as species, for the concept of genus to be maximally informative. (I'm not saying all genera should be the same size, just that there is a most useful average size).

Even if Euphorbia are all closely related, putting 2000+ species into one genus does not seem like a very informative thing to do.

Ralph, an adequate response to the issues you raise would require a long essay or an hour's lecture.
I'm sorry, but you're going to have to be patient 'cos I don't have time to do this here and now.

So what follows is a very quick back of the envelope kind of response. I also really need images to convey quickly what I want to say, and I don't have these to hand. Key responses to most of the points you've raised:

• (1) Yes of course a genus should be based on relatedness. This is essential.
  (2) The number of species in any given genus is totally irrelevant. Size isn't everything, in fact it's meaningless in this context. To give you an example from genera I'm most familiar with, Aloe has 550+ spp. whilst Aloidendron has only 7. Both are good genera using currently applicable criteria, so clearly size is totally irrelevant.
  (3) It's not just taxonomy that's involved here since there are nomenclatural, i.e. priority issues of naming. This point actually relates more to Senecio than to Euphorbia.
  (4) Euphorbia does indeed have synapomorphies unique to it, i.e. features that NO other genus has. So all 2,000 spp. share certain derived characteristics which no other genus in the Euphorbiaceae has.
  (5) You say "There has to be some optimum average relatedness to determine a genus". Nope sorry, no one has ever tried to define a genus in this way. There is no unified concept of a genus based on such criteria.

Sorry Ralph but I must end here 'cos I'm going to Patagonia with the CSSA.

[ralphrmartin]

Thanks for the reply Colin.

While Wikipedia is not necessarily authoritative, it does say that (i) there are no objective criteria for grouping species into genera, other than they are supposed to be similar, and (ii) in order for a genus to be descriptively useful, it must have reasonable compactness. I presume the latter is a reference to the number of species in it, or if not that, then to the similarity of the organisms.

Yet Euphorbia is neither compact in terms of size, nor compact in terms of similarity of organisms (containing succulents, arable weeds, trees, etc etc).

So if Euphorbia meets this requirement, in what sense is it compact?

[Colin Walker]

Thanks for the reply Colin.

While Wikipedia is not necessarily authoritative, it does say that (i) there are no objective criteria for grouping species into genera, other than they are supposed to be similar, and (ii) in order for a genus to be descriptively useful, it must have reasonable compactness. I presume the latter is a reference to the number of species in it, or if not that, then to the similarity of the organisms.

Yet Euphorbia is neither compact in terms of size, nor compact in terms of similarity of organisms (containing succulents, arable weeds, trees, etc etc).

So if Euphorbia meets this requirement, in what sense is it compact?

I can assure you, Ralph, that no taxonomist I know aims for "compactness". Well, at least not in their day job.

[edds]

I agree! 'Compactness' is not a term I've heard much in classification but it seems they're talking about it being fairly restrictive to group closely related species together and not end up as a useless catch-all genus.

[John Foster]

Senecio was a huge genus of 2,000 or so spp. but is gradually being split up. So eventually NO succulent spp. will remain in this genus since they're evolutionary far removed from Senecio vulgaris, the common groundsel, which is the type sp. of Senecio.

Colin - what about borderline-succulent species? Will they remain in Senecio? I'm thinking of macroglossus, ruwenzoriensis, oxyriifolius, tropaeolifolius, angulatus.

[ralphrmartin]

I can assure you, Ralph, that no taxonomist I know aims for "compactness". Well, at least not in their day job.

Fair enough Colin. Moral: Don't believe everything you read on the Internet!

Still, I reckon what they need is some helpful computer scientists to show them how to do optimal classification based on information theory. Stir, Stir (why is there no wooden spoon emoticon).

[MatDz]

Fair enough Colin. Moral: Don't believe everything you read on the Internet!

Still, I reckon what they need is some helpful computer scientists to show them how to do optimal classification based on information theory. Stir, Stir (why is there no wooden spoon emoticon).

I wanted to share some smart papers with complex equations, but it's a real rabbit hole and I had better things to do over the weekend Dozens of "internal cluster(ing) validity indices" exist, but I couldn't find one that takes "cognitively optimal cluster size" into account. I bet there's good research on that, just hidden in various other fields (online grocery shopping categorisation comes into mind where companies spend lots of £ to optimise users' shopping journey and make them find what they want easily).