Echinofossulocactus for id

[ralphrmartin]

Not a great match between that photo, Phil, and my plant, is it?

In particular E. crispatus seems to have flattened centrl spines (much broader than deep) which my plant doesn't.

So much for keys and descriptions.

It's probably yet another case of species being a human concept, and not everything out there fits into our neat pigeonhole system. I'll just enjoy the plant for what it is.

[Pattock]

Keys are only there to help, you can't rely on them to provide a correct identification.

[ralphrmartin]

A problem with many botanical keys is a lack of understanding of dichotomy (or at least proper application of it) by the person who constructed the key, leading to keys which simply don't work.

Two classic problems are exemplified by

Spines up to 1cm }
Spines over 2cm } You are sunk if your plant's spines are anywhere between 1 and 2 cm

Short red spines }
Long black spines } You are sunk if your plant has short black spines, or long red spines

A proper dichotomy should
(a) involve only one character, not several like the second example,
(b) should always lead to one alternative or the other, whatever the facts, not leave cases leading to neither outcome like the first example.

I'm not pretending it is easy to write keys which follow these requirements, but failing to do so results in many keys simply not working when faced with an unknown specimen. Because of this, I prefer an approach which lets you put in the information you do have (e.g. a plant may not be in flower to tell the flower colour), and uses some statistical analysis to tell you what you are most likely to have. The problem with that is you need to study whole populations of plants, to know that e.g. species A has spine length 1 cm with a standard deviation of 3mm,. while species B is more variable and has a spine length of 2cm with a standard deviation of 9mm. Having done that, though, given an unknown plant with spine length 1.2cm, you can work out the probability of it belonging to each species. Add in a few more characteristics (number of spines, number of ribs, flower colour, etc, etc) and further calculations can tell what the most likely answer is, and how likely (70% chance you have this, 20% chance you have that, <5% each it is one of several other species). Of course, this needs some maths, but as "everyone" has a mobile phone these days, it would be easy enough for a program to ask the questions, collect the answers, and do the sums.

All of this assumes, of course, that everyone agrees what the species are, and someone goes and measures plenty of specimens. Unfortunately, I don't suppose those assumptions are too realistic.

[David Neville]

Keys are only there to help, you can't rely on them to provide a correct identification.

The late great Gordon Rowley used to say that a good, worthwhile botanical key should enable precise species identification. If it doesn't work, then it's not a proper key.

I've never heard anyone say before that keys are only here to help......

[edds]

Completely agree with both David and Ralph.

Too many keys, especially in books intended for a hobbiest audience seem to pick 30 common species or forms and try and sort them out with as few steps as possible rather than being a technically accurate key that will sort the full range of features within a group.