One of my ongoing projects is to produce a “beginner’s guide” to freshwater diatoms. I wrote one about twenty years ago but the taxonomy is now seriously out of date. I also illustrated it with line drawings rather than photographs. Strange to think now but this was the last years of film photography and the quality of photomicrograph that we now take for granted was a lot harder to achieve. Nonetheless, I like to think that it played a useful role, if not as the definitive guide to the diatoms of the UK then as a “phrase book” that helps them get understand the foreign language that hardcore diatomists talk.
Most of us refer to identification guides, informally, as “keys” yet, paradoxically, experienced biologists often skip the keys and flick through the pictures, using pattern recognition in preference to probabilistic reasoning to name specimens. There is a reason for this: most keys are not very good. I’ll go further: by the time a biologist has enough experience of a group of organisms to write a key, s/he has forgotten what it is like to be a beginner. What is missing from most keys is empathy with the ignorant.
Let’s take as an example, an early couplet in the key in Freshwater Diatoms of Central Europe which asks if the cells you are looking at have internal septa. Septa are thin, silica plates which project into the cell space (see figure a. below) and are useful diagnostic characters for some araphid diatoms, in particular. Most keys to freshwater diatoms would have a couplet such as this at an early stage. However, using septa as a diagnostic characteristic carries some disadvantages because they are part of the structure of girdle bands not the valve. In the case of Tabellaria, the most common septa-bearing genus in freshwater (see image below), the cells often disintegrate during slide preparation so that our struggling beginner is faced with a few valves (b.) and a larger number of girdle bands (c. and d.). Most of the features that are useful for identification are on the valve itself but here, just to spice up the beginner’s experience, the septa are on the girdle bands. So the beginner has to detect a thin plate of silica (virtually a flat piece of glass) that is mounted between a slide and a coverslip (two more flat pieces of glass) in order to progress to the next step which asks about characteristics on an entirely separate structure.
Tabellaria flocculosa from the River Broom in north-west Scotland with septa indicated by arrows. a. whole frustule in girdle view; b. valve; c. and d. girdle bands. Scale bar: 10 micrometres (= 100th of a millimetre).
Just to make matters more confusing, some diatoms (Rhoicosphenia and some Gomphonema, for example) have “pseudosepta”, which are similar to septa but are part of the valve itself rather than the girdle band. This should not be a problem when using keys because questions about septa usually come after questions about whether a raphe is present or not. That should have steered our beginners away from Gomphonema and Rhoicosphenia except that one valve of Rhoicosphenia has a very short raphe that a beginner might well have missed.
Rhoicosphenia abbreviata from the River Derwent, north-east England with pseudosepta indicated by arrows. a. and c. inner (concave) valve; b. outer (convex) valve (pseudosepta present but not in focus); d. whole frustule in girdle view. Scale bar: 10 micrometres (= 100th of a millimetre). Photos: Ingrid Jüttner.
Whilst, in theory, the logical structure of a key should take the user infallibly to the right taxon, in practice, users tend to use the key only until the point where they encounter a couplet that cannot be easily tackled. At this point, they switch from probabilistic reasoning to pattern recognition – they flick through the images, in other words, until they find one that matches the specimen that they are trying to identify. Then they use the descriptions to confirm (or not) their hunch. The key may fail because the writer assumed too much knowledge on the part of the user, because the specimen is not “typical” (that’s for another post!) or the user’s equipment is not as good as the writer expected. I suspect that the first of these is most often the case, because the experts who write the keys have, quite simply, forgotten what it is like to be a beginner.
I do use keys a lot when teaching because I think that the repetition of a series of (more-or-less) logical steps drives home the elements of diatom morphology that beginners need in order to put names on different genera and species. Once they have got these basics in their heads, then I am happy for them to switch to pattern recognition rather than probabilistic reasoning. What I suspect happens is that schemata of most of the genera get lodged in their memory, and they can then use this information to find the right set of images from which to match their unknown specimen. The key is an important part of this process so effort put into writing a really good key should be worthwhile.
The problem lies in understanding what we mean by “really good”. There is a risk that we define the quality of identification guides in terms of taxonomy whereas didactics plays an equally important role and we cannot assume that someone who is an expert on the former is as knowledgeable about the latter. The Field Studies Council AIDGAP guides set a fine example by insisting on end-user testing to ensure the usability of keys, but these principles have not filtered through to the wider academic community. Remembering what it was like to be a complete beginner is a good start.
That’s laid out the theory. Next job is to turn this into practice ….
More about the development of keys to identify diatoms in “The decline and fall of a CD-ROM”
I also contributed to a European standard which set out the principles behind writing good keys for applied ecology:
EN 16164:2013. Water quality – Guidance standard for designing and selecting taxonomic keys. Comité Europeen de Normalisation (CEN), Geneva, 12pp.