How to use keys: Some Summer Rainfall Examples
There are many forms of key in use by botanists. If you intend using keys to identify your plants you will need to understand how to "read" a key. In this issue we compare some of the features of the more common keys and philosophies behind them.
Indented Sequential Key
Here the degree of indentation determines the choices to be assessed. Where keys are long, these are prefixed with letters or (less frequently) numerals. To use the key, find the relevant pair of entries (these will have the same degree of indentation and will always precede an entry of shorter indentation), select an option and then proceed to the next line below the chosen option. This sequence follows the process that would be used in drawing up the key, and allows for easy checking and testing of the key once it is compiled.
However, it is not so easy to use: Finding alternate pairs of entries to assess may be difficult, as they may be separated by up to dozens of entries, and may occur on different pages. For any choice the key is easy to follow as the entry immediately below the option chosen is to be read. It is most suitable for keys of a page or two in length. Perhaps its biggest advantage is quickly scanning over entry pairs after having made an error, allowing rapid reassessment of choices made. It is also convenient, in that once one is familiar with the key, a very rapid scanning of the commoner used entries is possible. However, it wastes much paper and such keys are not much used now days.
e.g. Dyer's Genera of southern Africa, Phillips & Hutchinson's Flora Capensis, Levyns' Cape Peninsula, Ross's Acacias.
Key to Protea species of the Summer Rainfall Region
The following key has been compiled in order to facilitate identification of Protea species from the summer rainfall region of South Africa, Lesotho and Swaziland. The author is Janine E. Victor (National Botanical Institute, Pretoria), and the source of the key is unknown. Some terms have been explained (in parentheses) and others simplified. Note that the key is not perfect - be very careful and note features in square brackets: and if Pretoria has identified your plants as P. simplex, look again. (The Protea numbers are from Rourke).
A. Diameter of flowerheads wider than 70mm
A' Diameter of flowerheads narrower than 70mm
Numerical path key
Here the alternatives to be assessed are presented in pairs. The next entry to be assessed is referenced by its entry number: thus the entry may be anywhere in the key. Such keys require more planning in their presentation, and often the entry numbers have to be revised many times when making the key. The paths can be complicated, but are relatively easy to follow. Alternative entries to be assessed are always presented together allowing for easy comparison.
Long, complicated and involved keys are usually presented in this format, in which case the entry number is often given together with the entry number of the source. However, the format does not lend itself to rapid evaluation of entries by people familiar with the key. This is currently the most popular type of key, possibly because it takes up the least amount of printed space.
e.g. Dulpher's Ericas, Gibbs Russell's Grasses, Palgrave's Trees, Rourke's Protea, Stewart's Wild Orchids.
Key to Protea species of the Summer Rainfall Region
1. Flowerheads wine glass-shaped, involucral bracts erect with most of inner surface hidden during flowering goto 2
1. Flowerheads bowl-shaped, involucral bracts splayed to nearly horizontal with much of the inner surfaces clearly visible during flowering goto 4
2. Inner involucral bracts bearded, slightly spoon-shaped (section Speciosae). Leaves 120-250 mm long, grey, with a 10-15 mm long stalk. P. lorifolia
2. Inner involucral bracts oval, not spoon-shaped nor bearded (section Exertae), curving outwards at the tips. Heads small: 50-70 mm long. Leaves 50-110 mm long, green. P. subvestita
2. Inner involucral bracts spoon-shaped, not bearded (section Ligulatae). Leaves 80-170 mm long, silver-haired or green-hairless, tightly clustered at ends of branches, no stalk. P. roupelliae goto 3
4.  Stems hairless. goto 6
4. Stems hairy when young, maturing to sparsely hairy. [Section Lasiocephalae Savanna Proteas]. goto 5
5: Savanna Proteas
6. Shrub or tree, 1-8 m tall, erect or multi-stemmed from an underground rootstock. Flowerhead medium-sized: style 40-60 mm long. Leaves variable: 70-250 mm long, 4-45 mm wide. Involucral bracts white, cream, pink or red, short silvery-haired (rapidly shed) or hairless. Petal tips with dense tawny or white hairs, rapidly shed. Bark (old) black, deeply fissured. P.c. caffra
6. Small shrubs less than 1 m tall (rarely to 1.5 m: Pr drac), multi- or single-stemmed (Pr nubi). Flowerheads small: styles 20-40 mm long (45-60 mm: Pr drac). [Section Leiocephalae: Grassveld Proteas] goto 7
6. Large shrubs or trees greater than 2 m tall (occ. 1 m: Pr rubr), single-stemmed when mature (multi-stemmed in Pr laet). Flowerheads large: style 60-80 mm long (50-55 mm: Pr laet). [Section Patentiflorae: Mountain Proteas] goto 9
7 : Grassveld Proteas
7. Plant erect, highly branched from a single trunk. Leaves 40-60 mm long, 7-13 mm wide. Flowerheads small: styles 35-40 mm long. P. nubigena
7. Plants resprouting from a lignotuber, multi-stemmed. Leaves 60-140 mm long. goto 8
8. Flowerhead medium-sized: style 45-60 mm long. Leaves 25-45 mm wide. Stems erect, unbranched, 5-10 mm diam. P. dracomontana
8. Flowerhead small: style 20-40 mm long. Leaves 7-30 mm wide. Stems erect, usually unbranched, 2-5 mm diam., usualy short P. simplex
8. Flowerhead small: style 30-35 mm long. Leaves 5-20 mm wide, secund (pointing upwards). Stems creeping, rarely erect, often branched, 2-4 mm diam. P. parvula
9 : Mountain Proteas
9. Outer involucral bracts densely hairy. goto 10
9. Outer involucral bracts hairless. Trees 4-8 m tall. goto 11
10. Flowerheads large: style 60-70 mm long. Leaves 30-65 mm wide. Single-stemmed. Involucral bracts densely covered in a thick brown layer of hairs. Petal tips white-haired. P. rubropilosa
10. Flowerhead medium-sized: style 50-55 mm long. Leaves 15-35 mm wide. Usually with 2-6 stems arising from a rootstock. Involucral bracts densely covered in fine silver-white or rusty hairs. Petal tips tawny-haired. P. laetans
11. Flowerheads large: style 65-80 mm long. Leaves 25-50 mm wide, occ. slightly curved, deep green, very woody. Involucral bracts greenish-white. P. comptonii
11. Flowerheads large: style 60-65 mm long. Leaves 8-16 mm wide, sickle-shaped, grass-green. Involucral bracts deep red/black. P. curvata
Simple list "key"
This is not a key in the true sense of the word, but a list of diagnostic or distinctive features. It is most commonly used to list features of groups which cannot readily be separated by simple characters, but which require complicated combinations of characters for description. [these groups may be surprisingly easy to recognize, but impossible to define]. Such a "key" cannot be too long, although the Mesembs key lists almost 40 taxa, and Erica lists 41 sections (in five subgenera). Although short on space, they are not convenient to use as often several entries satisfy the listed characters, and all entries have to be read to reach a conclusion. They are suitable, however, when presenting a choice of taxa which all differ in a single obvious feature (such as leaf length), in which case the list is often presented as a "Table".
e.g. Baker & Oliver's sections of Erica, Herre's Mesembryanthemaceae classification, Reynold's sections of Aloe.
Key to the Arborescent Proteas of the Summer-Rainfall Region
This key was compiled by Jutta von Breitenbach for all arborescent proteas and modified by omitting species from the winter-rainfall region. It was published in Journal of Dendrology 6 in 1986. The numbers refer to the South African Official Tree List.
Flowerheads as broad as or broader than long. Bracts forming a bowl, opening widely.
The Compound Key
The ideal key combines features of all the above. Entries can be numbered, indented and tabulated, thus combining the best of the above systems.
The ideal key also provides as short a path as possible to a conclusion [every choice is a possible error, every choice requires some time to read, assimilate and decide]. Thus, ideally, common species should have very short paths if the key is intended for popular use (but professionals would not often use keys for such species). A balanced key can give answers to each species in 2n = X choices, where n is the number of choices and X is the number of taxa to be evaluated: the number of choices is thus dependent on the number of taxa to be considered. Thus in Figure 1, there are four species and the minimum number of evaluations required is 2 (22 = 4) per species. A less balanced key requires fewer decisions for distinctive (but not necessarily common) taxa, but requires on average more than 2n = X evaluations: Thus Figure 2 requires on average 2.25 choices per species.
The only way to shorten a key to less than the 2n = X limitation is to present triplets for evaluation, but these are a nuisance unless features are obviously distinctive. It is debatable as to whether
is any better than:
Thus although keys can replace simple lists, there are times when lists are suitable. However, in long keys such lists are easily misread when paired entries are the norm, and are thus seldom used.
The other issue about keys is whether they should reflect natural groupings or present the shortest route to an answer. The advantage of a quick, efficient key is its ease of use and time saved. However, any errors made can only be discovered at the end, and this requires meticulous and time-consuming checking of all entries starting again at the beginning. The natural key has the advantage that for each natural grouping a diagnostic summary can be made, allowing for checking at convenient levels, and thus shortening the amount of entries to be checked if an error is made. It also allows users proficient with a group to readily use only a relevant portion of the key, rather than having to start at the beginning (i.e. Which Grassveld Protea is this?) or end at the end (i.e. Is this a Mountain Protea?). The natural key thus allows users to understand the group rather than merely end up with a name. This is fine if you are interested in the group, but laborious when you are trying to compile a plant checklist for Barberton and you have three Proteas in hand plus four hundred other plants waiting to be worked out.