Plant Identification

by Anna Lawrence

  One might suppose that a useful way to research a field guide would be to ask local experts ‘what is that plant’ and then ‘how do you recognize it’. Of course, it is essential to involve local experts, or other experts if you are a local expert, especially to locate and identify species in the field; but they may not know entirely how they are recognizing the plants they know. When asked, some secretly bemused experts latch onto any obvious feature, struggling to invent rules on the spot, although with practice and experience of thinking analytically, or of teaching, a few can translate experience into reliable rules better than most.

  Psychologists point out that recognition is largely based on ‘ gestalt’, or ‘spirit of the whole’, rather than analysis of successive questions; on computers, ‘neural networks’ can be trained to behave in a similar way. However, when the neural network has been trained reliably to distinguish a spanner from a hammer, it is not then possible to open up the circuit and obtain a set of rules about how to recognize hammers and spanners that could be used elsewhere. So, do not be surprised when your expert’s brain struggles to speak the rules for separating two species which that expert can reliably distinguish. Unless, of course, some rules can be remembered from the field guide or expert who first taught them.

  Identification tools, like field guides or herbaria, should enable us to identify plants using our powers of recognition or analysis or, most likely, a combination of both.

  Authors have to strike a balance between recognition and analysis that suits the intended users of their field guides.

  Box 5.2 explains why you cannot create a field guide by simply publishing the information that experts have in their head. Another reason that field guides cannot emulate the mind of experts is that texture, taste, smell and three-dimensional perspectives are not well dealt with by currently practicable media, and are often very hard to characterize anyway. Recognition seems to involve a set of many fuzzy mental images or memories, covering all forms of a species; yet field guides need to be concise and intelli-gible when summarizing variation and must therefore stick to characters that are more reliable and with precise limits. (see Box 5.3).

  One way or another, it is essential mentally to break up your plants into parts if you intend to make good use of your data on a computer, even simply to store data in a database. Furthermore, characters have to be defined if you want to promote identification by analysis. Even if you are to promote recognition-based identification and have no computer, you will find it a useful discipline to define and qualify characters – make a table or spreadsheet of species, characters and character states – in preparation for your field guide. The logical and consistent thought about the plants that this demands will

  Identification 93

  BOX 5.3 CHARACTERS

  A character is any observable aspect of a plant – such as flower colour or a preference for swamps – that can be referred to with a word or phrase reliably and objectively, and itemized or quantified in terms of ‘character states’:

  •

  Binary characters, such as ‘thorny’, have only two possible states: present and absent (or true/false, yes/no).

  •

  Other types of character are best specified by numbers or counts, perhaps a set or range of possible values (for example, number of petals) or lengths or, in some sophisticated approaches, a probability that a character state is present.

  •

  Categorical characters are specified in terms of multiple choice character states as words or codes – for instance, the character ‘leaf arrangement’ can have states

  ‘opposite’, ‘whorled’ or ‘alternate’, which would normally be codified (for example, O, W, A, or 1, 2, 3) for efficiency on a computer.

  The choice of how to define a character is more about semantics, computers and convenience than botany. There is no right or wrong way of defining any particular character.

  Leaf arrangement could equally well be defined as a number – for example, number of leaves per node, with opposite defined by a character state of 2. Similarly, categorical character states can be thought of as, and translated to, a suite of binary characters, such as ‘opposite leaves?’ (true or false?). In this case, other linked binary characters, such as

  ‘alternate leaves?’, would also be needed; the binary option is the best if any species can have both opposite and alternate leaves.

  Not all people in different cultures, not all languages, agree on how plants are subdivided into separate parts or characters, nor on how the various character states are to be defined (see, for instance, Berlin et al, 1973; Hunn, 1977; Atran, 1990; Davidoff et al, 1999); in fact botanists often do not agree with each other. Although, as Mendel (1822–1884) showed with his peas, some characters reflect individual genes, there is not necessarily anything so fundamental about them. The variation around the world in the boundaries between our words for colour or taste, or plant types and parts, form a whole anthropological discipline on their own. Fortunately, the degree of ‘common-sense’

  universal agreement seems far more impressive (Witkowski and Brown, 1978; Brown, 1991; Wierzbicka, 1994), particularly for subjects that we as individuals find the easiest to distinguish and define. Colours and tastes are highly subjective, difficult to make a consistent classification of even in our own minds; so it is not surprising this is where cultural differences seem to be greatest.

  force you to clarify details you might otherwise leave unresolved or unexplored, and will help you to choose the best order of species and choice of images, even for a picture guide.

  Botanical identification tools, in general, tend to promote analytical identification because this is the easiest way to make a publication accurate and concise; identification rules or diagnostic characters are independently testable and easy to communicate and copy. Botanists are traditionally trained to think and write in this analytical way, and it provides a discipline that is useful for science students. The process of successive approximation, of gradually closing in on a solution, identifying the broad characteristic of the family or genus first, then gradually specifying the detail and pinpointing the

  94 Plant Identification

  BOX 5.4 GENERAL CRITERIA FOR A USEFUL SET OF

  DIAGNOSTIC CHARACTERS

  Describing a set of diagnostic characters across many species can be a lengthy process, so prioritization of some generally useful diagnostic characters for your particular combination of users, budget, access method and species is essential. Obviously, some types of character need only be researched in subsets of species (for example, yam shape or taste among Dioscorea). A good character set for a particular set of species should have many of the following attributes:

  •

  They vary much more between species than within species, or at least limits can be defined for each species that often differ between species.

  •

  They should complement each other – that is, some should differ among species that otherwise have all other recorded characters the same. Avoid wasting time documenting characters that are more or less correlated with others that you have already recorded – for example, latex in the petiole, when latex in the bark has already been considered – even if there are some minor differences.

  •

  They vary in a way that can be summarized and communicated reliably, with little effort. Smells and tastes, while distinctive, are often very hard to communicate on paper and across cultures. The character should be easily observable and reliably interpreted by your specific user group and by you (see Chapter 9). For instance, non-seasonal characters from low down on a large tree are better than seasonal details of the crown. The perfumed scent of leaves and bark of certain plant families (Annonaceae, Lauraceae, Zingiberaceae, etc.) are notable exceptions because, although not initially clear to many users from a textual description, they can be quickly learned and define large groups that otherwise have few completely distinctive vegetative features.

  �
��

  They are straightforward to research. Characters that are often recorded in books or herbaria might have this advantage if you have easy access to these. Start by looking in existing Floras and monographs to see what characters they use. However, beware that you will still have to concentrate on field characters that we discuss in Chapter 6.

  •

  The best characters to use in keys are often those which can most easily be reduced to simple and clear codes in a database field. If you need a sentence or even more than a few letters or digits to define a character state clearly, then it will often not be very easy or efficient to use in a key, although a short list of discrete options (for example, countries in which a species occurs) can work well.

  As an example, the range in the number of pairs of lateral nerves on a leaf is a useful field character to consider, even though it varies on the leaves of one plant, because it is easily recorded, easily researched and the lower limit of many species will be higher than the upper limit of others. The texture of leaves, however, is less useful as it is harder to record, research and communicate. You will come across hundreds of such characters – ‘cut wood with a slight smell’; ‘older trees with rather dark crowns’ – but do not ignore them altogether as they can be useful to help users confirm the correctness of an answer suggested by the other characters and the access method. You might relegate them to note fields rather than coded into neat categories in your database. As a last resort, if you do use such characters in a key, try and ensure that they are the last questions or nearly so.

  Identification 95

  species, forges a mental and information framework useful for describing and learning new species. However, not everyone needs such a mental framework; they just need to know a few names or details for plants that they come across.

  Many successful field guides make good use of identification through recognition, using pictures with little diagnostic text, and there is no reason why these cannot also help users develop a useful, systematic mental framework. Field guides for birds and some other animal groups encourage more browsing than plant guides. Stevenson et al (2003) even distinguish bird field guides from ‘keys’, as if by definition a field guide is always less analytical – a distinction which does not apply to plant guides. Larger animals, in general, are more amenable to pictorial guides than plants as there are fewer of them, and visual signals which humans can see have often been important in their own evolution. We consider below the increasing role that browsing might have in modern plant field guides.

  Your major task is to discover and document or illustrate characters that enable plants to be distinguished in the field. There may be no one in the world who consciously ‘knows’ all of the diagnostic rules you need in your field guide – existing botanical works will probably not contain all the answers, either – so do not expect to find them entirely by asking experts or reading books.

  Decide on a good basic set of key characters as soon as possible after you have chosen your species; this will help you to focus your attention on unknown facts during fieldwork and help you take more useful photographs. Some requirements for these characters are explained below and in Chapter 6; others are more general (see Box 5.4).

  TYPES OF GUIDE: TYPES OF ACCESS

  Field guides provide botanical information in two main forms: the actual descriptive material, including text or pictures, and the indexing, page arrangement and keys –

  called here the access methods – to help readers identify or reach plants in which they are interested. Our definition of a field guide has identification as a main feature, so an access method that enables navigation to the right species based on its details is generally crucial. A suitable access method can be defined in consultation with the intended users (see Chapter 3). More essentially, it must be tested before finalizing – see Chapter 9 for methods and examples.

  The descriptive element of a field guide should allow verification of the identity of species independently of the more concise or cursory information used for access. In fact, the same information ideally functions in both ways, on different occasions: if the species has been identified initially by browsing pictures, the textual description should allow confirmation; if access to the species was via analysis and reading of the text, then the pictures should help to confirm it independently:

  •

  Promote structured browsing – that is, an effective combination of analysis and recognition. Place pictures within groups where groups are defined in a sequence, based on simple characters (for example, leaf arrangement and margin details): users can jump directly to the right groups based on rules or experience; then browse pictures to reach a matching species; then maybe check the text to identify subtle differences between similar species (we are trying to achieve this effect for image libraries of random plant images in the Virtual Field Herbarium).

  96 Plant Identification

  BOX 5.5 TYPES OF STATIC GUIDE AND TYPICAL

  ASSOCIATED ACCESS METHODS

  •

  Books and other standard printed works are the typical examples, and many access methods have evolved.

  •

  Posters or large single information sheets for public display of information may serve as a field guide, especially if arranged as a fold-out map. Access to particular species on a poster may not be deemed crucial; but a good design may throw particularly outstanding patterns of similarity into sharp relief from a distance, with finer detail nested for closer examination.

  •

  Laminated identification cards display all information on one sheet (for example, the Chicago Field Museum’s Rapid Colour Guides featured in Case study 4.4, page 88).

  These do not present much of an access problem because there are few enough species to see all on one page. They are designed to facilitate browsing. If several such sheets are used together in a ring binder, the format of the guide as a whole becomes dynamic: if the sequence of cards is fixed, we have a special type of book. Likewise, collections of actual herbarium specimens or copies of them designed for field use (as with the Chicago Field Museum’s Rapid Colour Guides) may have a static format.

  •

  ‘Novelty’ (not very commonly employed) devices, such as identification wheels –

  tables arranged in a circle, with the more general character states towards the middle, and the species on the rim, but with no moving parts. These rarely represent a full guide, however, and typically lack descriptions.

  •

  There are also certain computer versions of any of the above – for example, basic word-processed documents without hyperlinks and basic websites with no dynamic content.

  Simple guides may work by browsing alone; this approach may even work for the more dogged enthusiasts with larger reference books. However, there will usually be too many species in a tropical forest guide to make browsing of all descriptions an efficient option, so a specific access method is usually needed.

  We can also distinguish between static and dynamic formats of field guides:

  •

  In static guides (normal field guides), the content, sequence and format are fixed at the time of publication. These include virtually all early field guides, notably books, and a majority of current ones (see Box 5.5).

  •

  In dynamic formats, the order and sometimes other aspects of the content of a guide can be rearranged during use, or at least after initial publication, either on a computer or by using devices such as card indexes and ring binders. It can be assumed that dynamic format field guides will become commoner as computers become smaller, cheaper and more field usable (see Box 5.11).

  Access methods in static guides:

  The primary sequence, keys and indexes

  Primary order of species in a field guide

  Static guides have the advantage that the order of the species is fixed and access and other aspects of information layout and style can therefore be optimized for this order.
>
  Identification 97

  For instance, the page sequence is fixed in a book, so indexing (by page numbers) to common and scientific names is trivial, as are indexes of plant appearance or keys.

  In a dictionary of words, in alphabetical order, an unknown word is found by matching the first letter, then the first two letters, and so on, until by successive approximation the correct word and its definition is found out of many thousands, hopefully in a matter of seconds. If the main aim were to find a word for a given concept, a book can be arranged as a thesaurus, with words for similar concepts placed together. Similarly, the order of species in a field guide should suit the main aim of the guide. There may be various indexes to help find information by alternative criteria – for instance, the diagnostic keys, and local and scientific name indexes; but the primary page order of the guide forms a crucial part of the access design for a field guide. There are two main choices:

  1 Name order: your typical users may think they know a name for a plant, but need to look up information about that plant in your field guide, perhaps to confirm its identity. If this were the main type of query, it would be best to arrange the information in alphabetical order of plant names, like a dictionary. The guide as a whole is then its own index. The only question, then, is should the order be common or scientific name? That will depend largely upon the relative importance you attach to these names – that is, user demand – and the reliability of local names (see Box 4.1, page 63). Keys can then provide a second access method, based on plant appearance:

  •

  Local or trade name order may be preferable for a guide with a limited readership, where most people are reliably using the same names for the same plants, and if most queries will be to confirm field details based on these names.

  •

  Scientific name order with a separate index of local names is the obvious solution for a technical or varied readership book that is sorted primarily by name. Alphabetical order of genus and species is useful if there are few species per family and people are likely often to access a species via its scientific name, perhaps having found it in a separate key or index. If there are many species per family, placing the species in family, genus and species alphabetic order, while not perfectly nature order, allows some general descriptive information to be included for each family and genus, saving space at species level.