Plant Identification
Page 20
•
Various mechanical arrangements, as with a slide rule, where bits of card can be slid against each other or two circular cards are rolled over each other (as in wheel guides; see Table 5.1i). Usually these will be small – in effect, a stand-alone index or access method – and therefore designed for use in conjunction with books.
•
Computer databases used as multi-access keys – for example, a Lucid database. A few of these for particular groups of plants are available on CD-ROMs, such as Euclid for identifying 690 Eucalyptus
spp., produced by Brooker et al (at
www.anbg.gov.au/cpbr/cd-keys/Euclid/) (see Figure 5.2).
Web pages or otherwise linked HTML files or documents are designed to react to user input in a similar way to multi-access keys. Guide pages may appear different to different users – for example, with alternative languages, fonts, with more or less jargon, perhaps always including drawings rather than photographs, and so on.
Interactive computer identification and e-keys
Interactive identification on computers includes both interactive documents (including web pages) and specialized software tools built around databases and polyclaves. At the time of writing, computers are rarely ideal as methods for field use, particularly for the rural poor in the tropics; but this may become less so as the internet permeates ever-smaller tropical settlements, as wire-free links are developed, and as devices for using this information become smaller, cheaper and more capable. Meanwhile, there is some relevance to the current manual because:
•
field guides do have some role in the office or laboratory;
•
some field visitor centres have computers;
•
static guides, especially modular ones, can be generated and printed from dynamic computer guides as well (see Box 5.13).
Get to know the intricacies of your word processor – use the hyperlinks and so on to facilitate cross-referencing of keys to groups (see Box 5.14).
Hyperlinked documents on a computer do not necessarily lead to greater accuracy or speed. However, cross-references are very useful for linking keys and other indexes to descriptions, and cross-linking species that are similar to each other, even in documents destined to be printed (see Box 5.15).
112 Plant Identification
BOX 5.12 MODULAR GUIDES: AN EFFICIENT SOLUTION TO
THE BIODIVERSITY IDENTIFICATION BOTTLENECK?
A one-species-per-card identification pack that can be assembled by users to suit their interests can be called a ‘modular guide’. Modular guides are dynamic field guides because the contents of the guide can be rearranged by the user. A project focused on producing modular cards would have a number of obvious benefits to the project workers, the funders and the ultimate users. The following conclusions were drawn from Forestry Research Programme (FRP) trials on the subject (see Case study 8.1, page 184):
•
Usable outputs would be available from very early on in the project, and progress could be followed and used monthly as successive cards were produced. A target of about 15 to 20 species per month, averaged over a year or so, is reasonable for a researcher with assistant. It should be possible to complete a 200 species guide in a year. This incremental approach is encouraging to the guide producers, and they can get feedback on efficacy before they produce too many species in a poor format.
•
Tests on the guides can easily be designed to pick out poorly performing cards – that is, those that turn out to be associated with poor identification accuracy, or which users do not like for other reasons. Cards can be added to or improved as time goes by (for example, to correct textual inaccuracies) without republishing a whole book.
•
The cards have to be designed as picture based since browsing rather than analysis is to be promoted in this approach. This reduces language problems and the need to produce detailed diagnostic keys. However, for larger sets of cards (more than 20) it is useful to reduce the burden on the browsers by defining broad categories (for example, slash exudate and texture) that will always break up the set of species into groups of manageable size, no matter which species are included.
•
If the infrastructure for browsing the cards is well planned, it is possible for different sorts of users to select very different sets of species, yet use the same framework – for example, a ring binder. Timber cutters might only choose cards for large trees; forest managers might select these and trees of conservation interest; forest guards might include only the species in their local forest.
•
The pictures should be designed to highlight crucial details in the same position on each card. Textual keys to particularly difficult groups can be slotted into the ring binder where necessary.
Such guides are not as easily published as printed sheets, especially if incomplete sets are to be sold. For this reason, the Photoguide to the Larger Trees of Ghana’s Forest has been published bound as a book (Hawthorne and Gyakari, 2006).
The internet per se may not currently be an appropriate medium for interactive tropical rainforest guides; but it is useful for distributing printable, free field guides.
Such a service would, anyway, be very useful for helping tropical plant identifiers in tropical herbaria, where even static format guides are in short supply.
There are various classes of computer-based identification systems (such as expert systems, statistical classifiers and neural networks; see Webb, 1999; Ripley, 1996); but for botanists, electronic multi-access key types (e-keys) have predominated and are a natural progression from paper-based multi-access keys and especially punched hole
Identification 113
BOX 5.13 DYNAMIC FIELD GUIDES FROM DATABASES
Many readers will be familiar with ‘Mail merge’ features of word processors, where a template document is arranged to have some static and some variable data. The variable data area is then filled in from a separate data table when the mail merge option is run.
For instance, a form may be:
Dear
The data table might be as follows:
Address
Name
Debt
Department of Plant Sciences
William Hawthorne
500.20
Environmental Change Institute
Anna Lawrence
264.30
Personalized letters are then sent to both parties. Another report template might summarize records from the database on one sheet. In this way, the data in the letter could then be said to be ‘repurposed’. This is a basic feature of all database packages (for example, Microsoft Access or Foxpro), where reports can be generated from data in various formats.
Botanists may be familiar with the use of one set of data to generate herbarium labels, then checklists and so on. Today, images can also be added to the reports so that a field guide of user-definable format and content can be generated from a field guide database. This is obviously useful if you want to make field guides customized for particular areas or forests because it is a trivial matter in any database to filter out some of the records. On the other hand, compared to static formats, there is less scope for the personal touch, to optimize paper space or to allow more room for difficult species.
If the database interface is the internet, field guides can, in principle, be customized for many types of user. This approach, embryonic but developing at the time of writing, is bound to evolve over the coming years, with data exchange potentially facilitated by the formal separation of style from information using XML (World Wide Web Consortium, 2002). Ideas vary from simple production of sheets with pictures and text to full use of the internet, with cross-linked databases and shared resources.
Further information can be found from the Openkey (www.isrl.uiuc.edu/
~telenature/about/index.html#openkey),
Hopscotch (www.hopscotch.ca/guides-on-demand/about.html), Enature (www.enature.com) and Bibe (www.biobrowser.org) Initiatives; and also on the Virtual Field Herbarium (http://herbaria.plants.ox.ac.uk/VFH).
cards (see Case study 5.1). There are at least 15 published e-key packages; but some, such as Pankey, have been superseded. Others are only available as published keys with data or, like CABIKEY and Linnaeus II, linked to publishers and may not be suitable for small research projects; so the practical choice for most field guide creators is more limited. At the moment, Lucid (see Figure 5.2) is probably the leading e-key package designed for both authors and users. See Pankhurst (1991, 1993), Bridge et al (1998) and Dallwitz et al (1998), Dallwitz (2000).
114 Plant Identification
BOX 5.14 WORD PROCESSOR FILES, ACROBAT PDF FILES
AND INDEXES
Word processors are all you need to make an interactive key: web pages have many of the same features. A word processed field guide, possibly saved as Adobe Acrobat (PDF) or HTML files, can be published directly on the internet. Here are some aspects of electronic documents that can help you to make your key more interactive and information rich:
•
Use tables (with invisible grid lines) to help format keys, with species names or groups aligned in a right-hand column.
•
Use outline-numbered styles for the group headings. Insert cross-references (hyperlinks) where relevant from the end point in one key to the start of the next key so that users can navigate back and forth, as in web browsing.
•
Create indented key levels as styles (special fonts, etc.), using a number or other symbol as the bullet to designate each level. Create styles for descriptive text and species names. Also use different colours or fonts for the different levels of indent in a key. Use of consistent styles for different types of information is helpful in itself for maintaining consistency, but should also facilitate transfer to an XML format for data exchange.
•
Use the outline level functions for keys to groups (level 1), sub-groups (2) and even indentation levels within keys. Users can then collapse or expand outline levels, or navigate using the document outline.
•
Create a glossary and designate the technical words in it as bookmarks or headings.
You can hyperlink these words in the main text to the book mark to facilitate glossary use by people reading the document on the computer.
•
For guides with many species (hundreds), create documents for large groups as separate files. You can index and interlink these documents in Word and Acrobat, so the files can be used independently or as a whole guide.
•
Include references to useful websites.
•
Experiment with the index functions in your word processor: it is possible to make indexes to glossary entries, and to species names, synonyms and local names without all of these alternative names necessarily being visible on the relevant page. Indexes to glossary entries can be developed as a quick, basic multi-access key.
In Acrobat PDF files, experiment with photographs that are of higher resolution than needed for printing. Users can zoom into photographs to investigate details of the plant.
Available e-keys work in a similar way, based on a matrix of characters (sometimes called ‘features’) or character states (‘attributes’) by taxa, in a database that can be sorted or queried in any order. A standard called DELTA (Description Language for Taxonomy) exists for communicating between character databases. Some programmes allow users or authors to score the characters for reliability, ease of use or interpretation, rarity, etc. The easiest characters can be considered before the more difficult. Other un-coded textual information and pictures may also be shown, and picture browsing is also more or less supported.
At the time of writing, the image browsing, zooming and related functions are much more limited than in dedicated image database software; this aspect shows the greatest
Identification 115
BOX 5.15 EXPERIMENTS ON THE VALUE OF INTERACTIVITY IN KEYS
In interactive documents, hyperlinks can be used extensively in a bid to facilitate identification. Surprisingly, Wright et al (1995) and Edwards and Morse (1995) found that users were slower and less accurate with a hypertext version of a key, and were more inclined to become ‘trigger happy’ and less likely to retrace their steps than with a paper key. Morse and Tardivel (1996) tested on zoology undergraduates a dichotomous key (with or without the hypertext links discussed below) and a multi-access key to woodlice. Multi-access keys were generally slightly slower to use than paper dichotomous keys and did not produce the right answers significantly more or less often (the overall accuracy was 74 per cent). Paper keys were associated more often with confident, but wrong, answers than electronic ones with hyperlinks.
Source: from Euclid – an implementation of Lucid software with Eucalyptus database –
see www.anbg.gov.au/cpbr/euclid/euclid.html
Figure 5.2 Screenshot of an e-key in action potential for future improvement. Until then, it is an open question how far computer guides will be able to facilitate recognition-based identification, even if the computers can be taken where they are needed on a budget that is realistic for rural, tropical communities.
116 Plant Identification
CASE STUDY 5.1 AUSTRALIAN RAINFOREST KEY
One of the best working examples of an inter-
active key for rainforest plants is Hyland and
Whiffin’s rainforest key, which also shows an
interesting evolution from card-based
polyclave to a computer-based key. In 1971,
this covered 584 taxa on 80-column computer
cards, with 48 bark features and 45 leaf
features. This grew until 1982, when it covered
799 taxa, the limits of the cards – with the
same features, but extra information added for
families and geography. This was converted to
a computer format in 1983 covering 1058 taxa,
and this was coupled with a leaf atlas and
extra computer data for flowers, fruits and
seedlings. In 2000 the system had expanded
to all 1733 Australian tree species in a
Windows interface. By now, four people were
involved. The most recent 2003 version covers
shrubs and vines. Unfortunately, the system is
not designed for other authors to use, unlike
the Lucid system where similar packages –
ready filled and available with data, like Euclid
– are increasingly common.
Characters in similar software are
normally defined per species; but the rainfor-
est key is more painstakingly based on an
underlying core database of specimen-level
records, facilitating the update of the dataset
when taxonomic opinions of the specimens
shift.
Figure 5.3 The evolution of
the Australian rainforest key
Sources: Images reproduced by kind permission of Trevor Whiffin; text based on www.anbg.gov.au/cpbr/cd-keys/rfk/history.html
Identification 117
CHOOSING ACCESS METHODS AND THE MEDIUM
FOR YOUR FIELD GUIDE
There are many factors to weigh when selecting the format appropriate to your needs.
Here is a summary of some of the pertinent points:
•
Since all user groups are different, use the consulation methods described in Chapter 3, and consider performing accuracy, confidence and other tests on your own field guide users to optimize your proposed access method (see Chapter 9) – you may well be surprised by the results.
•
Critics who decry paper keys are often confusing the access method, which could hardly be simpler, with the (technical, jargon-rich, picture-less) c
ontent of traditional biological keys: ‘For non-experts traditional keys are often difficult to use because a scientific language is used and illustrations are often lacking or placed elsewhere’
(CABI, www.pest.cabweb.org/Identification/BIOSYS.HTM). Until computer identification systems are as cheap, robust and future proof as field guide books, traditional dichotomous keys should be the first default type of key you try out with your users.
Computer-based dynamic guides and field use
•
In e-keys, users can avoid or delay questions that they cannot answer. This is particularly valuable for fieldwork, where only fragments of plants may be available. The flexibility of data display in an e-key is undoubtedly a huge boon.
•
Some e-keys do allow commoner species or more important characters to be prioritized; this option is usually taken for granted in static keys; but potentially, at least, an e-key could adjust this option depending upon where you are and the species remaining. Sensitivity to user skill is another key asset in principle, although this is hard to implement. An e-key could prioritize characters based on user skill, species remaining, location, and feedback from past instances of the use of this character.
This would be harder to set up, but probably worth it for e-keys with large markets.
We look forward to further developments in this direction.
•
Modern students are generally comfortable with the ‘point-and-click’ browsing approach to information, and interactive software may draw some to tropical botany if the interface seems modern, regardless of the content or actual usability.
An interactive CD version of your book need only involve hyperlinks to be a useful accessory.
•
If a decision is bothering you as author, or there is a conflict among potential users (Which species should I include? Should I emphasize local or Latin names?), maybe you could supply both formats, or species lists, with software ‘switches’ to change between them.
•
Evolution of e-keys is slow compared to other software, but they do increasingly allow reasonable illustration of taxa, with multiple images, maps and textual information, to be viewed when necessary to confirm an identification by selecting the appropriate link. In book format, all of this information would soon add up to a key heavier than a portable computer.