Lucid digital identification keys: their role in on-line taxonomy
With developments arising from the Global Biodiversity Information Facility, Species 2000 and other initiatives to create on-line taxonomic databases, computer-based identification keys will have several, increasingly important roles to play. Foremost among them is the capture of taxonomists’ expertise, particularly of those about to retire, and making this expertise readily accessible for non-experts to make identifications themselves. For many potential users of taxonomic information – such as ecologists, conservation managers, quarantine inspectors, plant health officers and students – computer-based identification keys, that enable them to identify and name specific taxa, are likely to be the only effective portals by which they will be able to access the wealth of taxonomic data that is rapidly being made available on the Internet.
As with many other break-throughs in biology, when we look at the main computer-based identification systems currently available, entomologists appear to have played a central role in their development. The DELTA system, one of the earliest matrix key systems, emerged from CSIRO Entomology in Australia, CABI-KEY emerged from the International Institute of Entomology in the UK and Lucid, the system that I will focus on in this issue of Embiopteran tools, also emerged from an entomologically focussed organisation – the Cooperative Research Centre for Tropical Pest Management.
In this short article on Lucid computer-based identification systems I will provide a brief summary of what these software tools can do and illustrate how the software is being used to develop Lucid keys for a range of end-users, particularly associated with quarantine and biodiversity related activities. I will conclude with a discussion of future issues, including plans and ideas for the future as well as the vexed question of sustainability, an issue that confronts all those involved in biological software development.
What is Lucid identification software?
The Centre for Biological Information Technology (CBIT) (at The University of Queensland, Brisbane, Australia) currently produces two identification software systems – Lucid3 – the latest release of the Lucid matrix key system – and Lucid Phoenix – a new identification tool that enables printed dichotomous keys to be easily converted into interactive, web-based dichotomous keys. Both of these identification systems, like other systems such as Delta and Linnaeus II, include two components –
A key development tool or builder – that allows taxonomists or others to clone their knowledge base into a form that is readily accessible by other people.
A key interface or player – through which end-users interact with the Lucid key that has been developed, either as a CD-ROM or via the Internet.
Therefore, both Lucid systems have two important features. First, they are generic, in the sense that they can be used to develop identification or diagnostic keys to just about anything. Second, since Lucid development tools are relatively simple to use, they can be used by a wide range of key developers – from specialist taxonomists through to students learning the basics of key development. In developing these Lucid systems, a major objective has been to contribute to taxonomic capacity building in two ways – by enabling identification keys to be easily developed and by increasing the availability and usefulness of these keys by making them available on CD or via the Internet.
Let’s look at these two Lucid systems in more detail.
The Lucid3 (Matrix) key system
The Lucid3 system is the latest version of the Lucid matrix key system. In a matrix key, the process of making an identification involves selecting those states from a list of character states that best describe the specimen to be identified. These character states can be selected (or de-selected) in any order, resulting in a shortening (or lengthening) of the list of remaining taxa that best match the described specimen. Key developers can include line drawings and images to help the user to correctly differentiate between different character states and can attach information sheets on each taxon – to help users confirm their identification and to provide additional information and links for that taxon.
Fig.1. Screen shot from the recent Lucid key – “Pest Thrips of the World” – showing images for the states associated with the character “Head, pronotal and metanotal sculpture”.
The earliest version of Lucid, designed for PC/Windows, was released in 1998. The second version, which was also for PC and released in February 2000, enabled keys placed on a web site to be accessed across the Internet. However, in order to access the key, users need to have an application version of the player installed on their hard drive – a model analogous to the Acrobat Reader.
The latest version – Lucid3 (Build 1) – which was released earlier this year, is Java-based, allowing the Builder to run on any operating system that supports Java Runtime Environment (JRE), including Microsoft Windows, Mac OSX, Linux and Sun Solaris. The Player in this initial release of Lucid3 is in the form of an applet. This means that key developers can compile their key with the applet and deploy it on a web site. When a Java-enabled browser is used to access a key in an HTML page, by clicking on the URL the applet’s code is automatically transferred to the user’s computer and opens up the selected key.
Lucid3 (Build 2) – due later this year – will have all the main features of the “Lucid2” builder but also include a number of new features. This next release will also include a full featured Player in the form of an application – to be installed on the user’s computer. One of the new features included in the Lucid3 player is the ability to switch from a view of taxa remaining as a list of names to a gallery of images, allowing the user to switch to a “gestalt” method of identification when the taxa remaining list is relatively short. Finally, a server-side player is being developed that can be browsed directly on a web site – a form of the player likely to be more appealing to casual key users.
Lucid Phoenix (Dichotomous) key system
This second Lucid identification system, released earlier this year, is also Java-based and is accessed via an applet player. However, while new dichotomous keys can be built using this tool, the main reason for developing Lucid Phoenix has been to take advantage of the thousands of dichotomous keys that already exist. Published dichotomous keys can be converted into interactive, web-based or CD-based keys by first scanning the published key, using optical character recognition (OCR) software to convert it into a text file, and then importing it into the builder. A single key stroke will then deploy the key in the applet player, producing a key which is interactive, can be enhanced with multimedia, and delivered seamlessly across the Internet.
In using the Lucid Phoenix player, users are presented with one couplet at a time. As users select one of the options in the couplet, the key automatically follows the lead and presents the next relevant couplet. As the identification process proceeds, the history of these selections is shown in another window, while a third window displays the current list of taxa remaining.
Using a Lucid Phoenix key to make an identification has a number of advantages over traditional paper-based dichotomous keys. For instance, Phoenix keys can better deal with the “unanswerable couplet problem” – a well known short coming of dichotomous keys. If users of a traditional paper based key come to a couplet that they cannot answer – for instance, flower colour, where the plant specimen is not in flower – then it is difficult for the user to proceed. However, Lucid Phoenix keys can help overcome this problem by allowing the user to skip one or more problem couplets and be automatically guided through the remaining branches of the key that can be answered.
Fig. 2 A screen shot of a Lucid Phoenix key to Insect Orders – showing the four windows structure
Another enhancement in Lucid Phoenix keys, not possible with traditional dichotomous keys, is the opportunity to use filters. Lucid Phoenix allows users to identify those taxa of relevance for a particular situation (based on a regional species list, for example) and to filter out those taxa not of relevance. The Lucid Phoenix player then automatically prunes the original key to generate a customised key for the chosen set of taxa.