Digital Identification Tools in Regulatory Science and Practice

Abstract

Correct identification of potentially invasive organisms detected at the border, or which have already been introduced into a country, is critical to any biosecurity strategy. Taxonomic specialists have traditionally provided identification services. Unfortunately, taxonomy is a discipline in rapid decline globally, and taxonomists are often unavailable for routine identification. Molecular technology offers an alternative diagnostic tool, but currently is not generally available for the wide range of organisms that must be identified. Moreover, molecular diagnostics probably will not replace traditional identification techniques needed for rapid and confirmatory visual identification procedures. This chapter shows how the development of new software programmes and web technology provides a readily available and practical solution to the problem of achieving rapid and accurate identification of invasive pests intercepted by regulatory programmes. The current use of digital interactive keys and image databases indicate they have considerable potential to train and support biosecurity agency surveillance teams and situations in which the identification of invasive species is important. This chapter provides background information on the traditional, “dichotomous” key system and then describes the essential features of online dichotomous keys, matrix keys and image databases. Examples are provided on the way in which these tools are currently being used. The chapter concludes with a description of remote microscope diagnostics. Together with other developments (collaborative key development projects, cybertaxonomy, increased linkages between digital keys and image databases, and mobile phone technology) remote microscopy is likely to influence the future of digital identification aids.

Appendices

Appendix 1: Screenshots of Matrix-Based Digital Keys Developed in Various Software Packages and Applications

An interactive key to tribes of leafhoppers (Cicadellidae) (Dmitriev 2003 )

Beetle larvae of the world (Lawrence et al. 1999 )

North American bee-associated mites (OConnor 2003 )

1,200 Weeds of the 48 States & adjacent Canada (Old 2011)

Key to adults of beetles common in the European part of Russia and North Palaeartic (Lobanov 2005 )

An interactive key to North American Amelanchier (Campbell 2008 )

Visual generic grasses of Louisiana (Jones 2011 )

Appendix 2: Best Practice Guidelines for Making an Identification Using a Matrix (Lucid) Key

During an identification session, Lucid allows you to choose any question (i.e. a feature and its states) in its list at any time, but “stepping” through the key in a structured and sensible way will make your task of identification easier.

Familiarity with the specimen:

First, become familiar with the characteristics of the specimen you wish to identify. If you are also familiar with the Lucid key that you will use, then you may already know many of the specimen’s characteristics. Briefly reviewing these before you start will make it easier for you to proceed through the identification.

Note and use distinctive features:

In any key, some taxa may possess particularly distinctive features. Use of these may allow the taxon to be keyed out in a very few steps. At the very least, starting with particularly distinctive or striking features for the first character states selected may quickly reduce the list of Entities Remaining.

Answer easy features first:

Browse the list of Features Available and address easy features first. The principles of dichotomous keys, in which the couplets must be answered in a preset order, are very familiar to most key users who often automatically apply these principles to a matrix key. Although Lucid3 lists the features of a key in an initial sequence in the opening window, this does not mean that the features must be selected in that order. You can select any feature from any position in the list. [However, note that in some keys, where positive dependencies are used, you may be forced to answer specific questions before others become available.]

Most Lucid3 keys will have a wide variety of features, ranging from those dealing with obvious and simple features to those dealing with features that are minute, obscure or difficult to interpret. Always start by browsing the list of Features Available for obvious features that you can quite quickly answer, as opposed to getting stuck on the first one. Lucid is designed to overcome problems associated with difficult and obscure features.

Choosing multiple states:

Always choose multiple states (more than one state of a feature) if you are uncertain which state is the correct one to choose for a particular specimen. Lucid is designed to allow you to choose as many states as you require from any one feature. Within the programme’s logic, these states will be connected by an “or” link. This will cause Lucid to search for all taxa with the any of the states you select. As a general rule, if you are unsure which of two or more states your specimen has, then choose them all: that way, you can be sure that your target taxon will remain in Entities Remaining.

Finding the best feature to address next:

When you have dealt with all the obvious features, use Lucid’s “Best” function to suggest the best remaining feature. The Lucid Player has two “Best” modes, Find

Find Best and Sort Best:

• Find Best. In the Lucid3 Player, clicking the Best button will cause the Player to move to and open the best available feature. Next Best and Previous Best buttons on the toolbar allow navigation through the Features list, if you have difficulty in addressing the first feature nominated. If the list of entities in Entities Remaining changes after choosing a feature as suggested by Best, you should click the Best button again to recalculate the next best feature to address.

• Sort Best. Sort Best will reorder the Features Available list so that features are sorted from best to worst. After a Sort Best, scan the top of the list for features that you can answer most easily.

  Note that Sort Best only works using List View, as a tree representation of features cannot be sorted.

What if no taxa remain?

This will happen sooner or later in one of your Lucid sessions. If no taxa are listed in the Entities Remaining window, then it means that no taxa in the database match the selection of states you have made. Several explanations are possible; some of the most common are:

• You made an error in one or more states that you have selected. This is the most likely error for any situation in which no taxa remain.

• The taxon may be undescribed. In this case Lucid cannot identify the specimen because its features are not represented in the key’s data tables.

• The key author may have made an error when constructing the key. This is unlikely, but it can happen. If, after carefully checking all the features and states and checking that the specimen you are attempting to identify would be expected to be included in the key, then a key construction error may be present.

Whichever of the above situations is suspected, you must very carefully review your chosen features and determine which ones you are uncertain about. Try deselecting uncertain states one by one to see what effect each has. One or more taxa may move back into the Entities Remaining window. In difficult cases, you may need to “play” with the key, adding or deleting states progressively to try to find the best matching taxon.

What if several taxa remain?

Never assume that you will always end up with one taxon remaining. Some taxa in the key may be very hard to differentiate, except when using difficult or obscure features. Sometimes, after you have addressed all the features you may have a short list of taxa remaining instead of just one taxon. You are still much closer to an identification than you otherwise would have been. You may then have to carefully check the specimen against associated information (descriptions, images etc. for the remaining taxa) or refer to more advanced or specialist reference sources.

Checking the result:

When you have made a preliminary identification, check the other information (such as notes, descriptions or images) provided for the taxon. Getting a possible name for a taxon from a key is not the end of an identification. You may have made errors, or you may have a taxon that is not in the key. In these cases, the key may have provided you with the wrong name. The associated information will often give you a good indication as to whether the answer is correct.