How to use the key
The Pea Key is an identification system for all Australian native (presumed to be present in Australia prior to 1788) and naturalised (introduced species with self-maintaining 'wild' populations) species and sub-specific taxa. Every effort has been made to include the most up-to-date taxonomic information, but as several groups are presently under revision (or are in need of revision), the taxon list will require updating from time to time. Similarly, there may be taxa as yet unknown to science, or presently known only in horticulture, which may in the future become naturalised.
There are 67 features included in the key, pertaining to vegetative, floral and fruit macro-morphology, and geographic distribution. The features we have chosen are relatively easy to assess from any reasonably intact specimen, and in conjunction with feature state images and notes, are readily interpreted. Nevertheless, some types of data are more reliable than others. In general, features that show continuous variation, such as a measurement, a relative value, or the shape of a structure, will be less reliable than those with a discrete value (such as leaf type, or spines). Features that are prone to misinterpretation (but nevertheless constitute valuable data, such as leaf or leaflet shape) have been coded broadly, or coded for misinterpretation (see below).
For each included entity, features have been assessed across a broad range of material. In many cases, this has been with reference to relevant published taxonomic literature (where the author may have seen many hundreds of specimens per species) and herbarium specimens housed in the Australian National Herbarium (CANB), or State Herbaria. In a few cases, particularly where taxa are relatively new to science, are rare or seldom collected, and are perhaps awaiting formal description, features may have been scored with reference to a handful of herbarium specimens only. Thus, even 'reliable' features may prove unreliable at a future date. Any identification should be treated with caution, and ideally, cross-checked with reference to additional material.
An important feature of a LucID key is the ability to code for misinterpretation and missing data. When building the key, we have attempted to code the correct state(s) for each feature, but also to identify and code for common mistakes in user interpretation. For example, the distinction between pinnately and palmately foliolate leaves can sometimes be obscure. This may be coded as 'pinnately foliolate' [true] and 'palmately foliolate' [true, but by misinterpretation]. The entity in question will not be discarded when either state is selected. Missing or uncertain data has been treated in two ways. For numeric features (eg. pod length), a missing value has been coded from zero to the maximum measurement for that particular feature amongst related (con-generic) taxa. For non-numeric data, all features represented within related (con-generic) taxa have been coded as uncertain. Consequently, the 'expert' user may at times be surprised to find species in the 'Entities Remaining' list, apparently in error, although any loss of power is compensated by the increased generality of the key.
Hints on using the key
When using the key, we recommend feature selection in the order presented upon start-up. If in doubt about a particular feature, this should be skipped to avoid risk of mis-identification. In many cases, it is possible to make a successful identification with sterile material (ie. using vegetative features only). However, where more complete material is available, it is advisable to enter the key at different points (ie. selecting flower or fruit features first) as a means of checking the identification. Although geographic data has proven to be valuable, we recommend that these features are used as a 'last resort', as geographic sampling is always incomplete. In general, distribution by 'biogeographic region' is the most reliable geographic feature, and by 'botanical region', the least.
The Pea Key has been widely tested and we have found that it is often possible to make a correct identification in 5-10 key steps. Nevertheless, there are instances where small groups of taxa prove difficult or impossible to separate. Where this appears to be the case, using the 'best' or 'shorcuts' options may be helpful. Occasionally, an entity is encountered which is only readily recognised by features not included in the key's character list. In these cases it is usually possible to make a correct identification by comparing the specimen with the included images or description, and/or with reference to the published literature.
Last updated 23 March, 2009 by firstname.lastname@example.org