Identification Keys

In designing an interactive identification key for a specimen-based virtual flora model, it was necessary to develop a system that could handle the taxonomically complex species lists that are dynamically generated from georeferenced specimen data. For example, a list of all species found within 10 kilometers of the visitor’s center on the south rim of the Grand Canyon National Park would contain a mixture of grasses and cacti, few of which share morphological characters useful for the identification of these groups. The standard method of dealing with this issue is to establish separate interactive keys for each plant family that are managed by a key to the families (a hierarchy of keys), making this awkward to use for those not familiar with the plant family classifications. To overcome this problem, we have integrated a relational data representation of the DELTA data standard with a model of the taxonomic hierarchy. This structure allows for morphological characters to be assigned to nodes within the taxonomic tree where the morphology has the most relevance. For example, awn length would be assigned to Poaceae (grass family). When using the key, algorithms evaluate the active species list and display only the morphological characters that have a high relevance to the majority of species remaining in the list. Therefore, grass characters would not be displayed until the majority of remaining species consist of members of the grass family.

There are several direct benefits to a hierarchical, multi-state identification model. First, it enables the integration of family and genus keys into a single model, thus reducing the complexity of the interface to the end-user. Second, it reduces data entry by allowing for a morphological trait to be assigned to a family or genus level, and the inheritance of that trait to all species within that group, unless a different trait is explicitly assigned and overrides the inherited trait. For example, if one assigns opposite leaf placement for Lamiaceae (mint family), all children species would inherit that trait, thus reducing the need to enter that trait for each species. When entering characters in a hierarchical manner, it is only necessary to code characters at the species level that distinguish one particular species from others within that genus. Perhaps most importantly, it avoids the matrix-based perspective and allows for morphology to be entered in a manner similar to how dichotomous keys are built where morphology is addressed only for the group of species to which it is most relevant.

Upper San Pedro Identification Key
http://swbiodiversity.org/seinet/ident/key.php?cl=3&proj=&taxon=All+Species

Dynamic Key generated from georeferenced specimen data
http://swbiodiversity.org/seinet/checklists/dynamicmap.php?interface=key