Search Strategy for Flora-ID Northwest Databases

This page last modified on 22 March, 2008 - - Additions/Modifications likely

Class Notes by Ken Bowles. Website references are to
       http://www.kenbowles.net/sdwildflowers

Abstract

Inexperienced users of the Flora-ID databases are vulnerable to frustrating failures because they often make menu selections using UnSafe Attribute values.

This paper describes a method in which the Species Remaining list is first reduced to roughly 10 Species using generally Safe Attribute values.  After using the Gallery of photographs to verify that the specimen plant's species is still in the Species Remaining list, a method is suggested for refining the search (typically) all the way to final selection of the plant's correct species.

Summary of Method

The taxonomic database covering the flowering plants of San Diego County published by Flora-ID Northwest on CD-Rom is one of several regional databases built using the XID system for building Multiple-Entry databases. The Flora-ID databases are unusual among Multiple-Entry databases because they have taken care to avoid the Missing-Data problem that generally has made Multiple-Entry databases very difficult for inexpert users to exploit for searching the species identity of an unknown plant. For details on this problem, see our companion paper "Taxonomic Databases for Naive Users".  The main menus in the Flora-ID databases have no Missing-Data, while secondary menus on specialized topics (which do have missing data so as to minimize confusion) are all introduced by Gatekeeper Menus which themselves have no Missing-Data.

Unfortunately, an inexpert user needs to be aware of a second related problem which in this paper we'll call the UnSafe Attribute problem. In order to make productive use of the Flora-ID databases, a search strategy needs to be employed to minimize the impact of that problem. The UnSafe Attribute problem results from the fact that (until the Species Remaining list has been reduced to just a few), the user has no practical way to tell whether the Species of a specimen plant may have been removed from the Species Remaining list inadvertently. The correct species may have been removed from the list because of an Attribute selected in error from one of the many menus. Such errors occur commonly for a variety of reasons, which include the following (among others):

• Published data describing the plant species does not refer to an attribute value that the inexpert user may consider significant.
  
• The menu may list several seemingly similar values, or values the user may consider hard to distinguish - - and the user selects a value not used in the database for the correct species.
  
• The attribute values in published data, within range of a menu of interest,  may not cover all attribute values that associate with the specimen species in all regions. The gap in values may be a result of inadequate samples taken in the defining research depending on variations in soil chemistry, regional micro-climates, year to year rainfall differences, or other ecological factors.
  
• The database may have been prepared at a time before publication of newer defining data which resulted in changes in the accepted definition of a species.
  
• The user may be insufficiently familiar with the most commonly accepted interpretation of a descriptive term used for attribute values in the menu.
  
• Different botanists often differ slightly regarding the definitions applying to one or more terms used in a menu.  This commonly occurs when several different words are commonly used by different botanists to mean virtually the same thing.

The best way for a user to avoid this problem is to use the XID Gallery showing photographs of plants still in the Species Remaining list. That becomes practical only after the Species Remaining list has been reduced to roughly 10 species (at most 20), which allows comparing all the available photographs within a single scrollable Gallery screen.  The main objective is to assure visually that a plant resembling the example in question is still included in the set shown in the Gallery.  To reach this objective typically requires selecting 5 to 10 Safe Attributes.  

We define the Safe Attributes to be those relatively free of the misinterpretation issues listed above.  Unfortunately, the set of available Safe Attributes varies from plant to plant.  Illustrated examples of Safe vs. UnSafe species/menu combinations are given in later sections of this paper.

In brief, here are a few examples:  

• If the stem (i.e. cauline) leaves are simple (i.e. not compound with multiple leaflets, and not significantly lobed) then it probably is safe to use the menus dealing with "Petiole to stem leaf ratio" (i.e. whether sessile or not) and with "Stem leaf/leaflet margin" (i.e. whether entire, or toothed).
     
• If the leaf margins are actually toothed, then it usually will be safest to OR select (see illustration below) two or more of the attribute options involving "toothed".
      
• If the leaves are completely smooth (i.e. no hint of hair, glands, or other surface irregularity), then it may be safe to select "Smooth, no hair" from the "Stem leaf/leaflet upper surface" menu.
       
• If the flowers clearly are of the most common structure including Petals, Sepals, and  Stamens, then it probably is Safe to select attributes involving the numbers of those items. But if you have any doubt, then it may be UnSafe to select "Regular/radial" from the "Flower shape or type" menu.  

So judgement is usually required in making selections before reaching the stage where the Gallery Check is practical.

Other generally Safe suggestions:

• If it is obviously a Shrub (one that is more or less erect), then under "General: Woody or herbaceous Plant Habit: Woody Plants" it usually is safe to select  "Shrub erect or spreading".
  
•  If a Shrub with obvious Spines on the stems/branches then under General: Spines or Thorns" it is usually safe to select "Spines on Stems".
  
• If the plant has only a whirl of Basal Leaves, and no leaves on the stems, then under "Leaves: Leaf distribution on plant" it is usually safe to select "ONLY BASAL leaves"
  
• Length Attributes such as "Plant Height at Maturity" in the "General" menu, or "Stem leaf blade length" in the "Leaves: Stem Leaves" menu, or "Flower Width" in the "Flowers" menu are often Safe - - but with one caution.  The published ranges often don't extend quite to the length value you find on a real plant specimen. It would be very helpful to OR select several values surrounding your measured length (e.g. if you measure about 3ft, make sure to include a little more than 1m in your selected range).  Though this will slightly increase the number of species included within your selected range, it will greatly increase the probability that you will include the Species of your specimen.

If on reaching the Gallery Check you find no photos similar to your specimen plant, it will be necessary to back up to selectively remove one or more selected attributes - - followed by selection of an alternate attribute.  This can become a very time consuming and frustrating exercise, one that clearly is to be avoided.  Some practice with known species will help to avoid these situations.

Third Stage:  Verify Your Species Attributes

If the Gallery Check is successful, you can now continue to refine your search to end up (usually) with only the matching Species in your remaining list. First step is to discover the name of the matching Species by hovering your mouse cursor over the photograph that matches your specimen.  Next highlight (or double-click depending on the version of your XID program) the matching species in the list in the lower left XID panel.  (Note that in the "Options: Species description" menu of the menu-bar, the little "Data all[  ]" box should be clear i.e. not checked).  The Attributes list at end of the right XID panel will now show all Attributes coded in the database for the highlighted Species displayed in blue. You should notice that most of these attributes will more/less correctly describe your plant.  Click in the Attributes list over any of these (not already selected) that seems particularly appropriate for your specimen plant, in order to display the corresponding menu item in the upper left XID panel. This includes the number of Species within those remaining that exhibit that Attribute value.  If that number is substantially smaller than the number of Species remaining, it would be appropriate here to select that menu item in order to further your search.

Asteraceae Species

The Asteraceae (Sunflower) family species are so numerous, and so common, in San Diego County that they merit a specialized set of submenus in the Flora-ID database.  It seems that some familiarity with the Asteraceae is required to have a reasonable sense of which Attributes may be Safe when doing a search re. Asteraceae species. Note that the Phyllaries of Asteraceae plants are called "Inflorescence Bracts" in this database.

About OR Selecting

OR selecting applies within a single Attribute Menu.  Any one of the values that is marked for OR will be sufficient to assure that the correct Species remains in the selected list if that value for your Species is coded in the menu. For example, assume that with 4 Species Remaining, you find that the "Stem leaf blade length" menu shows that values ranging from 5 mm to 7 cm may apply to at least one of those 4 Species.  Figure 1 shows the two left XID panels that would apply at that point:

      Figure 1. Menu showing range of values.

Now suppose that you have measured (estimated) the leaf length to be between 1.6 and 2 inches (4-5 cm), but you are afraid the database may have coded leaves for the correct species to be just over 2 inches long.  So you Right-Click over the 4-5 cm menu item, and click "Mark OR" in the pop-up menu, getting the pop-up Dialog box shown in Figure 2.

Figure 2. OR Dialog Box

In addition to the check-mark automatically placed next to 4-5 cm, you click over the box next to 5-6 cm in order to turn on that check-mark too. You then click OK in the Dialog box, and the menu display changes to show  those two menu items with little Or symbols next to them.

This completes the OR selection for this menu, and you are left with 2 Species still in the Remaining list.

Safe vs. Unsafe Flower Shapes

 
Figure 3.  Safe flower has Petals, Sepals, and Stamens

Figure 4.  UnSafe flower. Unclear which are Petals and which are Sepals

Safe vs. UnSafe Leaves

Figure 5. Safe Leaves re. Leaf Type (Simple); Leaf Margin (Entire); Petiole to Stem Leaf Ratio (1/4 - 3/4 Total Length).

Figure 6. Safe Leaves re. Leaf Type (Simple); Petiole to Stem Leaf Ratio (Sessile; NO Leaf Stem); Margin (Toothed, directed forward OR Toothed, directed outward).

Figure 7. Safe Leaves re. Leaf Distribution on Plant (ONLY BASAL leaves)

Figure 8. UnSafe Leaves re. Stem Leaf Shape (left side: Compound; right side: Simple + Palmately Lobed)

Final Stage: Check Description Text for the Species

After reaching what you believe is the Species that most closely matches your specimen plant, you should always read the Description Text for that Species.

To do this either click or double-click the Species name in the lower-left XID panel. (Double-Click is required when using older versions of the XID program).  The Description Text appears following one or more photographs associated with the
Species in the right-side XID panel.

Figure 9. Shows the Description Text for the Species Rhamnus crocea (whereas my website photos are identified as Rhamnus ilicifolia).  This is a case where the Flora-ID database identifies the currently accepted Species (in the SDNHM Checklist v.4, and in The Jepson Manual, as Rhamnus ilicifolia) as Rhamnus crocea Ssp. ilicifolia. (Some of the photos on my website labelled Rhamnus ilicifolia are actually Rhamnus pilosa, an error which I'll correct when time permits.)

Figure 9. Description of Rhamnus crocea.

Discrepancies like this one appear in a small percentage of the Species in the Flora-ID databases covering San Diego County and California as a whole. 

A larger number of apparent discrepancies occur because the Flora-ID databases are not coded for Subspecies or Variants - - only for the parent Species.  The SDNHM Checklist v.4 includes a total of about 2500 flowering taxa including Subspecies and Variants, whereas the database includes only 1761 Species.  There are a number of situations in which the Plant Atlas database includes substantial numbers of two or more Subspecies per Species.  In some of those cases the differences of characteristics separating the Subspecies are sufficient that they will probably be re-named as separate Species eventually.  You can easily find various examples of this among the wildflower photos on my website.