Multiple-Entry Key for Asteraceae - Notes

This page last modified on 15 February, 2006

The Problem

The primary reference containing detailed descriptions of flowering plants in California is The Jepson Manual (James Hickman Editor, Univ California Press, 1993) which I'll reference in the rest of this note as "TJM".  This massive book was the result of a project associated with the Jepson Herbarium, and the product of cooperative efforts by a large community of professional botanists.  Preparation of a major revision to reflect current knowledge is now underway at the Jepson Flora Project.

In principle, one finds the detailed description of a plant species one wishes to identify by following the TJM system of Dichotomous Keys.  Starting from the root of a dichotomous key, one is presented with a branch-point offering two possible descriptions of some grouping of plants (Family or Genus or Species ... , or  subset thereof) and must decide whether the specimen fits one or the other but not both of those descriptions. After following one branch, another such branch is reached describing two dissimilar subsets of the first group chosen. Ultimately, one should reach the end-point of a branch which hopefully identifies the description of the grouping one wished to find.

But there is a fundamental problem with the logic of many Dichotomous Keys, where reaching the correct Goal requires the person searching to make every This/That choice correctly!!  In the larger botanic families, one may need enough information to make the correct choice at 20 or more branches. If one takes the wrong branch at just one such choice along the way, all is lost!  If one has to guess at a  branch point, for lack of the right equipment, knowledge, or permissions to take specimens of the plant, then the probability of losing doubles at each such branch point.

It's taken me six years of on/off efforts to identify the less common plants I've been photographing to reach the point where I can make sense and good use of the TJM keys on the smaller botanic families. Discovery of the correct family as a starting point has taken a lot of time poring over the popular field guides, and occasional patient consultation with several experienced botanists (who typically are extremely busy coping with unrelated problems).  But while it is fairly easy to recognize a Genus or Species belonging to the Asteraceae family, I have always had to give up in frustration when trying to use TJM's keys leading to identification of the Genus within Asteraceae.  It's only recently that I've come to learn that this experience is fairly common even among professional botanists!!  

So it's taken me until now to undertake a serious effort to present my photographs of at least 50 Asteraceae species from San Diego County - and to fix what I've found to be a number of embarrassing errors in my species labels still on this website.

The MEKA programs and SLIKS

Sometime around the early 1970's, when information science was first becoming a serious discipline worldwide, several groups of botanists recognized that a Multiple-Entry Key can be used to avoid this fundamental problem by allowing any of several (even many) paths through choices of descriptive characteristics to lead to the correct Goal.  A simplified description of the method describes it as similar to the "peekaboo" method of finding a correct choice by using punched cards such as the Hollerith Cards ("IBM Cards") used in early data processing machines and early computers.

Perhaps a more familiar analogy for people who use computers would be a large spreadsheet with each Row representing one Genus, Species, or other Goal, and each Column representing one Character.  (Where most folks might prefer the less technical term characteristic, and computer geeks would prefer property.)  For any single Goal, the cell corresponding to a Character known to be present would be marked "Yes", the cell for a Character known NOT to be present would be marked "No", and all other cells would be marked "Unknown".  The simplest search method amounts to finding all Goals for which some selected set of Characters have been marked "Yes".  In principle, this method is a lot more forgiving than the Dichotomous Key because there may be many different small sets of Characters which, if marked "Yes", will lead to just one Goal. In other words, the search need not fail if the key lacks information on some Character that you find is obviously present in a plant you are examining. (As described elsewhere, there are some subtle problems with this multiple-entry approach, and a number of variations in the logic to resolve those problems.)

Once computers came into widespread use, several groups undertook to develop computer programs for use in building and using multiple-entry keys in botany. The project most relevant here started around 1990 as a collaboration between Chris Meacham, who worked on a program he called MEKA (Multiple Entry Key Algorithm), and Thomas Rosatti, who as a botanist was looking for a way to avoid the tyranny of dichotomous keys. Both have been affiliated with the Jepson Flora Project and its antecedents. For details, see The Meka Home Page.

Though aware of the MEKA programs  and keys for Asteraceae almost since I started taking wildflowers seriously in 1999, there were two problems that prevented me from making use of them.

• Meacham's programs were awkward to use because they were developed to run under DOS, and I found them confusing to use under Microsoft Windows.
• Rosatti's keys were written to follow an alternative subdivision of the Asteraceae into "Tribes". I've found the Tribes to be a difficult approach to follow when trying to use just photographs to identify the Genus within Asteraceae.

Fortunately, Meacham has collaborated with Gerald Guala to produce a version of MEKA keys which operate directly on the Web using Javascript program code, and a revised format Guala calls "SLIKS".  Again, for details see The Meka Home Page.  Rosatti's Asteraceae keys, along with a growing number of keys for families other than Asteraceae, are now available for public use in MEKA/SLIKS format at
     http://ucjeps.berkeley.edu/keys/sliks/index.html

Asteraceae Key based on MEKA-Edit and Modified SLIKS

Considering my long work experience in analyzing environmental data with computer programming, I decided to try building my own MEKA/SLIKS key by starting with the TJM Genus level keys and descriptions for common Asteraceae species known to grown in San Diego County. The TJM keys start on page 174, and refer to 13 broadly defined Asteraceae groupings of genera.  In effect, my purpose was to find out if a simple MEKA key, containing only Characters one can observe visually from photographs taken in the field, might offer a faster way to arrive at the Genus of a plant.

After an initial week of effort, the results were amazing for the few sample species I tried to identify using the resulting key.  I concentrated on photographs of several somewhat unusual sample Asteraceae species to see what would happen. 

The results of this made it appear worthwhile to spend two months of additional effort to:

• add more detail, eventually totalling 161 key plant characters that I found commonly (but not universally) used in the TJM descriptions and Genus keys.  
• understand and refine Guala's JavaScript code for SLIKS so as to make the step-by-step search through the available characters more efficient (and less time consuming when I made choices not supported in the TJM descriptions and/or keys).
• verify the resulting tentative Asteraceae Key by checking identification and sample details using photographs of over 40 species that I had already identified by other means.
• search for Genus identifications for plants I had photographed recently, but not yet identified. I then used the Rosatti MEKA/SLIKS Genus level keys to narrow the search in each case to Species, which I could then verify manually in the printed TJM.

From all of this it's clear that I've been able to reduce the time I must spend to identify an unfamiliar Asteraceae Genus dramatically - at least by a factor of 10!  It's also clear that errors remain in the key from several sources:

• errors in coding a total of more than 17,000 characters using Meacham's MEKA-Edit program (my own errors - not the program's !!)
• errors in translating some of the non-standard terms used by TJM authors into terms that I've tried to make mean the same thing for all of the 111 Genera listed. (The non-uniformity of terms is recognized in the Author instructions for the TJM revision project.)
• errors in deducing the value of a character based on values of other related characters. SLIKS requires a "Yes", "No" or "Undefined" value for every character. One's intuition fails when, for example, the member Species in one Genus may have Leaf-Surface which can be Glabrous, Tomentose, Hairy, etc. when it might seem just one choice out of such a list should apply.
• errors that result when the TJM author has entirely left out certain aspects of one group of Genera, where those same aspects have proven to be pivotal in identifying members of another group of Genera.
• errors that result from inconsistencies when comparing a TJM Genus level key with the descriptions for Species within that Genus. This has proven to be a very subtle problem that I'll describe in more detail elsewhere.

What I've found amazing is that this key is so redundant (i.e. providing many different ways to reach the same Goal) that it "forgives" most of those errors. In computer-speak, it provides simple workaround solutions to verifying that the correct Genus has been found.  In general, details about the Genus and Species found should always be checked with other available sources.  So the main point of using the key is to save a lot of time in finding the most likely Genus for which those details will be checked.

Link to New Key, and Plan for Posting on This Website

Using almost any Windows browser, launch the Key Here.

I will illustrate use of the Key, based on characters you may observe in my photographs, by means of annotation "Notes" linked to the initial large format photograph for each species.  Use the red Notes link at left in the window that pops up when you select one of the Asteraceae species by clicking over the thumbnail image. (If the legend "[See Notes]" is missing from the text at top of the large format photograph, no Notes are available.)  I'll post the full set of photos plus the illustrated notes on using the Key for the corresponding Species, one Species at a time. To cover all of the Asteraceae species with available photographs will take the next month or two.  You'll find the names of the Species updated so far in the Recent Changes List (linked from my SDWildflowers home page) as this work progresses.

See my Simplified User Guide to get started.

To make successful use of this key, you'll need to be familiar with the general structure common to all Asteraceae species, and with the most basic technical terms used for the Characters of those species. Here are suggested references where you'll find the Asteraceae structure illustrated and the terms defined:

• California Desert Flowers, by Sia and Emil Morhardt, Univ Calif Press 2004, pages 29-32.
• The Jepson Desert Manual, Baldwin et. al editors, Univ Calif Press 2002, Glossary beginning with illustration of Composite (Asteraceae) Family, top of  page 21.
• The Jepson Manual (TJM as referenced earlier), Glossary beginning with illustration of Composite (Asteraceae) Family, top of  page 25.  If more than just casually interested in identifying California plants, you need to use this BIG book.
• Plant Identification Terminology, an Illustrated Glossary, by James and Melinda Harris, 2nd Edition, Spring Lake Publishing 2001.  No overall illustration of how parts of Asteraceae fit together, but illustrated definitions of Asteraceae terms that are used in TJM but not defined in the TJM Glossary.

-- Ken Bowles