Stage 3 - Differentiating Among (Sub)Species Within One Genus

Stage 3 - Differentiating Among (Sub)Species Within One Genus

This page last modified on 04 October, 2006

Not all Asteraceae family Genera known in San Diego County lead quite as easily to a (sub)Species identification as I've illustrated with the Efficient Stage 2 Search Strategy example in the previous note. A typical scenario will leave several members of the same Genus (i.e. that of the apparent most likely search solution) showing fewer than 5 State differences from the most likely (sub)Species. But all listed (sub)Species belonging to other Genera will generally show 5 or more differences from the most likely Entity at this point.

It's at this point where limitations of both the Lucid Multiple-Entry key, and the TJM Dichotomous key approaches will become more apparent if you compare them. This page will illustrate various of those limitations by presenting the identification of Hulsea californica plants through observations from my photographs.

The initial photograph, which you can display using the first (tiny colored photo ) icon next to the species name in the Entities Remaining tree, is in ScreenShot #1 showing results of a Stage-1 search:

The Stage-1 search was relatively easy because these plants are spectacularly woolly during the earlier stages of their blooming cycle. However giving priority to SubFeature/State pairs (e.g. found near top of the Entities Remaining list; then checked in the All-Characters-Profile page), showing the lowest available Key-wide counts, did turn up a false match in the Eriophyllum genus. I resolved this quickly using the methods described re. error-discovery during the Stage-2 search. This turned out to be a situation with very few Key-wide counts in the very low range well below 50 -- which would have indicated comparitive uniqueness.

Screenshot #2 shows the results of a Stage-2 search which started from the status shown above:

Things to note:

All of the additional States chosen correspond fairly obviously to Characters of the plants seen in my photographs.
All Entities Remaining, other than the four Hulsea species, differ from Hulsea californica by at least 6 States. In fact, except for the one Chaenactis, two Lasthenia, and one Layia examples, all the rest differ by at least 7 States. This reinforces confidence that the correct Genus is indeed Hulsea.
By carefully reviewing the Hulsea Genus-Profile page (click the document icon next to the Hulsea Entity), you can find a few additional States which will "demote" one or more of the three Hulsea Entities other than H. californica. But good judgement suggests that those additional States apply to H. californica only marginally. Indeed choosing them beyond this stage would tend to increase the noise background relative to H. californica.

A review of the Genus-Profile compared with the TJM key for the Hulsea genus (page 290) turned up the following categories of issues:

Errors and coding problems in the Lucid key. Some of these were a result of my failing to understand fully the implications of TJM editorial rules that are summarized in the TJM Conventions on Keys and Descriptions (pages 6-8 in TJM vs. pages 3-5 in The Jepson Desert Manual -- noting that the two are somewhat different!!). Also notable: These rules have been tightened and made more precise in the Guide for Authors of TJM2 which is now in preparation. These problems in my Lucid key may be expected to cause marginal within-genus search errors in other genera until they are discovered and corrected (Feedback E-mail on any such errors you discover will be much appreciated!!).
Inconsistencies in the TJM Genus-key which at-best require careful interpretation -- especially in relationships with the genus and species level descriptions if one assumes that the editorial rules have been followed.
Finely shaded differences, among the four Hulsea Entities, based on phrases that are really not practical for use as States in a multiple-entry key.
Within-Species variability of the States that apply to some Features -- thus making the plants at hand appear to fail a 100% match with any of the keyed Entities in the Genus.
Character (Feature) States eliminated from the TJM keys and/or descriptions for editorial (mainly space reduction) reasons -- probably on grounds that they did not contribute significantly to identification of the (sub)Species.

The following list of short notes discusses examples of these issues for the H. californica plants shown in my photos. The order of presentation more/less follows the TJM order of the main Features, which is also the order used in the Lucid key:

The TJM Hulsea genus-key contains the following couplet re. the Leaves: Surface: "Feature":
```
   1. Lower lvs gen glandular; gen barely hairy  .... H. mexicana and H. heterochroma
   1' Lower lvs ±woolly or long-soft-wavy-hairy  .... H. californica and H. vestita
    
```
My coding of the key appears to have treated "barely hairy" to mean "soft hairy". My photographs show that "long-soft-wavy-hairy" was accurate for some leaves on H. californica later in the maturity cycle, probably as a result of flattening of the earlier notable "woolly" appearance because of aging or environmental influences. A text search thru all 317 TJM online description files found no match for the "long-soft-wavy-hairy" term -- it occurs only in the Genus-key (which I used during coding the Lucid key occasionally -- but not for coding every (sub)Species), and perhaps only for this one species.
Bottom line: Adding special States to the Lucid key to remain faithful to the couplet above would make the key more confusing. It seemed better to refer even an inexperienced user to TJM regarding these fine descriptive details. The clue in this case was the Profile pages within-genus counts for "Leaves: Surface:" State lines for Soft Hairy and for Woolly.
Discovered using the within-genus counts, this time for "Plant: Surface:"
```
   Woolly ...... H. californica
   Hairy-Bristly ...... H. mexicana
    
```
This exposes issues about my coding of the Lucid key. The TJM description for H. mexicana says "± soft-hairy ...". For H. californica the overall plant description says "± woolly ...". Though not stated directly, the TJM Conventions on Keys and Descriptions imply that these States override the slightly different characterizations in the genus-key. But nothing here justifies the term "bristly" in this context.
When judging how many distinct State values to put in the Lucid key for "Plant: Surface:", I had decided to lump Hairy and Bristly into a single choice because the range of descriptive terms I had encountered (in TJM for the San Diego County universe of Asteraceae (sub)Species) did not seem to justify using two distinct terms. This Hulsea example raises the question of whether it's worth a key-wide review of the codes I've used for "Plant: Surface:".
Discovered using the within-genus counts, this time for "Leaves: Margin:"
```
   Entire ...... H. californica,  H. vestita       
   Toothed ...... H. mexicana, H. heterochroma
    
```
The TJM descriptions are the sources for all four of these. This comparison favors H. californica within the Genus. But it's noisy key-wide because "Entire" is listed for 175 Entities. However, once we're mainly concerned with the Genus, use of this to differentiate from H. mexicana does seem justified since a close look at my leaf photos doesn't show evidence of marginal teeth (though scalloped edges do suggest teeth if one doesn't look closely).
Leaves: Place: Reduced Upwards
Initially, I scored this Present in all four Hulsea (sub)Species, so it wasn't useful for differentiating within the Genus. But there are differences in the TJM Genus-key, mainly that the leaves of H. vestita are "abruptly smaller above rosette", while those of the other 3 entities are "gradually smaller upward".
```
   "gradually smaller upward"  ..... H. californica (TJM genus-key)
   "± reduced upward ..... (TJM genus description; applies unless overridden below)
   "abruptly smaller above rosette" ..... H. vestita  (TJM genus-key)
    
```
So I altered coding of the key to show the H. c. and H. v. scores as "?" to suggest that a user should refer to the TJM genus-key re. these subtleties. The Lucid Multiple-Entry key is really not designed to make these fine distinctions more easily apparent. To augment the Lucid key, the "?" notation in the Profile pages provides a flag suggesting that the user should refer to the TJM details. For identification of the plants shown in my photos, it seems we can conclude that those plants are not an instance of H. vestita.
Leaves: Shape: Oblanceolate
Continued analysis of the Ranked Entities Remaining list, when 10+ States were checked, turned up clues on when to look closer at the Genus-level keys. Example: for Hulsea vestita ssp. calicarpha I had checked Leaves: Shape: Oblanceolate on strength of the Genus description. But in the TJM key:
```
   7. Lvs gen oblanceolate, ...... H. nana
   7' Lvs gen spoon-shaped, woolly or long-soft-wavy-hairy, ...... H. vestita
   
```
I had also checked spoon-shaped for H.v. ssp. c. based on the Species level description, but missed the probable implication of the Genus-key that the leaves of this Species are NOT oblanceolate shaped. (TJM Introduction says "Gen" or "Generally" before some character in a description means that (sub)Species at more junior levels have that character present unless otherwise stated at the junior level.) In this case a probable flag pointing to the TJM Genus-key is the Profile page count block for Spoon-Shaped, i.e. [13 (3:4)], compared with the simple count block [86] for Oblanceolate. So I removed Oblanceolate from the Lucid key for H.v. ssp. c. This caused the Profiles of all 3 of the other Hulsea (sub)Species to indicate splits that may be explained in the TJM Genus key.
Inflorescence: Peduncle: Length
This State is specified only for H. californica and H. vestita within the genus. In TJM, this is only in the Genus-key (not in the description - though the Conventions say it should be repeated there!!). My trial search with the Lucid key showed that I had missed scoring this for H. vestita. Having corrected this, I find that in the Genus-key
```
   1. "Lower lvs gen glandular, gen barely hairy..." ..... H. heterochroma, H. mexicana
   1' "Lower lvs gen ± woolly or long-soft-wavy-hairy"
         6. ...."Upper peduncles < 5cm" .....  H. californica
         6' .... "(or peduncles gen. > 5cm)" ..... H. vestita
    
```
In my photos, the UPPER peduncles mostly do appear to be slightly shorter than 5cm, but in the more mature-stage plants there were much longer peduncles lower down. The implication is that in using the Lucid key, one should have checked Leaves: Surface: Woolly -- since there's no TJM information on the peduncles length for H. heterochroma. and H. mexicana. (including no mention of peduncles in the Genus-description). But as the plant bloom cycle became more mature, it was more correct to say that the lower leaves were long-soft-wavy-hairy, a composite adjective I had decided was too specific to include in the Lucid key.
Experienced users of the TJM keys know that numeric specifications, particularly those measuring lengths, should be interpreted with caution -- it's not unusual for a specimen to exhibit a length that falls outside the length range given in the key. In this example, a user of the Lucid key would be well advised to check the Genus-profile page for this subFeature before entering a length obtained from the plant specimen being examined during an ID search. The lack of any numbers at all in the profile boxes corresponding to two of the four Entities in the genus might be regarded equivalent to a flag suggesting one should check the TJM Genus-key. Further review of the Lucid key may suggest that I should add a more attention-getting notation in these empty Genus-profile boxes for this and similar situations.
Phyllaries: Shape: Linear-Lanceolate
NOT (Ob)Ovate-Oblong, which I had originally included in the Lucid key, as suggested by the Hulsea Genus-description. The TJM Species descriptions for H. californica, H. heterochroma, and H. mexicana all specify "phyllaries linear-lanceolate,..." implying (according to the TJM Conventions) that this State overrides the value specified in the Genus-description.
Ray Flowers: Ligule Length (mm): 18
18 mm is the value I estimated from the photographs, but that estimate is obviously only approximate. TJM descriptions give the following values:
```
   15-20 ...... H. californica
    6-10 ...... H. heterochroma
   12-18 ...... H. mexicana
    6-10 ...... H. vestita ssp. calicarpha
    
```
Had I entered 19, or even 18.1, in Screenshot #2 above H. mexicana would have been demoted by one additional difference level.
Ray Flowers: Distinctive Characteristics: {not selected in Screenshot #2 example}
```
   Corolla tubes hairy .... H. heterochroma (TJM description), H. mexicana (TJM genus-key)
   Corollas puberulent .... H. californica (TJM description; not visible in photos)
   Corollas puberulent .... H. vestita (TJM description)
    
```
I coded this in the Lucid key as "Ligules or Tubes Hairy" for H.h. and H.m. I've altered the key to say "Corollas hairy or puberulent" for all 4 Hulsea Entities; But I coded H.c. and H.v. as "?" to suggest the user should refer to the genus-key.
Fruit: Surface: Hairy
What the TJM descriptions say:
```
   silky-hairy ...... H. californica
   very hairy ...... H. heterochroma
   sparsely hairy ...... H. mexicana
   moderately hairy ...... H. vestita (species-wide)
    
```
These fine shadings are not common or precise enough to merit inclusion in a multiple entry key. Nor is Fruit mentioned in the TJM Genus-key.