[Taxacom] Three Treasure Troves of botanical systematics

[Kenneth Kinman]

Dear All, 
     When I want information on the higher taxa of angiosperms,
I go to one of three sources first: either (1) Dallwitz et al.
(www.delta-intkey.com), especially at family level; (2) Reveal (now
available through www.plantsystematics.org); or (3) last, but not least,
is Stevens' APG website (where details of proposed clade synapomorphies
are available at one convenient and informative site). Below I will
refer to the latter as "Stevens-APG", as it contains lots of very
pertinent information to my classification of Rosopsida.  Given that
my classification will likely not be published anytime soon (if ever), I
will here at least present some more details (no "alchemy" involved) in
defense of my proposed division of Class Rosopsida (eudicots) into four
Subclasses (which are structurally the very heart of this
classification). It is reposted below for your convenience. And here is
a link to the "Stevens-APG" website:
http://www.mobot.org/MOBOT/research/APweb 
      Stevens' Gunnerales page is an excellent place to
start, since it has a lot of pertinent information (and references to
the literature), and it also notes an unfortunate change in terminology.
What I call "Core Eudicots" (and what he also formerly called "Core
Eudicots") excludes Gunnerales. But he now includes Gunnerales (to
conform to APG), so please be aware that what I still call "Core
Eudicots", he now refers to as the "Rosids et al. + Asterids et al."
clade (for information on this clade, see his Dilleniales page).
  
       My first Subclass Ranunculidae% is obviously
paraphyletic (and the % sign clearly labels it as such). Simply put, it
includes all eudicots with a primitive (plesiomorphic) floral morphology
(including that of Gunnerales). Stevens obviously recognized the
importance of this plesiomorphy, but at some point (Version 8?) the
website was changed to emphasize certain synapomorphies instead. I would
strongly encourage him to change it back, because a major, complex
plesiomorphy can sometimes easily trump a minor synapomorphy in
importance, and classification should reflect this (especially something
like floral morphology, which can be not only rather complex, but also
far more easily observed than either genes or chemicals).          
     My second Subclass Dilleniidae%% (basal "core eudicots") is
admittedly going to be more difficult to get accepted. Being doubly
paraphyletic (indicated by %%), it's going to be similar to the
difficulty of trying to defend (at least to strict cladists) a doubly
paraphyletic Class Reptilia%%. Subclass Dilleniidae is all "core
eudicots" (possessing synapomorphies of that group) which also lack the
synapomorphies of the two exgroups (Rosidae and Asteridae). Thus it is a
truncated clade. Note that it includes what is often split off as a
separate subclass Caryophyllidae, as well as parts of other proposed
subfamilies, such as Cornidae and the polyphyletic Hamamelididae.     
     My third Subclass Rosidae is a well-defined clade and
informally referred to as "Rosids" in the APG system.   See
"Stevens-APG" for proposed synapomorphies discovered thus far.   And
my fourth Subclass Asteridae (euasterids), is also informally recognized
by APG as a clade (as "Asterid I + Asterid II" or as "Euasterids I +
Euasterids II"). Whether this subclass should perhaps be expanded to
include their Cornales and Ericales as well, is a debatable option. I
presently believe euasterids is a more defensible and homogeneous clade,
but I would be open to arguments of expanding it to include Cornales and
the ericalean Orders (rather than just the euasterid clade).   
   
        In summary, I would add that my reclassification
is (in my opinion) an optimal consensus classification of four major
groupings (Subclasses). Traditionalist classifications not only
recognize too many subclasses, but they also often don't agree with one
another as to what Orders belong to each. Reveal has done a wonderful
job comparing his own consensus to other traditional classifications.
However, these traditional classifications are oversplit, confusing, and
far from being concise enough for even many professional botanists to
wrap their brains around (even more so because several competing systems
disagree so often). 
        It's no wonder APG rejected that approach, but
they then went to the other extreme and totally abandoned ranks
altogether. Therefore, between Order and Phylum level, the APG approach
is a messy and confusing conglomeration of informal clade names. It's
not really a "classification" at these important, broad levels of
interrelationships. ANYWAY, those botanists who strongly prefer to
continue operating at one of these two extremes will obviously continue
to do so. But I'm willing to bet there are plenty of botanists who would
like to see a more concise consensus classification along the lines that
I have outlined below.            
           --------Ken Kinman                   
P.S. For botanists who like their subclasses divided into superorders,
one could easily do so from my classification. For example, Rosidae
could be divided into Geranianae, Malvanae, Rosanae, Celastranae,
Oxidanae, and Violanae. By the way, Thorne and Reveal's (2007) Violanae
is almost certainly polyphyletic. Cucurbitales should be treated as a
separate Order, or better yet, they could have used the junior author's
own Superorder Cucurbitanae Reveal, 1994.

---------------------------------------------- 
Dear All, 
       I decided to update my Class Rosopsida
classification (especially since hotmail sort of "chewed up" the
classification I posted last spring). I think it is VERY important to
have such a middle-ground classification, if only to more clearly show
both the commonalities and differences between the two extremes of: (1)
the "traditional" ones, which are usually too split in my opinion, and
don't systematically store sister group information in a way that is
clear and retrievable, and (2) APG, which gets a little too overlumped
in some taxa, and sadly lacks the ranks of Class and Subclass to give
classification a balanced and cohesive structure (which "cladifications"
usually don't, especially the large ones). I attempt to combine the best
of both (cladistic and eclectic) into a single classification.  
    
     Most of the changes are in Subclass Rosidae. I have added
Order Picramniales, and I am now dividing APG's bloated "super"-Order
Malpighiales into just four separate Orders (Violales, Euphorbiales,
Podostemales, and Ochnales), coded as a polytomy (since their cladistic
relationships are still poorly known). These are the four Orders
recognized by Thorne and Reveal, 2007 (in their recent classification in
Botanical Review, Vo. 73, pp. 67-181). They do not use the name
Malpighiales, and APG does not use the name Euphorbiales, so this
thankfully minimizes confusion when those names are used. Frankly,
instead of Malpighiales sensu lato and sensu stricto, I'd now rather see
Malpighiales just disappear from use, as in Thorne and Reveal, 2007,
although I doubt that we will be that lucky anytime soon. I have also
made some other minor changes in coding to reflect updated sister group
information.                

Class Rosopsida (eudicots)              

     1 Subclass Ranunculidae%
            1  Ranunculales
            2  Sabiales
            B  Proteales
            3  Trochodendrales
            4  Buxales
            5  Gunnerales
            6  {{"Core Eudicots"}}
(Dilleniidae,etc.) 
 
  _1_ Subclass Dilleniidae%%
            1  Saxifragales
            B  Vitales
            C  {{Rosidae}}
            2  Dilleniales
            3  Santalales
            4  Berberidopsidales
            5  Caryophyllales
            B  Polygonales
            C  Nepenthales
            6  Balsaminales
            B  Polemoniales
            C  Primulales
            D  Styracales
            E  Theales
            F  Ericales
            7  Cornales
            8  {{Asteridae}}         

  _a_ Subclass Asteridae (euasterids)              
               1  Aquifoliales 
             B  Asterales
             C  Apiales
             D  Dipsacales
             2  Garryales
             3  Gentianales
             4  Lamiales
             ?  Boraginales
             5  Solanales           

  _b_ Subclass Rosidae (rosids)
             1  Geraniales
             B  Myrtales
             2  Crossosomatales
             B  Picramniales
             C  Sapindales
             D  Huerteales
             E  Malvales
             F  Brassicales
             3  Zygophyllales
             B  Fabales
             C  Rosales
             D  Fagales
             E  Cucurbitales
             4  Celastrales
             5  Oxalidales
             6  Violales
             A  Euphorbiales
             A  Podostemales
             A  Ochnales 
______________________________________ 
  NOTE: Although I now follow APG (Stevens-APG) in showing
Geraniales and Myrtales as sister groups, I still do not believe that
they are part of an exclusive "malvid" clade. I therefore continue to
show malvids as paraphyletic (two separate, but successive clades)
giving rise to the fabid clade (Zygophyllales to Ochnales). We also
differ in the exact cladistic placement of Dilleniales (not really
surprising given how much it jumps around in different phylogenetic
trees).