Straw broke "Ericales" back (eudicot classification)

[Ken Kinman]

Dear All,
     Oh mercy, what have I got myself into now?  I thought the rosids were bad, but the "dilleniids" are even worse.  Unfortunately the APG appears to have just swept the problem under two big rugs, "Caryophyllales" and "Ericales" (the latter appears to be even worse than their "Malpighiales").  Initially I just wanted to place Mitrastemonaceae in their "Ericalean" grouping, and it fell apart like a house cards (well actually more like a camel that needs to be broken into two or more parts).
      Put simply, the APG's Caryophyllales and Ericales are two big taxonomic "black holes" (wastebaskets if you want to put it another way).  If one separates off some of each, you get two intermediate groupings that are basically the old Dilleniid mess (but without all the unnecessary rosid stuff mixed in).  So here is my present classification of Class Rosopsidea (eudicots), actually here I'll call it Rosopsida for those who don't care for my standardized endings.  :-)  Clade 7 are the "non-core" Caryophyllales of APG, and "Clade" 8 is a split up Ericalean grouping that is probably paraphyletic at best.  Anyway, I have revived a number of traditional orders that got sucked into these molecular "black holes", which really bridge the gap between the core (traditional) Caryophyllales and the "Theales-Ericales" group which seems to be near the rosid-asterid divergence.
     Actually this is probably still too cladistically simplistic to be correct, but it is a step in the right direction in my opinion.  It's definitely a step back from APG's overlumping, but a small enough step that it shouldn't negate what contributions they have made.  Needless to say, trying to treat this "dilleniid" mess continues to be a challenge, and although I don't believe APG meant to be sweeping it under big taxonomic rugs, that is nevertheless what I believe has happened.  Too much faith was initially placed on rbcL (rubisco large subunit) sequences, just as some bacteriologists initially placed too much faith in rRNA sequences alone.  But we live and learn, and I am still in the process of learning more about this "dilleniid grade".  The following classification is definitely a work in progress (notes on Rafflesiales below*):

Class Rosopsida (eudicots)
        1  Ranunculales
        2  Proteales
        B  Sabiales
        3  Buxales
        B  Trochodendrales
        4  Gunnerales
        5  Caryophyllales (core only)
        6  Physenales
           Polygonales
           Nepenthales
        7  Primulales
           Ebenales
           Styracales
           Polemoniales
           Balsaminales
           Theales
           Ericales
        8  Cornales
        B  Aquifoliales
        b  Asterales
        c  Apiales
        d  Dipsacales
        C  Garryales
        D  Gentianales
        E  Solanales
        ?  Boraginales
        F  Lamiales
        9  Dilleniales?
           Berberidopsidales
           Santalales?
           Saxifragales
           Vitales
       10  Fabales
           Rosales
           Fagales
           Cucurbitales
       11  Zygophyllales
           Geraniales
           Crossosomatales
       12  Hypericales
           Malpighiales
           Euphorbiales
           Violales
       13  Linales
           Oxidales
           Celastrales
       14  Myrtales
       15  Malvales
        ?  Rafflesiales
       16  Sapindales
       17  Brassicales
---------------------------------
P.S.  The Rafflesiales remain incertae sedis near Malvales (following Nickrent, 2002).  Barkman et al.'s (2004) placement in the "Malpighialean grouping" (basically clade 12 above) seems premature, as I have state before (more to be published on that by others who know more about that group than I do).  However, I am still very tempted to follow Barkman et al.'s placement of Family Mitrastemonaceae in the Ericalean grouping (either within Theales or between Theales and Ericales sensu stricto).  And I am talking about a very restricted Theales (not the old bloated one with all those rosids in it).  Whether Dilleniales is near the base of rosids or more in the middle of the dilleniid "mess" is unclear (is it really near Vitales??).  Likewise, Santalales remains a problem, although it could go in either direction (my educated hunch is that it will be a rosid).  This is all giving me a royal headache, so I am call it quits for today.  :-)
            ---- Cheers,
                      Ken Kinman
__________________________________