Phylogeny

Family relations within the conifers

How the phylogenetic tree was found.

Theorecally it is possible to construct a phylogenetic tree from both morphologic and molecular data, but the molecur way is the only realistic possibility due to the following reasons:
  1. You can get almost as many molecular characters as you want, whereas the number of suitable morphological characters are very limited.
  2. It is almost impossible to weigh to morphological characters, whereas the problems of weighing the molecular characters are small.
  3. Using parsimony is the best choice when morphological data are used, whereas you are free to choose other assumptions when you use molecular data.

In order to get a safe tree when it was constructed in 2001 all the available informations and possibilities were included in the final family tree

  1. the 3 different genes rbcL, 18 S rDNA, and 28 S rDNA were used independant of each other.

  2. the three different algorithms maximum parsimony MP, maximum likelihood ML, and Neighbour Joining NJ were used independant of each other.
  3. the three different outgroups Angiosperm+Pteridophyte, Cycadales+Ginkgo, and Gnetales were used independant of each other.

Using 3 different genes, 3 different algorithms, and 3 different outgroups gave a total of 3x3x3 = 27 different trees. The rbcL-trees are shown in thesis3.rbcL-trees.pdf, the 18S trees are shown in thesis4.18S-trees.pdf, the 28S trees are shown in thesis5.28S-trees.pdf. The topology of all the 27 different trees were identical in relation to the conifer families except for some minor artifacts, which were partly that Cycadales+Ginkgo or Gnetales in a few of the trees were situated in the middle of the conifers and partly that the Taxaceae s.l. were not strictly united in all trees. A table containing an overview of the 27 different is shown in sum.table-of-27trees.Eng.pdf. All bases were weighted equal, i.e. no special weighting for trasitions/transversions or for first, second or third position bases, because equal weighting gave the best results.

Unfortunately most cladists (including the Scandinavian) were parsimonists in 2001, and they thought that one single perfected tree (using MP on one datamatrix with one outgroup) was much more safe than the integration of 27 trees (using 3 different genes, 3 different algorithms, and 3 different outgroups) into one sum-tree. As a matter of fact, they did not understand the common scientific principle of varying each factor per se.

Besides the above mentioned strictly systematic investigation in the thesis several other trees have been constucted varying other factors too like choice and number of taxa etc. reachíng a number of about 200 trees with different assumptions, but all reaching the same identical conifer family tree, apart from the above mentioned artifacts. As a matter of fact, just adding a few extra conifer taxa to a datamatrix changed the position of Gnetales from outside the conifers and created the artifact of Gnetales as sister to Pinaceae. A few of those trees can be seen here; from 1999: taxa-tree.rbcl.MP-str.cons.pdf ; in the report: "Phylogeny of Gymnosperms. Fact and Fiction" from 2001: phylogeny-2001.pdf; and in the report: "Phylogeny of Gymnosperms, Coniferales, Pinaceae, Taxaceae etc." from 2002: phylogeny-2002.pdf; where the different artifacts can been seen very clearly!

The above mentioned conifer family tree was confirmed by the more traditional investigation of Quinn et al. in 2002, the whole paper are shown in phylogeny-Quinn.2002.pdf, their parsimonic tree are shown in taxa-tree.str.cons.Quinn3.pdf and their "best" tree are shown in taxa-tree.rbcLmatK.Quinn4.pdf, which is reproduced from Kew Bulletin 57: 513 - 531 (2002) with kind permission of the Trustees of the Royal Botanic Gardens, Kew.