Reproducibility of results in phylogenetic analysis of mollusks: a reanalysis of the Taylor, Kantor, and Sysoev (1993) data set for conoidean gastropods
G. Rosenberg, Reproducibility of results in phylogenetic analysis of mollusks: a reanalysis of the Taylor, Kantor, and Sysoev (1993) data set for conoidean gastropods, AM MALAC B, 14(2), 1998, pp. 219-228
Reanalysis of the Taylor, Kantor, and Sysoev (1993) data set on conoidean g
astropods failed to reproduce their results. Taylor er al. found more than
900 trees of length 189; reanalysis yielded 32,700 trees of length 187. The
number of trees they found was limited by the memory available on the comp
uter used for the analysis. The Taylor rt al. consensus tree omitted the st
ated outgroup Benthobia (Pseudolividae); reanalysis including the outgroup
yielded 3,149 trees of length 193, in all of which Benthobia fell within th
e ingroup. Strict and majority-rule consensus trees differed considerably i
n topology from those with Benthobia excluded. Reanalysis excluding the hyp
othetical ancestor, whose character states Taylor et al. determined in part
by ingroup analysis, yielded additional topologies of consensus trees. Onl
y eight of 38 clades in the Taylor er al. tree appeared in all three strict
consensus trees; 17 clades were not supported by any of the majority rule
consensus trees. All three majority-rule consensus trees did support the tr
ansfer to Conidae by Taylor er nl. of the turrid subfamilies Clathurellinae
, Conorbinae, Oenopotinae, Mangeliinae, Daphnellinae, and Taraninae. This c
lade, however, did not appear in two of the strict consensus trees, so supp
ort for it is equivocal.
Additional problems with the analysis include incorrect character mappings,
use of characters primarily from one organ system, conflicts between text
and data matrix, choice of taxa, and inclusion of data from taxa not includ
ed in the cladistic analysis in formulating the classification. The Taylor
er al. data set does nor support strong inferences about conoidean phylogen
y, and there is not yet convincing evidence for abandoning the traditional
classification of the group. Nonetheless, their data are an immensely valua
ble contribution to be built on as information about conoidean taxa, charac
ters, organ systems, and outgroups accumulates.