Hybrids and Phylogenetic Systematics III. Comparison with Distance Methods

Abstract

Three distance methods including two hierarchical clustering methods [average linkage (UPGMA) and Neighbor Joining, NJ] and one non-hierarchical ordination method (Multidimensional Scaling, MDS) were used to analyze data sets including taxa of known hybrid origin and their parents. Whereas goodness of fit measures indicated reduced success of MDS and parsimony analyses with increasing numbers of hybrids, neither UPGMA nor NJ analyses were affected by number of hybrids. Compared to analyses with the same number of taxa, all of which were species, goodness of fit measures indicated that all analytic methods had more difficulty handling species than hybrids. These measures are unlikely, therefore, to provide any signal of the presence of hybrids. Results were also compared in terms of placement of hybrids relative to parents and other taxa. UPGMA (and raw distance values) reliably placed many hybrids closest to one or both parents. In contrast, NJ placed hybrids in the same manner as parsimony: usually as the basal member of a lineage in~luding one or both parents along with other taxa. These results support the idea that NJ yields topologies that are similar to explicit phylogenetic methods. However, in the special case of hybrids, these topologies, in parallel with those from parsimony analyses, provide no indication of the presence or special evolutionary history of hybrids. If identification of hybrids is the goal, these results suggest that pairwise distances are more useful than any of the methods explored.

Phylogenetic Relationships and Morphological Evolution in Penstemon subg. Dasanthera (Veronicaceae)

“…Our observations are consistent with previous studies of the topological behavior of hybrids using distance methods (McDade 1997). Previous studies demonstrate that hybrids often cluster with one or both parents in hierarchical analyses (Heiser et al 1965;Bemis et al 1970; Barrett and Rhodes 1976;Schilling and Heiser 1976;McDade 1997).…”

“…Previous studies demonstrate that hybrids often cluster with one or both parents in hierarchical analyses (Heiser et al 1965;Bemis et al 1970; Barrett and Rhodes 1976;Schilling and Heiser 1976;McDade 1997). However, hybrids between distantly related species may affect clustering among other taxa in the analysis, and hybrids that share one parent often cluster together (Heiser et al 1965;Schilling and Heiser 1976).…”

Evidence for an allopolyploid complex in New ZealandPolystichum(Dryopteridaceae)

“…11) does not implicate an allopolyploid origin for the octoploid lineages: they simply appear as separate and distinct lineages. As reported by McDade (1992McDade ( , 1997, topological position alone is insufficient to identify individuals resulting from hybrid events. However, with the knowledge that some of the lineages are polyploid, the results of the AFLP analyses can be used to distinguish between the very different topological expectations of autopolyploidy and allopolyploidy, as outlined above.…”

Numerical phenetic analysis of Olinia rochetiana sensu lato (Oliniaceae)

“…It was performed by (2) 19 Surface of inflorescence axis, glabrous (0), sparsely pubescent (1), markedly pubescent (2) 20 Inflorescence unit glabrous (0), sparsely pubescent (1), markedly pubescent (2) 21 Hypanthium surface glabrous (0), sparsely pubescent (1), markedly pubescent (2) 22 Peduncle surface glabrous (0), sparsely pubescent (1), markedly pubescent (2) 23 Pedicel surface glabrous (0), sparsely pubescent (1), markedly pubescent (2) 24 Inflorescence organisation loose /sparse (0) or dense/compact (1) 25 Internodes along inflorescence axis shorter than inflorescence units (0) or longer (1) 26 Flowers respond to galling by swelling up (0) or elongating to narrow tubercles (1) 27 Leaf texture glossy, soft & smooth (0), stiff to leathery (1) or thin to papery (2) 28 Leaves clustered at base of inflorescence axes and units (0) or present at every internode (1) 29 Secondary veins curving (0) or parallel each other (1) 30 Abaxial surface of petal glabrous (0), sparsely pubescent (1), markedly pubescent (2) 31 Adaxial surface of petal glabrous (0), sparsely pubescent (1), markedly pubescent (2) 32 Style surface glabrous (0), sparsely pubescent (1), markedly pubescent (2) 33 Adaxial surface of bract glabrous (0), sparsely pubescent (1), markedly pubescent (2) 34 Abaxial surface of bract glabrous (0), sparsely pubescent (1), markedly pubescent (2) 35 Leaf discolorous (0) or concolorous (1) 36 Secondary veins visible (0) or invisible (1) on adaxial surface of leaf lamina 37 Secondary veins visible (0) or invisible (1) on abaxial surface of leaf lamina 38 Leaf margin flat (1) or sparsely in-rolled (0) 39 Bract coloration light to yellow (1) or green (0) 40 Bract texture papery (0) or leathery (1) 41 Bract caducuous (0) or persistent through anthesis (1) 42 Leaf shape elliptic (0) or not (1) 43 Leaf shape oblong (0) or not (1) 44 Leaf shape obovate (0) or not (1) 45 Leaf apex emarginated (0) or not (1) 46 Leaf apex mucronate (0) or not (1) 47 Petal shape oblong (0) or not (1) 48 Petal shape obovate (0) or not (1) 49 Petal shape spathulate (0) or not (1) 50 Petal apex truncate (0) or not (1) 51 Petal apex shortly mucronate (0) or blunt (1) 52 a Leaf shape index 53 a Number of secondary veins counted on both sides of midrib 54 a Angle of branching of secondary veins 55 a Mean areole length measured on adaxial surface 56 a Mean areole width measured on adaxial surface 57 a Mean number of veinlets per areole counted on adaxial surface 58 a Mean number of branches per veinlet measured on adaxial surface 59 a Angle of leaf apex measured on adaxial surface 60 a Angle of leaf base measured on adaxial surface Quantitative continuous characters are represented by the mean values per individual specimen measured in millimeters (mm) while quantitative discontinuous characters represent maximum values. a Characters defined as in Hickey (1973), Hill (1980) and Herman et al (1987) calculating the similarity matrix between OTUs using the average taxonomic distance coefficient from the standardised matrix, clustering the OTUs by using the unpaired group method of arithmetic averages (UPGMA), computing the cophenetic values and the cophenetic correlation using COPH and MXCOMP, respectively, to measure the distortion between the original distance matrix and the resultant phenogram (Crisci et al 1979;McDade 1997;…”