Untitled Document

I developed or modified several methods to improve the detection of both taxonomic and character incongruence: PABA, sliding window analysis, convergence factor c, O/E, and hypothetical species. The Partition Addition Bootstrap Alteration (PABA) approach methodically investigates alterations of bootstrap support when partitions are added and thus pinpoints both sources of conflict, taxa and partition. Using a permutuation method similar to the ILD test it is possible to assess the significance of the results. Furthermore, Wilcoxon-Signed-Rank test can reveal if a partition should be included in the analyses. The development of a program is in progress. I modified the traditional sliding window analysis to determine different groups of genetic variation within a data set in that it now approximates a Gaussian distribution. Factors or values (c, O/E) were developed to detect saturation more objectively than it is done to date. Furthermore, employing a hypothetical progenetic species I revealed the problems of assessing homology of morphological characters in the case of supposed progenetic species.

Trees reconstructed from genes may not only represent speciation events, but also processes of discord like gene duplication or horizontal gene transfer. These processes are molecular evolutionary events, but even in analyses using several concatenated genes they potentially obscure the 'true' species phylogeny. Furthermore, molecular data are often incongruent with morphological hypotheses due to these processes, but also due to problems in assessing homology of morphological features (e.g. primary absence or secondary loss). Thus, the precise detection of conflicts and their source will improve the reliability of phylogenetic hypotheses as well as shed light on the molecular evolution of genomes, gene clusters, and single genes. However, besides this taxonomic incongruence tree reconstruction might be hampered by character incongruence (e.g. saturation).

		Fig. 1. c and O/E values of nucSSU and nucLSU data sets of annelid taxa. All O/E values below 1 in combination with small c values point to saturation (red arrow). This figure is from Struck et al. (2008).

		I developed or modified several methods to improve the detection of both taxonomic and character incongruence: PABA, sliding window analysis, convergence factor c, O/E, and hypothetical species. The Partition Addition Bootstrap Alteration (PABA) approach methodically investigates alterations of bootstrap support when partitions are added and thus pinpoints both sources of conflict, taxa and partition. Using a permutuation method similar to the ILD test it is possible to assess the significance of the results. Furthermore, Wilcoxon-Signed-Rank test can reveal if a partition should be included in the analyses. The development of a program is in progress. I modified the traditional sliding window analysis to determine different groups of genetic variation within a data set in that it now approximates a Gaussian distribution. Factors or values (c, O/E) were developed to detect saturation more objectively than it is done to date. Furthermore, employing a hypothetical progenetic species I revealed the problems of assessing homology of morphological characters in the case of supposed progenetic species.
Fig. 2. Phylogeny of Eunicida using four different genes (=partitions). Taxa with ctenognath jaws were previously thought to be basal. Red boxes indicate nodes with conflicting signal by PABA (see Table 1 below). This tree is from Struck et al. (2006).

Table 1. PABA results of concatenated analysis of Fig. 2. Red indicates conflict due to either nucLSU or COI, whereas green indicates that the crucial node 5 is supported by all partitions. Node5 is crucial, because it assures that taxa with ctenognath type jaws are more derived than previously expected. This table is from Struck et al. (2006)
		Gene
		nucSSU as			mtLSU as			COI as			nucLSU as
Node	BS value	2nd	3rd	4th	2nd	3rd	4th	2nd	3rd	4th	2nd	3rd	4th
1	63	30	35	36	-3	3	1	-4	4	7	11	24	33
2	74	43	30	30	34	25	22	12	-5	4	23	8	9
3	100	88	42	n.a.	21	n.a.	n.a.	26	n.a.	n.a.	59	42	n.a.
4	79	36	31	7	6	25	12	-41	-26	-20	37	31	4
5	84	50	56	45	8	11	3	8	14	9	22	24	16
6	100	55	49	n.a.	24	3	n.a.	-12	2	n.a.	42	20	2
7	100	36	15	n.a.	12	2	n.a.	9	1	n.a.	39	7	1
8	90	22	47	61	-1	8	-2	38	40	35	-6	10	15
9	100	83	69	95	77	-1	n.a.	42	n.a.	1	-35	-39	n.a.
10	100	83	45	29	48	30	n.a.	28	9	n.a.	37	10	n.a.
11	94	25	32	22	27	18	1	20	25	13	9	24	12
12	90	59	28	4	39	23	1	13	0	-9	41	32	12
13	90	85	77	81	20	1	-3	-8	-14	-9	14	-1	-6
Average over all nodes		53	43	41	24	12	4	10	5	3	23	15	10
Average over nodes 1-3, 5-7, 10-11		51	38	32	21	13	7	11	7	8	30	20	12