Supplementary information to Donoghue et al. ‘Synchrotron light shed on the nature of fossilised embryos’

Supporting data. Included as supplementary data to this manuscript are Quicktime movie files containing the reconstructed slice data on which the 3D rendered models and segmentations presented within the main manuscript were based. These files are provided as downsampled versions of the original, to provide ready access to the data. However, the original data, including both projections and reconstructed slices, are available upon request.

Supplementary Figures 1 and 2. Theseare tomographic reconstructions of the divisions between compartments of blastoderm (dark infill) in the putative germ band of Pseudooides (MESIG 20064), and marginal furrow (light infill); 1 external view; 2 internal view provided by virtual sectioning away the opposite hemisphere of the embryo thus showing the division of the germ band penetrating to depth but not fully enclosing the compartments.

SRXTM Methodology. The synchrotron-radiation X-ray Tomographic Microscopy (SRXTM) investigation was performed at the Tomography station of the Materials Science beamline of the Swiss Light Source at the Paul Scherrer Institute. The X-ray energy was optimized for maximum contrast at 17.5 keV. The magnification of the X-ray microscope ranged from 20x to 40x. On-chip binning (2x) was selected to improve the signal-to-noise ratio, resulting in isotropic voxels that ranged in dimensions from 0.7 µm to 1.4 m. Projections (1001 or 501, depending on magnification) were acquired equiangularly over 180°, were online post-processed and rearranged into flat- and darkfield-corrected sinograms, and reconstruction was performed on a 16-nodes Linux PC farm using highly optimized filtered back-projection routines. Thus, the full volumetric information was available within a few minutes of the end of the scan. Slice data derived from the scans were then analysed and manipulated using AMIRA ® < software for computed tomography on 2.5 GHz G5 Power Macintoshes with 8 GB DDR SDRAM.

Segmentation of Introvert of Markuelia. Using Amira 3D visualisation software, discrete scalids were isolated by defining their structure, with respect to 3 orthogonal slices, and interpolating between these highlighted data. The position of scalids, in rings, in specimen GMPKU 2205 of Markuelia hunanensis (fig 2a-f) was then determined by the following criteria:

  1. The three rings are positioned at different rostro-caudal depths, determined by the position at which the scalid emerges from the wall of the prepharyngeal lumen, or the position of the tip of the scalid. If the body of a scalid was positioned at a rostro-caudal level different to the surrounding scalids, it was considered to be in a different ring from the surrounding scalids.
  1. The rostro-caudal differentiation of the scalids revealed a staggered, but not discrete relative positioning of adjacent scalids. Thus, the successive rings of scalids were revealed to be intercalated. This information was then used to verify interpretation of individual scalids to their respective rings.
  1. There is some variation in the morphology of the scalids. Scalids in the third ring are slightly larger than those in the inner two rings. If intercalation, or rooting failed to give a definitive answer to the ring assignment of a scalid, size was used to assign it to the second or third ring by comparison with those scalids already assigned.

Cladistic analysis

This analysis is based on the dataset presented in ref(1), but codings for Markuelia modified in light of new information provided through SRXTM analysis. The complete, revised codings for Markuelia are as follows:

Markuelia 111011100?????????????12100?00100000--??00??1111?1?1?01???????????????11---?????00-??000

- denotes inapplicable; ? denotes missing data or unknown; & denotes polymorphic character

Search techniques

Cladistic analyses were carried out using PAUP* 4.0b10(2); branch and bound analyses were undertaken on small partitions of the dataset while heuristic searches (100 replicate random sequence addition with 10 trees retained at each step; steepest descent search option on) were utilised in analysis of larger partitions and of the complete dataset, because of computational limitations. Character evolution was resolved using PAUP* 4.0b10 and MacClade 4.05(3). All characters were unweighted and unordered in the preliminary analyses of the dataset. Gastrotricha was used as an outgroup (notionally designated paraphyletic for PAUP analyses) to root trees. Bremer support values were calculated for each of the nodes using RADCON(4), and bootstrap values were calculated using PAUP* 4.0b10 (1,000,000 ‘fast’ heuristic search option); these values are presented on the node to which they relate. Because underlying support for a phylogenetic hypothesis, as expressed through Bremer support, can be obscured by rogue taxa, the dataset was also subjected to Double Decay Analysis(5) (DDA) using RADCON, which provides a means of identifying component subtrees that exhibit support that is equal to, or greater than the phylogenetic hypotheses based on the entire dataset. The results of DDA are discussed and presented at each dataset partition.

In all instances, results from primary unweighted analyses were subjected to a posteriori reweighting, where characters are weighted in proportion to their performance in the primary analysis, as measured by the rescaled consistency index. This is undertaken as a test of the most parsimonious trees (MPTs) resulting from the primary unweighted analysis, not to reduce the number of competing MPTs. The justification for this approach is provided in the supplementary notes to ref(6).

Results and experimental analysis of the dataset

As per ref(6), the objective of the analysis was to determine the phylogenetic position of Markuelia, not a means to resolve the interrelationships of extant ecdysozoans/introvertans. Thus, an initial analysis included only extant taxa to determine whether, in the absence of fossil taxa, the resulting hypothesis of relationships concurs with contemporary hypotheses based on comparable datasets. Branch and bound analysis of the dataset yielded a single MPT (Fig. Supp1) at 124 steps (CIex [consistency index excluding uninformative characters] 0.7652; RI [retention index] 0.8200; RC [rescaled consistency index] 0.6415) in which all nodes show high statistical support. Two rounds of a posteriori reweighting yielded a single tree with the same topology (84.88333 steps; CIex 0.9337; RI 0.9530; RC 0.0.8965).

Figure Supp1: Most-parsimonious tree derived from analysis of extant taxa alone. Numeric values to the right and left of each branch are bootstrap and Bremer support values, respectively.

Inclusion of Markuelia into this analysis resolves its phylogenetic position unequivocally as a sister-taxon to Scalidophora (Fig. Supp2; 127 steps; CIex 07458; RI 0.8101; RC 0.6188). Thus, uncertainty over arthropod-nematoid-scalidophoran interrelationships is not germane to the affinity of Markuelia. Most branches are well supported by bootstrap and Bremer Support. The branch subtending Markuelia plus all other scalidophorans is among the least well supported, but this topology is fully supported by a posteriori reweighting (84.9167 steps; CIex 0.9214; RC 0.9451; RC 0.8787)

Figure Supp2: Most parsimonious tree derived from analysis of extant taxa and Markuelia hunanensis. Numeric values to the right and left of each branch are bootstrap and Bremer support values, respectively. † denotes extinct taxon.

Analysis of all taxa using the heuristic search option yields two equally most parsimonious trees that differ in the relative relationships of Markuelia and two Burgess Shale taxa, Ancalagon and Fieldia, the strict consensus of which leaves their relative relationships unresolved (194 steps;CIex 0.5608; RI 0.7340; RC 0.4200). Few of the branches are supported by bootstrap, although many are well supported by Bremer Support. Successive rounds of a posteriori reweighting yields three most parsimonious trees in which the equivocation among the trees derived from the unweighted dataset is resolved, but where the phylogenetic relationships of Cricocosmia are unresolved (Fig. Supp3; 86.06136 steps; CIex 0.8029; RI 0.8923; RC 0.7267).

Figure Supp3: Most parsimonious tree derived from analysis of all taxa. Numeric values to the right and left of each branch are bootstrap and Bremer support values, respectively. † denotes extinct taxon.

1.X. Dong, P. C. J. Donoghue, J. Cunningham, J. Liu, H. Cheng, Evolution & Development7, 468 (2005).

2.D. L. Swofford. (Sinauer Associates, Washington DC, 2002).

3.W. Maddison, D. R. Maddison. (Sinauer Associates, Inc, Sunderland, MA, 1992) pp. 404.

4.J. L. Thorley, R. D. M. Page Bioinformatics16 (2000).

5.M. Wilkinson, J. L. Thorley, P. Upchurch, Systematic Biology49, 754 (2000).

6.X. Dong, P. C. J. Donoghue, H. Cheng, J. Liu, Nature427, 237 (2004).