{\rtf1\ansi\ansicpg1252\cocoartf1504
{\fonttbl\f0\fswiss\fcharset0 Helvetica;}
{\colortbl;\red255\green255\blue255;\red0\green0\blue10;}
{\*\expandedcolortbl;\csgray\c100000;\csgenericrgb\c0\c0\c3922;}
\paperw11900\paperh16840\margl1440\margr1440\vieww17480\viewh18340\viewkind0
\pard\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx6480\tx7200\tx7920\tx8640\pardirnatural\partightenfactor0
\f0\b\fs24 \cf0 Supplementary Figures\
\pard\tx720\tx1440\tx2160\tx2880\tx3600\tx4320\tx5040\tx5760\tx6480\tx7200\tx7920\tx8640\pardirnatural\partightenfactor0
\b0 \cf0 \
Supplementary Figure 1. G1PDH full tree, 111 sequences, 190 positions, inferred under LG+C60 model. Root position inferred using relaxed uncorrelated lognormal clock model with a Yule prior and LG substitution model. MAD root indicated with asterisk. \
\
Supplementary Figure 2. GGGPS full tree, 133 sequences, 129 positions, inferred under LG+C40 model. Root position inferred using relaxed uncorrelated lognormal clock model with a Yule prior and LG substitution model. MAD root indicated with asterisk. \
\
Supplementary Figure 3. GGGPS large subclade, 98 sequences, 166 positions, inferred under LG+C40 model. Root position inferred using relaxed uncorrelated lognormal clock model with a Yule prior and LG substitution model. MAD root indicated with asterisk. \
\
Supplementary Figure 4. DGGGPS full tree, 177 sequences, 97 positions, inferred under LG+C50 model. Root position inferred using relaxed uncorrelated lognormal clock model with a Yule prior and LG substitution model. MAD root indicated with asterisk. \
\
Supplementary Figure 5. GpsA full tree, 84 sequences, 169 positions, inferred under LG+C60 model. Root position inferred using relaxed uncorrelated lognormal clock model with a Yule prior and LG substitution model. MAD root indicated with asterisk. \
\
Supplementary Figure 6. Glp full tree, 51 sequences, 199 positions, inferred under LG+C40 model. Root position inferred using relaxed uncorrelated lognormal clock model with a Yule prior and LG substitution model. MAD root indicated with asterisk. \
\
Supplementary Figure 7. GlpK full tree, 77 sequences, 363 positions, inferred under LG+C60 model. Root position inferred using relaxed uncorrelated lognormal clock model with a Yule prior and LG substitution model. MAD root indicated with asterisk. \
\
Supplementary Figure 8. PlsC full tree, 74 sequences, 57 positions, inferred under LG+C50 model. Root position inferred using relaxed uncorrelated lognormal clock model with a Yule prior and LG substitution model. MAD root between clade comprising of sequences in red and other sequences. \
\
Supplementary Figure 9. PlsY full tree, 60 sequences, 104 positions, inferred under LG+C50 model. Root position inferred using relaxed uncorrelated lognormal clock model with a Yule prior and LG substitution model. MAD root indicated with asterisk. \
\
Supplementary Figure 10. G1PDH full tree, 123 sequences, 173 positions, inferred under LG+C60 model. Root position inferred using 3-dehydroquinate synthase\cf2 (DHQS), five glucerol dehydrogenase (GDH) and five alcohol dehydrogenase (ALDH) sequences as an outgroup \
\
\cf0 Supplementary Figure 11. Glp full tree, 63 sequences, 183 positions, inferred under LG+C60 model. Root position inferred using \cf2 12 \cf0 FAD-dependent oxidoreductase sequences \cf2 as an outgroup \
\
\cf0 Supplementary Figure 12. Gpsa full tree, 96 sequences, 148 positions, inferred under LG+C60 model. Root position inferred using \cf2 six \cf0 hydroxyacyl-CoA dehydrogenase\cf2 (HACDH) and 6\cf0 UDP-glucose 6-dehydrogenase\cf2 (UDPGDH) sequences as an outgroup \
\
Supplementary Figure 13. Gpsa full tree with eukaryotic sequences, 113 sequences, 159 positions, inferred under LG+C50 model.\
\
Supplementary Figure 14. Glp full tree with eukaryotic sequences, 80 sequences, 190 positions, inferred under LG+C50 model.\
\
Supplementary Figure 15. PlsC full tree with eukaryotic sequences, 96 sequences, 54 positions, inferred under LG+C50 model.\
\
\cf0 Supplementary Figure 16. Unrooted G1PDH full tree, 111 sequences, 190 positions, inferred under LG+C60 model. \
\
Supplementary Figure 17. Unrooted GGGPS full tree, 133 sequences, 129 positions, inferred under LG+C40 model. \
\
Supplementary Figure 18. Unrooted DGGGPS full tree, 177 sequences, 97 positions, inferred under LG+C50 model. \
\
Supplementary Figure 19. Unrooted GpsA full tree, 84 sequences, 169 positions, inferred under LG+C60 model. \
\
Supplementary Figure 20. Unrooted Glp full tree, 51 sequences, 199 positions, inferred under LG+C40 model. \
\
Supplementary Figure 21. Unrooted GlpK full tree, 77 sequences, 363 positions, inferred under LG+C60 model. \
\
Supplementary Figure 22. Unrooted PlsC full tree, 74 sequences, 57 positions, inferred under LG+C50 model. \
\
Supplementary Figure 23. Unrooted PlsY full tree, 60 sequences, 104 positions, inferred under LG+C50 model \
\
Supplementary Figure 24. Gpsa full tree, 96 sequences, 148 positions, inferred under LG+C50 model. Root position inferred using \cf2 six \cf0 hydroxyacyl-CoA dehydrogenase\cf2 (HACDH) and 6\cf0 UDP-glucose 6-dehydrogenase\cf2 (UDPGDH) sequences as an outgroup \
\cf0 \
\
\cf2 Supplementary Table 1. Highest marginal posterior probabilities for each gene tree for both the molecular clock and MAD rooting methods.\
\
Supplementary Table 2. Comparison of Bayesian Information Criterion (BIC) scores for the models selected by IQTree compared to the model used by Yokobori et al. (LG+gamma) for the outgrip rooting analyses. Lower BIC scores indicate better fit of model to data. \
\
Supplementary Table 3. Accession numbers for sequences used in phylogenetic analyses. \
\pard\pardeftab720\ri-340\partightenfactor0
\cf2 \
}