Supplementary Information for Phylogenetic analyses of ray-finned fishes (Actinopterygii) using collagen type I protein sequences

Ray-finned fishes (Actinopterygii) are the largest and most diverse group
of vertebrates, comprising over half of all living vertebrate species.
Phylogenetic relationships between ray-finned fishes have historically
pivoted on the study of morphology, which has notoriously failed to
resolve higher-order relationships, such as within the percomorphs. More
recently, comprehensive genomic analyses have provided further resolution
of actinopterygian phylogeny, including higher-order relationships. Such
analyses are rightfully regarded as the ‘gold standard’ for phylogenetics.
However, DNA retrieval requires modern or well-preserved tissue and is
less likely to be preserved in archaeological or fossil specimens. In
contrast some proteins, such as collagen, are phylogenetically informative
and can survive into deep time. Here, we test the utility of collagen type
I amino acid sequences for phylogenetic estimation of ray-finned fishes.
We estimate topology using Bayesian approaches and compare the congruence
of our estimated trees with published genomic phylogenies. Furthermore, we
apply a Bayesian molecular clock approach and compare estimated divergence
dates with previously published genomic clock analyses. Our
collagen-derived trees exhibit 77% of node positions as congruent with
recent genomic-derived trees, with the majority of discrepancies occurring
in higher-order node positions, almost exclusively within the Percomorpha.
Our molecular clock trees present divergence times that are fairly
comparable with genomic-based phylogenetic analyses. We estimate the mean
node age of Actinopteri at ~293 million years (Ma), the base of Teleostei
at ~211 Ma and the radiation of percomorphs beginning at ~141 Ma (~350 Ma,
~250–283 Ma and ~120–133 Ma in genomic trees, respectively). Finally, we
show that the average rate of collagen (I) sequence evolution is 0.9 amino
acid substitutions for every million years of divergence, with the α3 (I)
sequence evolving the fastest, followed by the α2 (I) chain. This is the
quickest rate known for any vertebrate group. We demonstrate that
phylogenetic analyses using collagen type I amino acid sequences generate
tangible signals for actinopterygians that are highly congruent with
recent genomic-level studies. However, there is limited congruence within
percomorphs, perhaps due to clade-specific functional constraints acting
upon collagen sequences. Our results provide important insights for future
phylogenetic analyses incorporating extinct actinopterygian species via
collagen (I) sequencing.