A molecular timescale for the origin of red algal-derived plastids

In modern oceans, eukaryotic phytoplankton is dominated by lineages with red algal-derived plastids such as diatoms, dinoflagellates, and coccolithophores. These lineages and countless others representing a huge diversity of forms and lifestyles all belong to four algal groups: cryptophytes, ochrophytes, haptophytes, and myzozoans. Despite the ecological importance of these groups, we still lack a comprehensive understanding of their evolution and how they obtained their plastids. Over the last years, new hypotheses have emerged to explain the acquisition of red algal-derived plastids by serial endosymbiosis, but the chronology of these putative independent plastid acquisitions remains untested. Here, we have established a timeframe for the origin of red algal-derived plastids under scenarios of serial endosymbiosis, using a taxon- and gene-rich phylogenomic dataset combined to Bayesian molecular clock analyses. We find that the hypotheses of serial endosymbiosis are chronologically possible, as the stem lineages of all red plastid-containing groups overlapped in time. This period in the Meso- and Neoproterozoic Eras set the stage for the later expansion to dominance of red algal-derived primary production in the contemporary oceans, which has profoundly altered the global geochemical and ecological conditions of the Earth.

Archaeplastida in the Paleoproterozoic Era, between 2.1-1.6 bya20. The origin of red algae has 87 been estimated to be in the late Mesoproterozoic to early Neoproterozoic (1.3-0.9 bya)20, after a 88 relatively long lag following the emergence of Archaeplastida. However, an earlier appearance in  In this study, we combined phylogenomics and molecular clock analyses to investigate 105 the chronology of the origin and spread of complex red plastids among distantly-related 106 eukaryote lineages in order to test the general rhodoplex hypothesis15. We assembled a broad

116
The phylogeny of eukaryotes. Molecular clock analyses rely on robust tree topologies. To 117 obtain our reference topology, we derived two sub-datasets from the full dataset of 320 protein-  suggesting that these topological differences derived from the chosen evolutionary model and not 147 from the use of Bayesian or ML inference ( Supplementary Fig. 5 (Table 1), and we followed a conservative approach when interpreting the 160 fossil record by choosing only fossils widely accepted by the paleontological community. We 161 also included one generally uncontested biomarker to constrain the emergence of extant Metazoa 162 in order to expand the otherwise sparse Proterozoic fossil record (24-isopropyl cholestane)29. Our 163 analyses placed the root of the eukaryote tree well into the Paleoproterozoic Era (Fig. 2). This Era 164 also saw the origin of primary plastids in the common ancestor of Archaeplastida, which likely 165 took place between 2,137 and 1,807 mya. Crown group red algae were inferred in the 166 Paleoproterozoic between 1,984 and 1,732 mya. These age ranges are provided as a conservative 167 approach encompassing all performed analyses (with the exception of those using t-cauchy 168 distributions with long tails, see below).

169
From red algae, plastids then spread to distantly-related groups of eukaryotes by at least 170 one secondary endosymbiosis. Plastid phylogenies have consistently shown that this red algal 171 donor lineage belonged to a stem lineage of Rhodophytina30, i.e., it lived after the split of 172 Cyanidiophyceae and the rest of red algae (Fig. 2). Based on Fig. 2, we inferred a time range for 173 this donor lineage to be between 1,675 and 1,281 mya. The origination period for the lineages 174 currently harbouring red algal-derived plastids were inferred as follows: cryptophytes between 175 1,658 and 440 mya; ochrophytes between 1,298 and 622 mya; haptophytes between 1,943 and 176 579 mya; myzozoans between 1,520 and 696 mya (dates refer to the 95% HPD intervals in Fig.   177 2). Thus, the stems of all extant lineages containing red algal plastids-along which these plastids

226
The eukaryote tree of life and the rhodoplex hypothesis. In the last 15 years, the tree of 227 eukaryotes has been extensively rearranged based on phylogenomics. We assembled a dataset 228 containing a dense eukaryotic-wide taxon sampling with 733 taxa and analysed subsamples of it 229 with a variety of mixture models to resolve several important uncertainties that are key for 230 understanding plastid evolution. Notably, we recovered the monophyly of Archaeplastida, which 231 has previously been strongly supported by plastid evidencee.g.,11,32 but not by host (nuclear) 232 phylogenetic markerse.g., 33-36. This topology is consistent with a recent exhaustive phylogenomic includes the red algal plastid-containing cryptophytes, was placed as sister to Archaeplastida. The association of Cryptista and Archaeplastida has been suggested before, often even disrupting the 238 monophyly of Archaeplastida35,36, but our analyses robustly recovered the sister relationship of 239 these two major eukaryote clades. Another contentious placement has concerned the supergroup 240 Haptista26,35,36,39, which includes the red algal plastid-containing haptophytes, as well as its 241 possible relationship to the orphan lineage Ancoracysta twista40. The best-fitting catgtr model 242 favoured the position of Haptista as sister to TSAR and placed A. twista deeper in the tree, 243 unrelated to Haptista. We observed that the model (catgtr vs. lgc60) rather than the method or 244 program is determinant in placing A. twista relative to Haptista-lgc60 always placed these two 245 lineages together, both in ML and Bayesian, but catgtr never did-and that variation in the taxon-246 sampling did not drastically modify this association. Proterocladus antiquus in ~1,000 Ma old rocks, taken as evidence for a much earlier appearance 281 of multicellularity in this group of green algae44, is in line with our results (Fig. 2). Similarly, 282 multicellular organic-walled microfossils with affinity to fungi were recently reported in the 1-283 0.9 Ga old Grassy Bay Formation45, which pushes back the emergence of fungi by 500 Ma 284 compared to previous studies46. In our analyses, fungi were estimated to originate even before 285 (1,759 to 1,078 Ma), consistently with the presence of multicellular organisms around 1 bya.

286
More generally, an early Paleoproterozoic origin of eukaryotes would also be in line with records 287 of aggregative multicellularity appearing more than 2 bya, although the eukaryote affiliation of 288 these fossils is debated47,48.  (Fig. 2), because the other models assumed the 307 monophyly of haptophytes with cryptophytes and the acquisition of plastids in an ancestor of that 308 group. In both compatible models, the secondary engulfment of a red alga by a stem cryptophyte 309 can be constrained by the minimum age of the plastid donor lineage, i.e., the age of the last 310 Rhodophytina common ancestor (Fig. 2). This would indicate a rather early plastid acquisition by