1.1-billion-year-old porphyrins establish a marine ecosystem dominated by bacterial primary producers

Significance The oceans of Earth’s middle age, 1.8–0.8 billion years ago, were devoid of animal-like life. According to one hypothesis, the emergence of large, active organisms was restrained by the limited supply of large food particles such as algae. Through the discovery of molecular fossils of the photopigment chlorophyll in 1.1-billion-year-old marine sedimentary rocks, we were able to quantify the abundance of different phototrophs. The nitrogen isotopic values of the fossil pigments showed that the oceans were dominated by cyanobacteria, while larger planktonic algae were scarce. This supports the hypothesis that small cells at the base of the food chain limited the flow of energy to higher trophic levels, potentially retarding the emergence of large and complex life. The average cell size of marine phytoplankton is critical for the flow of energy and nutrients from the base of the food web to higher trophic levels. Thus, the evolutionary succession of primary producers through Earth’s history is important for our understanding of the radiation of modern protists ∼800 million years ago and the emergence of eumetazoan animals ∼200 million years later. Currently, it is difficult to establish connections between primary production and the proliferation of large and complex organisms because the mid-Proterozoic (∼1,800–800 million years ago) rock record is nearly devoid of recognizable phytoplankton fossils. We report the discovery of intact porphyrins, the molecular fossils of chlorophylls, from 1,100-million-year-old marine black shales of the Taoudeni Basin (Mauritania), 600 million years older than previous findings. The porphyrin nitrogen isotopes (δ15Npor = 5.6–10.2‰) are heavier than in younger sedimentary sequences, and the isotopic offset between sedimentary bulk nitrogen and porphyrins (εpor = −5.1 to −0.5‰) points to cyanobacteria as dominant primary producers. Based on fossil carotenoids, anoxygenic green (Chlorobiacea) and purple sulfur bacteria (Chromatiaceae) also contributed to photosynthate. The low εpor values, in combination with a lack of diagnostic eukaryotic steranes in the time interval of 1,600–1,000 million years ago, demonstrate that algae played an insignificant role in mid-Proterozoic oceans. The paucity of algae and the small cell size of bacterial phytoplankton may have curtailed the flow of energy to higher trophic levels, potentially contributing to a diminished evolutionary pace toward complex eukaryotic ecosystems and large and active organisms.

T he succession of primary producers in the oceans shaped marine ecology through Earth's history (1). Primary producers form the base of the food web. Their cell size, elemental stoichiometry, and cell density influence the flow of energy and nutrients to higher trophic levels (2), presumably setting limits for ecosystem complexity. The composition of phytoplankton communities in Earth's early oceans, including anoxygenic and oxygenic phototrophic bacteria and eukaryotic algae, may thus have set the pace for the emergence and radiation of different groups of filter feeders, grazers, and predators, including the proliferation of modern protists ∼800 Ma and the appearance of eumetazoan animals some 200 My later (1) (Fig. 1B).
Based on the fossil record, chlorophyll c (Chl c) algae, including dinoflagellates, coccolithophorids, and diatoms, were the major energy and carbon source in the oceans of the past ∼250 My (Fig. 1B). By contrast, Paleozoic (541-251 Ma) and Ediacaran (635-541 Ma) oceans were presumably dominated by primary endosymbiotic algae (Archaeplastida), encompassing the red (Rhodophyta) and green algae (Chlorophyta) (1). Deeper yet in time, reconstructing the succession of primary producers becomes challenging. Phytoplankton without a preservable cuticle or skeleton are rarely preserved in the body fossil record, and there are no uncontentious fossils of planktonic bacteria or algae in the pre-Ediacaran (3). However, biomarkers, the molecular fossils of biological lipids, can provide complementary information about primary producers. For example, hydrocarbon fossils of carotenoid pigments extracted from sedimentary rocks have been used to detect phototrophic green (Chlorobiaceae) and purple sulfur bacteria (PSB) (Chromatiaceae) in 1,640-My-old marine ecosystems (4,5), while the concentration of eukaryotic steranes, relative to bacterial hopanes, may provide basic information about the ecological relevance of Precambrian algae (6). The relative abundance as well as diversity of steranes dramatically increased in the brief interlude between the Sturtian and Marinoan snowball Earth glaciations of 659-645 Ma, heralding the rise of planktonic algae as important primary producers in the oceans (Fig. 1A). Steranes in the time interval 645 to ∼500 Ma have a strong predominance of stigmastane (a structure with 29 carbon atoms, C 29 ), revealing Chlorophyta, a division of green algae, as dominant primary producers from the late Cryogenian to early Paleozoic ( Fig. 1) (7,8).
Preceding the Cryogenian rise of planktonic algae, the oldest clearly indigenous eukaryotic steranes appear in the geological record ∼900-800 Ma, albeit in low concentrations relative to bacterial biomarkers (7,8). These Tonian (1,000-720 Ma) steranes display a primordial distribution with a nearly 100% predominance of cholestane (C 27 , red circles in Fig. 1A), which may be related to the activity of rhodophytes (9) or heterotrophic eukaryotes (8).

Significance
The oceans of Earth's middle age, 1.8-0.8 billion years ago, were devoid of animal-like life. According to one hypothesis, the emergence of large, active organisms was restrained by the limited supply of large food particles such as algae. Through the discovery of molecular fossils of the photopigment chlorophyll in 1.1-billion-year-old marine sedimentary rocks, we were able to quantify the abundance of different phototrophs. The nitrogen isotopic values of the fossil pigments showed that the oceans were dominated by cyanobacteria, while larger planktonic algae were scarce. This supports the hypothesis that small cells at the base of the food chain limited the flow of energy to higher trophic levels, potentially retarding the emergence of large and complex life.
Going yet further back in time into the mid-Proterozoic (1,800-800 Ma), evidence for algae becomes scarce. Based on molecular clocks, the last common ancestor of all algae originated broadly between 1,900 and 1,400 Ma (10,11), predating the oldest unambiguous fossil of a eukaryotic phototroph, the ∼1,050 Ma (12) benthic red alga Bangiomorpha (13) (Fig. 1B). However, clearly indigenous biomarkers for algae have not yet been found in sedimentary rocks >900 Ma despite preservation of bacterial hopanes (Fig. 1). Whether this lack of steranes is caused by preferential degradation of algal organic matter (14) or reflects genuine paucity of eukaryotic phototrophs, remains unresolved.
Here, we present an approach to gauge the activity of bacterial and eukaryotic primary producers in the Precambrian. Based on Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS), we identify geoporphyrins in 1,100-My-old black shales. Porphyrins are the molecular fossils of (bacterio)chlorophylls. Their nitrogen isotopic composition can provide quantitative information about dominant phototrophs in past ecosystems (15). Some groups of phototrophs possess a characteristic offset (e por ) between the nitrogen isotopic composition of the whole cell and chlorophylls, independent of the nitrogen source (15), and under suitable conditions this isotopic offset is largely unaffected by degradation processes in the water column and bottom sediments (16). Therefore, the nitrogen isotopic composition of porphyrins in sediments and sedimentary rocks may preserve information about primary producers present in an ancient environment, largely independent of physical and chemical conditions such as food source and diagenesis.

Depositional Environment
Eleven black shales studied for their molecular content (Table 1 and SI Appendix, Table S1) come from the En Nesoar and Touirist formations of the 1,100 Ma El Mreïti Group (17) deposited on an epicratonic platform of the Taoudeni Basin on the West African Craton (18). The black shales, with a carbon content of up to 32%, accumulated beneath anoxic ferruginous and occasionally sulfidic waters (19) during maximal flooding of the craton in a quiet subwave base environment possibly protected by offshore stromatolite reefs (18). The black shales contain micrometer-thin laminae of organic matter and pyritized filamentous sheaths that are interpreted as benthic microbial communities, presumably of heterotrophic and/or chemosynthetic microorganisms thriving beneath anoxic waters (19), and irregularly shaped, discrete or bedding-parallel accumulations of organic particles typical of planktonic debris (SI Appendix, SI Geology and Samples).

Results
Biomarker Syngeneity. Potential contaminants introduced in the laboratory were monitored using comprehensive accumulatory system blanks, and compounds that may have permeated into rock samples were quantified and eliminated by comparing the concentration of individual hydrocarbons in the exterior and interior portions of the rock (20)(21)(22). All compounds reported here are demonstrably indigenous (SI Appendix, SI Syngeneity Assessment).
Hydrocarbon Biomarkers. All El Mreïti Group shales yielded pentacyclic terpanes, including αβ-hopanes with 27-35 carbon atoms as well as gammacerane ( Fig. 2A). A complete series of C 31 to C 36 2α-methylhopanes was also present in low concentrations. The 2α-methylhopane index, defined as the abundance of C 31 2α-methylhopane relative to C 30 αβ-hopane, was low, ranging between 0.35 and 0.77%. Although hopanes were well above detection limits, C 26 to C 30 steranes were not detected in any sample (Table 1 and SI Appendix, Fig. S1), consistent with a previous study on Atar Group shales that also failed to detect steranes (23). Based on determination of detection limits, the maximum level of steranes that may be present beneath the noise is extremely low. The ratio of maximum hypothetical C 27 steranes relative to C 27-35 hopanes (S/H max ) falls into the range of 0.0002-0.0014 (average, 0.00065; n = 11; Table 1). The El Mreïti Group shales were also devoid of the saturated carotenoid derivatives lycopane, β-carotane and γ-carotane. However, two of the most dominant biomarker classes in the aromatic hydrocarbon fraction were C 13 to C 27 2,3,6-trimethyl aryl isoprenoids (2,3,6-AI) and C 13 to C 30 2,3,4-trimethyl aryl isoprenoids (2,3,4-AI) (Fig. 2B), the degradation products of aromatic carotenoids.  Fig. 3). FT-ICR MS isoabundance-contoured plots confirm C 29 to C 32 DPEP (2) and C 29 to C 31 Etio (1) as the most abundant Ni-porphyrins, and C 32 DPEP and C 30 Etio as the highest concentrated VOporphyrin ( Fig. 4). Also identified were Ni-and VO-based di-DPEP (4, 5, and/or 6), rhodo-Etio (7), and rhodo-DPEP (8) (26). Each of these compound classes occurs as homologous series across a wide carbon number distribution with more than 40 carbon atoms ( Fig. 4A) due to side-chain length variation off core structural motifs. Although these side-chain elongated porphyrins may include bacteriochlorophyll (BChl)-derived structures (27), sample limitation prevented characterization of precise alkyl side-chain configurations (SI Appendix, SI Additional Details on Biomarker Identification and Interpretation).

Nitrogen and Carbon Isotopic Composition of Individual Ni-Porphyrins.
For the richest sample (#13), it was possible to measure the carbon and nitrogen isotopic compositions of six isolated and purified Niporphyrins (SI Appendix, Fig. S3G). Based on molecular masses for isolated compounds determined by HPLC-atmospheric pressure chemical ionization (APCI)-MS, purified porphyrins correspond to C 28 to C 31 Etio and C 30 and C 32 DPEP structures (Table 2). Measured C/N ratios of the purified porphyrins (C/N = 12.7-22.1) were higher than predicted stoichiometries (C/N = 7.0-8.0; Table  2), which suggests that isolates still contained other nonporphyrin compounds that were carried over across multiple steps of chromatographic purification (SI Appendix, Fig. S4). However, the APCI mass spectra of the isolated fractions (SI Appendix, Fig. S3 A-F) all show relatively clean molecular ions diagnostic for individual Niporphyrins and only minor foreign masses. This indicates that the contaminants are poorly ionized by APCI and likely represent hydrocarbons. The presence of nonporphyrinic nitrogen-containing compounds is unlikely in these fractions due to their higher polarity relative to porphyrins and absence of significant signals in the APCI mass spectra (see SI Appendix, SI Methods for a detailed discussion about avoidance of artifacts and contaminants). The δ 15 N por values measured for six porphyrins ranged between 5.6‰ and 10.2‰, with corresponding values of e por between −0.5 and −5.1‰ (Table 2).

Discussion
Bulk Nitrogen Isotopes. The bulk-rock nitrogen isotopic composition (δ 15 N bulk ) of the En Nesoar and Touirist Formation shales falls into the narrow range of 4.2-6.1‰ (n = 11, average = 5.1‰, Table 1) and is slightly higher than those of the Mesoproterozoic Belt Basin (−1.1-5.2‰) (28). Although δ 15 N bulk of organic-rich shales, such as those studied here with total organic carbon (TOC) between 2.1 and 31.7%, commonly closely reflect the nitrogen isotopic composition of primary bulk organic matter (29), two factors must be considered before interpreting the data: first, the contribution of inorganic nitrogen to δ 15 N bulk values, and second, isotopic fractionation effects associated with diagenesis and heterotrophic reworking of photosynthate.
Although inorganic nitrogen sources are usually negligible in organic matter-rich samples (16), the El Mreïti Group shales may contain up to 0.1% inorganic nitrogen (see intercept on N bulk axis in SI Appendix, Fig. S5) (30). However, it is plausible that the conversion of smectite to illite during burial diagenesis of the clay-rich  samples yielded substitution of potassium by ammonium (e.g., ref. 31). As the ammonium would have largely derived from the decomposition of organic matter, and as the substitution process is not coupled to major isotopic fractionation (32), δ 15 N bulk probably largely reflects the nitrogen isotopic composition of the organic matter. This is further confirmed by the absence of a correlation between TOC and δ 15 N bulk .
Secondary processes in the water column or bottom sediment may alter the original N-isotopic composition of the primary organic matter. In modern environments, aerobic heterotrophic reworking of primary organic matter in marine regions with low organic carbon flux and high oxygen exposure times increases δ 15 N bulk by +3 to +5‰ (33). However, δ 15 N bulk values of high TOC sediments do not have such positive isotopic offsets and faithfully record the N-isotopic composition of primary producers (29). Conversely, in anaerobic waters, δ 15 N bulk may decrease by 3‰ under laboratory conditions, and a depletion of 1.2‰ has been observed in anoxic sediments of Lake Lugano (34). However, in two large modern anoxic marine basins, the Black Sea and the Cariaco Trench, no negative N-isotopic offsets were observed (35,36). Moreover, a study of Black Sea sediments found that the isotopic difference between δ 15 N bulk and δ 15 N por faithfully records e por of phytoplankton (16). Thus, δ 15 N bulk of modern organic-rich shales likely reflect δ 15 N of contemporaneous phytoplankton.
Our data extend the record of e por , the offset between δ 15 N bulk and δ 15 N por , by nearly 1 Gy. Problems associated with the extrapolation of nitrogen isotopic information over such immense periods of time, into an eon with potentially different degradation processes, are discussed in detail in SI Appendix, SI Taphonomic Biases. However, even taking possible nonuniformitarian processes into account, it is likely that e por of highly organic-rich Proterozoic shales, such as those studied here, reflect the nitrogen isotopic composition of primary biomass. Moreover, even large alterations of δ 15 N bulk in the range of −3‰ to +10‰ would not impact our conclusions.
Porphyrin Nitrogen Isotopes (δ 15 N por ). The measured δ 15 N values of porphyrins (δ 15 N por ) range between +5.6‰ and +10.2‰ and are at the isotopically heavy end of data observed in modern and Phanerozoic (541 Ga to present) aquatic systems (Fig. 5). Geoporphyrins are believed to reflect the nitrogen isotopic composition of the precursor chlorophylls (δ 15 N Chl ). For example, the isotopic offset between total organic matter and porphyrins remains largely constant during burial (16), and demetallation and metal-complex formation during chlorophyll diagenesis have commonly very little effect on nitrogen isotopic fractionation (37). Thus, the isotopic composition of the Mesoproterozoic geoporphyrins likely represent δ 15 N of the precursor (bacterio) chlorophylls (SI Appendix, SI Taphonomic Biases).
Biological Origins of the Geoporphyrins. The previously oldest geoporphyrins come from the early Cambrian (38), so the structures reported here extend the record by 600 My (SI Appendix, SI Other Reports on Proterozoic Porphyrins). However, the structures of the 1,100 Ma porphyrins, when viewed in isolation, provide limited biological information. The major (bacterio) chlorophylls and their most important biogenic sources are summarized in SI Appendix, Table S2. General evidence for photosynthetic activity in the Taoudeni Basin is provided by C 30 to C 32 DPEP (2), which may originate from Chl a, b, and c, and BChl a, b, and d (39). More specifically, 7-nor DPEP (2d) may originate from Chl b and Chl c 3 (40,41), and 17-nor DPEP (2f) from Chl c (42). Chl b is an accessory pigment in higher plants, green algae, and cyanobacteria (43), although it can also be chemically generated by alteration of Chl a (44), while Chl c is found in chromophyte algae and cyanobacteria (45). Furthermore, the etio (1), butano (3), diDPEP (4, 5, 6), and rhodostructures (7, 8) may have the same biogenic sources as the corresponding DPEP compounds. Hence, the identified porphyrins do not point to any specific chloropigment source and may derive from most groups of oxygenic phototrophs plus a possible contribution from anoxygenic phototrophic bacteria.
More information can be obtained from the nitrogen isotopic composition of the porphyrins. Phototrophic organisms have a specific nitrogen isotopic offset (e por ) between total biomass and chloropigments (e por = δ 15 N biomass − δ 15 N Chl ) (Fig. 6). In culture experiments, freshwater and marine cyanobacteria, purple nonsulfur bacteria, and green and red algae were grown on different nitrogen sources (15). Despite a wide range of nitrogen isotopic compositions of bulk organic matter (δ 15 N bulk = −20 to +20‰) and (bacterio)chlorophylls (δ 15 N Chl = −30 to +25‰; SI Appendix, Fig. S6), the isotopic offset e por is quite uniform within the same group of organisms but relatively distinct between groups. e por reflects the difference in nitrogen isotopic fractionation between chlorophyll biosynthesis and the synthesis of the bulk of other nitrogen-containing metabolites, mainly proteins (16). Furthermore, e por is independent of the nitrogen source (NH 4 + , N 2 , or NO 3 − ) and its isotopic composition (15). Assimilated nitrogen species are initially converted to NH 3 , which is then incorporated into amino acids from which cellular proteins and chlorophylls are synthesized. Although enzymes converting different nitrogen species to NH 3 fractionate nitrogen differently and therefore affect the absolute value of δ 15 N cell , the isotopic difference between cellular proteins and chlorophylls remains constant within the clade. Accordingly, e por is also largely independent of redox conditions during cell growth (15,46).
The El Mreïti Group porphyrins have e por values ranging from −5.1 to −0.5‰ (Table 2) (15,(47)(48)(49)(50)(51)(52)(53). Particulate organic matter in modern oceans, recent sediments beneath anoxic waters, and Phanerozoic shales all yield positive e por values overlapping with the algal signal (Fig.  6), indicating that eukaryotic algae were the dominant export producers in these systems. By contrast, the new Mesoproterozoic e por values clearly fall into the range of cyanobacteria. However, the assignment of the Mesoproterozoic e por values to cyanobacteria has some remaining uncertainty. Most organismspecific values in Fig. 6 are based on laboratory data from a limited number of species and may not fully reflect natural conditions or variabilities. Moreover, the compilation is missing some groups of phototrophic bacteria that may have been important in the Mesoproterozoic, particularly Chloroflexi and Chlorobiaceae. Thus, the Mesoproterozoic porphyrins probably have a dominant cyanobacterial source plus an unknown contribution from anoxygenic phototrophic bacteria. Notable eukaryotic algal input, however, is unlikely based on the considerable and consistent offset between the positive e por of algaedominated Phanerozoic sediments and the negative e por of the Mesoproterozoic porphyrins.
Hydrocarbon Biomarker Evidence for Primary Producers. Carotenoid derivatives of the 2,3,4-and 2,3,6-AI series are present in all El Mreïti Group shales (Fig. 2B). 2,3,4-AI may be derived from the biological carotenoid derivatives okenane, renieratane, and renierapurpurane, which are all diagnostic for PSB of the family Chromatiaceae (4). The 2,3,6-AI series is most commonly derived from aromatic carotenoids with the cholorobactane, isorenieratane, and/or β-isorenieratane skeletons found in green sulfur bacteria (GSB) (Chlorobiaceae). As 2,3,4-and 2,3,6-AI are abundant in all Taoudeni bitumens, PSB and GSB probably played an important ecological role. While both families produce BChl a (54), some PSB also produce BChl b (55) and GSB additionally synthesize BChl c, d, or e (SI Appendix, Table S2). These BChls would all have contributed to the porphyrin pool of the extracts. However, their impact remains uncertain because e por is unknown for GSB, and only a single value is available from a lacustrine PSB species (e por = +5.8‰) (11).
The ratio of eukaryotic steranes over bacterial hopanes (S/H) is a first-order estimate for the relative contribution of eukaryotic to bacterial biomass to sediments. Although hopanes were abundant in the black shales of the El Mreïti Group, steranes remained beneath detection limits. Quantification of background noise shows that steranes, if present, fall below S/H = 0.0002-0.0014 (Table 1), which is one to two orders of magnitude lower than the Tonian average (1,000-720 Ma; S/H = 0.06 ± 0.10), and two to four orders of magnitude lower than typical Ediacaran to Phanerozoic values (635 Ma to present; average S/H = 0.75 ± 1.1) (Fig. 1A) (8).
Most molecular clocks estimate the origin of the last common ancestor of all extant eukaryotes (LECA) >1,400 Ma (e.g., ref. 10). LECA possessed all enzymes required to biosynthesize most modern common sterols (56). Thus, the absence of steranes in 1,100 Ma El Mreïti Group shales must be related to preferential heterotrophic degradation of sterols or to an ecological insignificance (or absence) of crown group eukaryotes, particularly eukaryotic algae, during deposition of the black shales. According to the "mat-seal" hypothesis, biomarkers from planktonic algae are generally underrepresented in pre-Ediacaran bitumens due to strong heterotrophic reworking of algal debris on the surface of oxygen-producing, shallow-water microbial mats (14). However, this mechanism cannot apply to the black shales of the Touirist Formation, which formed during maximum flooding of the West African Craton beneath relatively deep, anoxic waters (19). Alternatively, sterols may have suffered severe recycling as planktonic algal debris sank slowly through the water column (14). However, the preservation of cyanobacterial porphyrins also excludes this mechanism. Senescent picoplanktonic cyanobacterial cells sink considerably slower than larger algal cells (57), causing preferential degradation of cyanobacterial over algal porphyrins, a process observed in the modern Black Sea (16). Such preferential degradation of slower sinking bacterial biomass would have elevated, not attenuated, the sterane/hopane ratio. Most importantly, however, porphyrins are degradation sensitive, and their mere existence demonstrates that photosynthetic organic matter suffered little degradation during deposition of the El Mreïti Group shales. Therefore, the lack of detectable steranes is not likely related to severe and selective degradation of algal biomass, but presumably reflects primary paucity of sterol-producing eukaryotes (for further discussion, see SI Appendix, SI Taphonomic Biases).
The Phototrophic Community and the Dominant Primary Producer.
The biomarker assemblage of the 1,100 Ma El Mreïti Group shales suggests that the community of phototrophs included cyanobacteria, PSB, and GSB. Based on the absence of steranes and low e por values, eukaryotic algae did not play a detectable role. Although ornamented and process-bearing organic-walled microfossils interpreted as unambiguously eukaryotic are present in El Mreïti Group sediments (58), they do not display evidence of algal affinity. Particularly the process-bearing taxon Trachyhystrichosphaera aimika can be locally abundant in subtidal sediments, and its morphology rather suggests osmotrophy, as was proposed for older Mesoproterozoic microfossils (Tappania) (59).  Fig. S3.
† Structures tentatively assigned based on molecular weight determined by APCI. ‡ Detector response of the elemental analysis-isotope ratio mass spectrometry system (1 mV is equivalent to ∼1 ng N). Procedural  However, based on biomarker data, none of these organisms was a quantitatively important contributor of organic matter to black shale facies in the Taoudeni Basin. The biomarker distribution of the El Mreïti Group shales is typical of the mid-Proterozoic (14). All mid-Proterozoic successions, when analyzed under strict exclusion of contamination, fail to yield diagnostic eukaryotic steranes despite the presence of bacterial hopanes. This includes the 1,640 Ma Barney Creek Formation (5), 1,390 Ma Xiamaling Formation (60), 1,310 Ma Velkerri Formation (14,61,62), 1,100 Ma Nonesuch Formation (14), and 1,000 Ma Lakhanda and Ui groups (14) (Fig. 1A). Absence or low activity of eukaryotic algae thus appears to be a common attribute of mid-Proterozoic open water environments wherever biomarkers are preserved.
The mid-Proterozoic biomarker compositions are a strong contrast to younger assemblages. 900-800 Ma, the first clearly indigenous, diagnostic eukaryotic steranes emerge (6, 7) followed by a permanent transition from dominantly bacterial to eukaryotic export production 659-645 Ma (Fig. 1) (8). After 645 Ma, steranes are almost always present in marine ecosystems (8) and the N-isotopic composition of Phanerozoic porphyrins indicates a persistent dominance of algal over cyanobacterial carbon burial, even during Oceanic Anoxic Events (Fig. 6) (50).
The incumbency of bacteria through the mid-Proterozoic and the perceived late transition to a more eukaryotic world is one of the most intriguing, yet unresolved, phenomena of Precambrian biology. Explanations may include sulfide poisoning of eukaryotes in euxinic Proterozoic basin margins (63,64) and severe paucity of biolimiting nutrients such as nitrogen (63) and phosphate (8,65). However, Proterozoic marine environments, including the Taoudeni Basin (19), were largely ferruginous rather than sulfidic (66), and nutrient limitation scenarios are hard to maintain for depositional environments with extremely carbonrich sediments such as the black shales of the 1,100 Ma El Mreïti Group and 1,390 Ma Xiamaling Formation (67) with organic carbon contents >30%. Such shales likely formed beneath nutrient-replete waters (67) and not during conditions of severe nitrate or phosphate limitation (57). The exclusion of algal cells in areas of high nutrient flow is hard to explain, even in the absence of grazing pressure (2). Thus, adequate explanations for the prevalence of bacterial phytoplankton may require addition of self-sustaining ecological feedback loops, such as the shading effect of dense planktonic cyanobacterial populations, which may exclude algae with higher light requirements (57).
While the paucity of planktonic algae in the mid-Proterozoic remains enigmatic, the persistence of bacterial phytoplankton has profound ecological and evolutionary consequences. In the water column of most mid-Proterozoic marine environments, slowsinking bacterial plankton was presumably recycled within the microbial loop, leading to diminished transport of nutrients to the deep ocean and upwelling zones, and a reduced flux of organic carbon to bottom sediments (57). As permanent burial of reduced carbon is offset by a net release of molecular oxygen into the atmosphere, the sinking behavior of bacterial phytoplankton may thus have contributed to mid-Proterozoic atmospheric oxygen levels significantly lower than at present. Low atmospheric oxygen concentration and the aerobic degradation of bacterial biomass within the water column may have contributed to the pervasively anoxic nature of mid-Proterozoic marine basins (64,(68)(69)(70).
In modern oceans, relatively large eukaryotic phytoplankton cells create food webs with efficient nutrient and energy transfers to higher trophic levels, fueling the proliferation of ordersof-magnitude larger grazers and predators (2,71). By contrast, in mid-Proterozoic pelagic environments with phototrophic bacteria at the base of the food web, a large proportion of carbon and energy may have been lost within the microbial loop. Based on the fossil record (58), such trophic systems may only have sustained relatively small protists at their apex. In combination with low atmospheric oxygen levels (68,69,72) and unstable redox conditions (73), the inefficient flow of energy from the base of the mid-Proterozoic food web to higher trophic levels may have inhibited the evolution of complex eukaryotic ecosystems and large, animal-like organisms.

Materials and Methods
Extended methods are available in SI Appendix, SI Methods. Briefly, biomarker analysis was performed at the Australian National University (ANU) under standard operational procedures and included a thorough syngeneity assessment (20)(21)(22). Hydrocarbons were analyzed on a Waters AutoSpec Premier double-sector mass spectrometer interfaced with a 6890 GC (Agilent), fitted with a DB-5MS capillary column. Porphyrin determination using reversed-phase HPLC-UV/Vis followed the standard operating procedure, and used the porphyrin injection standards of Geoscience Australia (Canberra) (24,25), and was conducted at the ANU using an Agilent 1100 system containing a degasser (G1379A, JP 40714), a Quat Pump (G1311A, DE 40926343214), and a DAD (G1315B, DE40521369). Molecular characterization of porphyrins was performed at the National High Magnetic Field Laboratories in Tallahassee, Florida (26), using a 9.4-tesla FT-ICR mass spectrometer. For bulk analyses, ∼250 μg of total organic matter of decalcified rock powder was weighed into a tin capsule with 4 mg of vanadyl oxide (VO 2 O 5 , Alfa Aesar, 99.6%) and the bulk N and C content as well as their isotopic composition were measured using a Sercon 20-22 IRMS connected to an ANCA-GSL CHN elemental analyzer. The average SD measured on samples was 0.06‰. Compound-specific isotope analyses on Ni-porphyrins were performed at JAM-STEC (Japan) following the methods of Ogawa et al. (74).