Carotenoid biomarkers in Namibian shelf sediments: Anoxygenic photosynthesis during sulfide eruptions in the Benguela Upwelling System

Significance

Anoxygenic photosynthesis by phototrophic sulfur bacteria is prevalent in microbial mat ecosystems and in restricted, highly stratified aquatic environments. This limited distribution reflects their simultaneous requirements for an anoxic habitat, reduced sulfur to supply electrons for carbon fixation, and an appropriate light regime. Although these conditions were often satisfied in ancient seas, as shown by the distinctive carotenoid and chlorophyll pigments preserved in geological samples going back as far as 1.65 billion y, we can find no record of these organisms growing in today’s generally well-ventilated oceans. An array of carotenoids in sediments from the Namibian shelf suggests that green sulfur bacteria, despite their sensitivity to oxygen, can proliferate during episodic toxic gas eruptions in the Benguela Upwelling System.

Abstract

Aromatic carotenoid-derived hydrocarbon biomarkers are ubiquitous in ancient sediments and oils and are typically attributed to anoxygenic phototrophic green sulfur bacteria (GSB) and purple sulfur bacteria (PSB). These biomarkers serve as proxies for the environmental growth requirements of PSB and GSB, namely euxinic waters extending into the photic zone. Until now, prevailing models for environments supporting anoxygenic phototrophs include microbial mats, restricted basins and fjords with deep chemoclines, and meromictic lakes with shallow chemoclines. However, carotenoids have been reported in ancient open marine settings for which there currently are no known modern analogs that host GSB and PSB. The Benguela Upwelling System offshore Namibia, known for exceptionally high primary productivity, is prone to recurrent toxic gas eruptions whereupon hydrogen sulfide emanates from sediments into the overlying water column. These events, visible in satellite imagery as water masses clouded with elemental sulfur, suggest that the Benguela Upwelling System may be capable of supporting GSB and PSB. Here, we compare distributions of biomarkers in the free and sulfur-bound organic matter of Namibian shelf sediments. Numerous compounds—including acyclic isoprenoids, steranes, triterpanes, and carotenoids—were released from the polar lipid fractions upon Raney nickel desulfurization. The prevalence of isorenieratane and β-isorenieratane in sampling stations along the shelf verified anoxygenic photosynthesis by low-light-adapted, brown-colored GSB in this open marine setting. Renierapurpurane was also present in the sulfur-bound carotenoids and was typically accompanied by lower abundances of renieratane and β-renierapurpurane, thereby identifying cyanobacteria as an additional aromatic carotenoid source.

Lipid biomarkers sequestered in sediments provide valuable information about the communities that occupy the overlying water column. Terpenoid membrane lipids and photosynthetic pigments with preservable hydrocarbon cores can be recognized in rocks and oils to reveal microbial communities that lived in the geologic past. Previous studies have shown that aromatic carotenoids produced by anoxygenic phototrophic green sulfur bacteria (GSB) and purple sulfur bacteria (PSB) are particularly useful in this respect because they signify strong redox gradients in water bodies and in microbial mats. Moreover, aromatic carotenoids are notably abundant in organic matter that was deposited during mass extinction events (1, 2) and during periods of ocean anoxia (3–7). More recently, a study that examined carotenoid distribution through geologic time found that, rather than just recording significant episodes of global ocean anoxia and extreme environmental and geobiological change, sedimentary aromatic carotenoids were considerably more prevalent through time and across a diversity of ancient environments, thereby exposing the enigma that appropriate modern analogs are lacking (4).

Research on modern environments and cultured microbes reveals that the occurrence of aromatic carotenoids is determined by the microbial community composition, water chemistry, redox structure of the water column, and the prevailing light regime. The predominant microbial sources of aromatic carotenoids include some members of the Chromatiaceae (i.e., PSB) and Chlorobiaceae (i.e., GSB) (8). More recently, culture studies and the distributions of genes that code for carotenoid biosynthesis show that actinobacteria and cyanobacteria produce aromatic carotenoids (9–15). These carotenoids are also prominent in marine sponges and some other invertebrates, where it is likely they are biosynthesized by bacterial symbionts (16, 17). Phototrophic bacteria have carotenoid distributions tuned to the light regime in which they live. Some members of the Chromatiaceae use the pigment okenone as their primary aromatic carotenoid to harvest light in the range 500 to 520 nm, thereby allowing them to access light not filtered out by water or other aquatic phototrophs (18, 19). Clades of GSB produce chlorobactene, β-isorenieratene, and isorenieratene to harvest photons at the green to blue end of the light spectrum and, potentially, to protect against photobleaching or reactive oxygen species (20–24). The other essential requirements for GSB and PSB are the absence of oxygen and a supply of reduced sulfur compounds, such as hydrogen sulfide, to serve as an electron donor during photosynthesis. However, while PSB can tolerate and perhaps benefit from low concentrations of O₂, GSB are strict anaerobes (20, 25, 26).

Carotenoid occurrences in modern environments serve as analogs for interpreting fossil hydrocarbons in ancient settings. Meromictic lakes with shallow chemoclines provide examples for the type of environments where PSB and GSB flourish today and where we might expect to find okenane and chlorobactane in the fossil record (27–30). In contrast, large, restricted basins and fjords with deep chemoclines are seen as analog environments where isorenieratane would be the dominant carotenoid (31–33). Even so, modern analog environments rarely have carotenoid inventories that precisely mirror fossil distributions, and unrestricted marine environments are noticeably absent from suitable modern analogs, hampering the interpretation of GSB and PSB carotenoids in ancient marine sediments (4).

Reduced sulfur has a dual role. Not only is it an electron source for the growth of anoxygenic photosynthetic bacteria, but it also facilitates the diagenetic processes leading to carotenoid preservation. Sulfide and other reduced sulfur species serve as reducing and cross-linking agents, as has been widely demonstrated through the identification and prevalence of organo-sulfur compounds (OSC) in the geologic record (34, 35), including fossil carotenoids sequestered as OSC (36, 37). As a result, mild chemical desulfurization of polar lipid fractions can liberate an abundance of taxonomically informative hydrocarbons, allowing their analysis by gas chromatography (GC) (38, 39). Additionally, sulfide can reduce double bonds in the polyunsaturated, isoprenoid chains of carotenoids during early diagenesis in the water column and recent sediments (40). Accordingly, thorough studies of the sedimentary lipid inventory of marine settings should analyze both the free and sulfur-bound fractions to avoid missing compounds that are sequestered in the macromolecular fraction.

In the present work, we investigated the carotenoids that are preserved in sediments underlying the Benguela Upwelling System (BUS) offshore Namibia, which is a coastal marine environment noted for its high primary productivity, sulfide eruption events, and sediments rich in organic compounds, which record a wide range of lipid biomarkers for the prevailing plankton (41), including hopanoids and other compounds preserved as sulfur-bound lipids (42–44). Furthermore, satellite imagery has identified recurrent sulfur and sulfide plumes (45–47) with corresponding evidence of an active sulfur cycle (43, 48–51) and, therefore, the potential to support anoxygenic photosynthesis by GSB and PSB. Here, we report and compare the distributions of acyclic isoprenoids, steroids, triterpenoids, and carotenoids in the free and sulfur-bound hydrocarbons present in solvent extracts of the underlying sediments.

Environmental Setting and Sampling

The BUS operating offshore southern Africa is the site of a highly productive marine ecosystem. The Benguela current off the coast of South Africa and Namibia has cooled progressively, and the upwelling has intensified over tens of thousands to millions of years (52) due to changes in atmospheric and oceanic circulation patterns, as well as, possibly, uplift of the African continent (53). The area is of significant biogeochemical interest because of myriad processes linked to the nitrogen (54) and sulfur cycles (43) and to the sedimentary sequestration of organic carbon (55) and phosphorous (56).

One distinguishing feature of the BUS is the recurrent eruptions of hydrogen sulfide on the inner shelf (47). During these eruptions, hydrogen sulfide is released from the organic-rich, anoxic sediments into the overlying water column and atmosphere, causing mass die-offs of marine life, including fish and lobster, and odors and corrosion in coastal communities (45, 46, 48, 50, 57). These events last on the order of days to weeks according to satellite images that show the turquoise milky waters generated from the oxidation of hydrogen sulfide to insoluble elemental sulfur and polysulfide (45, 46). The water column can become anoxic and sulfidic at depths as shallow as 10 to 50 m (43, 46, 57). Sulfide eruptions occur most commonly during austral summer and are stronger in some years than others (45, 46, 50). The sulfide eruptions on the Namibian shelf represent a prime natural laboratory to test whether GSB and PSB can proliferate in the water column of a modern open-ocean environment, thereby contributing to sulfur cycling and providing a modern analog for the deposition of GSB and PSB pigments in fossilized ancient marine sediments.

Sediment samples were collected in March 2014 during a cruise of the research vessel Mirabilis belonging to the Namibian Ministry of Fisheries and Marine Resources as part of the 2014 Regional Graduate Network for Oceanography Discovery Camp (58). Samples were collected from regularly monitored stations between Walvis Bay and Luderitz at the 23°, 24°, 25°, 26°, and 27°S latitudinal transects in water depths ranging from 30 to 260 m. Some previously published data from these sites include information on water depths and depth of the photic zone (59) (Fig. 1 and SI Appendix, Table S1). Samples include sediment cores, core tops, and grab samples; the latter were collected when coring was not possible. A large proportion of recovered samples, and particularly those from between 23° and 25°S, smelled strongly of sulfide. Whole cores collected for pigment analyses were sectioned into 2- to 5-cm slices on deck after recovery. They were placed into labeled plastic zip-lock bags, frozen at –20 °C in the dark, and maintained in this state during transport and storage at the Massachusetts Institute of Technology.

Fig. 1.

Map of sample stations in the BUS offshore Namibia. The regularly monitored stations that were visited during the March 2014 cruise are marked by circles; the red filled circles represent the stations where samples were collected, and the black filled stations represent stations where samples were not collected. The stations are numbered in the order in which they were sampled. The base MERIS RGB image shows the distribution of sulfur plumes (green color) in quasi-true colors as they were on February 20, 2014 from MODIS Rapid Response data (NASA). Other publications (47) report historical data for even more intense plumes in the region of the inner-shelf stations 4 and 10. The outer-shelf stations, including stations 6, 7, and 8, are characterized by lower levels and sulfide and higher concentrations of oxygen in the water column (e.g., ref. 50).

Lipids were analyzed in two phases. Shortly after the initial sampling, pigments were extracted and analyzed together with a range of reference lipids extracted from eukaryotic and bacterial phototrophs and two saline lake sediments. A selection of samples, focusing just on surface sediment (0 to 4 cm), was reanalyzed in 2019 and 2020 with the specific purpose of comparing the distributions of free and sulfur-bound neutral lipids (for details, see SI Appendix, Table S1).

Results and Discussion

Contrasting Patterns of Free and Sulfur-Bound Hydrocarbons in Benguela Sediments.

The array of lipids detected in the nonpolar (NP) fractions of the total lipid extracts of Benguela sediments included n-alkanes dominated by C₂₅–C₃₁ molecules with an odd over even preference, C₂₇–C₂₉ sterenes, small amounts of hopanes dominated by 17α (H),21β (H) isomers, along with mixtures of C₂₀ and C₂₅ OSC (Fig. 2A). In contrast, an abundance of distinctive biomarker hydrocarbons was released from the polar fractions after the Raney nickel treatment (RN-NP). These were dominated by C₂₀ and C₂₅ acyclic isoprenoids, steranes, triterpanes, and carotenoids (Fig. 2B). Thus, a large proportion of the extractable material was either chemically reduced by the Raney nickel treatment (60) or composed of naturally sulfurized unsaturated and functionalized compounds. No attempt was made to characterize these OSC further because unambiguous identification is difficult in the absence of authentic standards.

Fig. 2.

Full-scan GC-MS total ion chromatograms of the nonpolar fraction from station 4 (A, before Raney nickel) and new NP fractions released upon Raney nickel desulfurization from the polar fraction (B, after Raney nickel) of the solvent extract. Identification of C₂₅ and C₃₀ HBI (haslenes and rhizenes, respectively) is tentative in the absence of an authentic standard. IS denotes the ai-C₂₂ alkane internal standard. Procedural blanks confirmed the purity of all reagents and solvents used in this process.

The compositions of the NP hydrocarbon fractions were striking in that they contained mixtures of unsaturated triterpenoids together with geological and biological isomers for steranes and triterpanes (Figs. 2–4 and Table 1). Such combinations of compounds are typically not found together unless fossil carbon compounds are admixed with contemporary materials. For example, in Fig. 3 A–F, which compares steroids and triterpenoids from the sample from station 4, we observe that the ααα (20R) biological isomer was dominant among the C₂₇–C₂₉ steranes, but there were significant abundances of the ααα (20S), αββ (20S), and αββ (20R) geological isomers together with the 20S and 20R βα-diasterane isomers (Fig. 3 A and B). Additionally, the steroid distributions in both NP and RN-NP included some sterenes (e.g., Figs. 2 and 3C). In the triterpanes, the C₂₉ and C₃₀ hopanes each showed the presence of αβ, βα, and ββ isomers but with different proportions of these isomers for each carbon number (Fig. 3 D and E). Some of the free hydrocarbon fractions showed significant concentrations of C₃₁ 22S geological isomers (Fig. 3F), with the resultant peculiarity that the typical thermal maturity parameters varied widely across the sample set as has been reported for other high organic sulfur systems (61, 62) (Table 1). The prevalence of sterane and hopane hydrocarbons in the NP fraction of the lipid extracts is due to rapid sulfurization and desulfurization reactions that convert functionalized sterol and bacteriohopanepolyol (BHP) precursors to their stable hydrocarbon cores (63). However, the presence of geological isomers—particularly diasteranes, αββ, and 20S steranes—is incompatible with the cooccurrence of sterenes, ββ-hopanes, and other unsaturated lipids, indicating that ancient, reworked organic matter is contributing to the biomarker inventory. Similar observations have been described for sediments of the Lawrentian fan in the Labrador Sea, where ancient shale-derived organics are mixed with contemporary material (64–66).

Fig. 4.

(A–H) GC-QQQ-MS reaction chromatogram comparisons of the aromatic carotenoid distributions from four stations (4, 10, 7, and 9) in the NP fraction (before Raney nickel) and new NP fractions released upon Raney nickel desulfurization of the polar fraction (after Raney nickel) of the solvent extract. Blue lines: 552→134 reaction chromatograms; green lines: 546→134 reaction chromatograms. Note the high signal/noise ratios (S/N) evident in the traces B and D compared to A and C as one indicator of the amount of carotenoid released upon desulfurization at these sites. Based on the distribution of sulfur plumes from MERIS RGB images (Fig. 1) and other published data (47), the inner-shelf stations 4 and 10 were within areas subject to regular and intense sulfur plumes, while stations 6, 7, and 9 were not. The RN-NP fractions of all Namibian shelf sampling station sediments contained isorenieratane (iso) and β-isorenieratane (β-iso) attributable to GSB with lower relative abundances of renieratane (ren), renierapurpurane (rnp), and β-renierapurpurane (β-rnp) attributable to cyanobacteria. The asymmetry evident in the chromatographic peaks of these carotenoids are features consistent with thermal immaturity (81). Anomalously, paleorenieratane (pal) was found in stations 6, 7, 9, and 10 after desulfurization but the relative abundances of these peaks are typically minor compared to those in the free hydrocarbon fractions. Critically, the paleorenieratane peaks lack the asymmetry evident in the other diaromatic carotenoids. These lines of evidence suggest that this signal is due to reworked ancient organic matter although an origin from a contemporary biological source of paleorenieratene cannot be ruled out.

Table 1.

A comparison of free and sulfurized biomarker ratios for the samples from different Namibian shelf sampling stations

	Samples	Maturity		Source			Environment
		Dia/Sterane	C31H S/S+R	S/H	C27/C27-C29	C29/C27-C29	Ga	MTTCI	C35 HHI
NP before Raney nickel	Station 2	0.28	0.50	0.46	0.40	0.35	0.02	0.88	3.93
	Station 4	0.16	0.20	0.16	0.49	0.27	0.01	0.89	0.75
	Station 5	0.36	0.49	0.30	0.38	0.38	0.01	0.88	2.30
	Station 6	0.36	0.52	0.33	0.40	0.37	0.01	0.89	2.97
	Station 7	0.40	0.54	0.34	0.40	0.37	0.02	0.89	2.83
	Station 9	0.34	0.55	0.63	0.43	0.33	0.02	0.92	5.43
	Station 10	0.20	0.47	0.47	0.50	0.28	0.02	0.91	2.02
	Station 10-grab	0.18	0.51	0.51	0.51	0.27	0.02	0.91	1.69
	Station X	0.41	0.54	0.72	0.41	0.34	0.03	0.89	6.09
NP after Raney nickel	Station 2	0.00	0.19	98.29	0.65	0.15	0.28	0.89	2.69
	Station 4	0.00	0.08	22.23	0.58	0.18	0.32	0.90	13.87
	Station 5	0.00	0.23	43.19	0.60	0.17	0.09	0.92	3.81
	Station 6	0.00	0.41	18.36	0.66	0.15	0.07	0.94	4.28
	Station 7	0.00	0.47	44.62	0.67	0.11	0.36	0.83	4.66
	Station 9	0.00	0.45	46.80	0.66	0.13	0.20	0.82	3.70
	Station 10	0.00	0.19	222.6	0.47	0.18	2.10	0.90	5.57
	Station 10-grab	0.00	0.09	148.0	0.46	0.21	2.14	0.93	8.93
	Station X	0.00	0.15	82.00	0.41	0.25	1.12	0.88	8.97
Average	Before Raney Ni	0.30	0.48	0.43	0.44	0.33	0.02	0.90	3.01
	After Raney Ni	0.00	0.25	80.69	0.57	0.17	0.74	0.89	6.28
	P values	<0.01	<0.01	<0.01	<0.01	<0.01	0.02	0.58	0.03

C₂₇/C₂₇-C₂₉, C₂₇ steranes/C_27–29 steranes; C₂₉/C₂₇-C₂₉, C₂₉ steranes/C₂₇-₂₉ steranes; C₃₁H S/S+R, C₃₁ αβ 22S homohopane/(C₃₁ αβ 22S homohopane +C₃₁ αβ 22R homohopane); Dia/Sterane, C_27–29 diasteranes/ C_27–29 (diasteranes+regular steranes); Ga, Gammacerane/C₃₀ αβ-hopane; MTTCI, 5,7,8-triMe-MTTC/(8-Me-MTTC+5,8-diMe-MTTC+7,8-diMe-MTTC); S/H, C_27–29 (diasteranes +regular steranes)/ C_27–35 hopanes. For steranes: diasteranes = βα(20S)-steranes + βα(20R)-steranes + αβ(20S)-steranes + αβ(20R)-steranes; regular steranes = 5α14α17α(20S)-steranes + 5α14β17β(20R)-steranes + 5α14β17β(20S)-steranes + 5α14α17α(20R)-steranes. For hopanes: C_27–35 hopanes = C₂₇ 17α 22,29,30-trisnorhopane + C₂₇ 17β 22,29,30-trisnorneohopane + C₂₈ 28,30-dinorhopane + C₂₈ 29,30-dinorhopane + C₂₉ 17α21β-hopane + C₂₉ 17β21α-hopane + C₂₉ 17β21β-hopane + C₃₀ 17α21β-hopane + C₃₀ 17β21α-hopane + C₃₀ 17β21β-hopane + C_31–35 homohopanes; homohopanes = 17α21β(22S)-homohopane + 17α21β(22R)-homohopane +17β21β-homohopane. For MTTCs, 8-Me-MTTC = 2,8-dimethyl-2-(4,8,12-trimethyltridecyl) chroman, 5,8-diMe-MTTC = 2,5,8-trimethyl-2-(4,8,12-trimethyltridecyl) chroman; 7,8-diMe-MTTC = 2,7,8-trimethyl-2-(4,8,12-trimethyltridecyl) chroman; 5,7,8-triMe-MTTC = 2,5,7,8-tetramethyl-2-(4,8,12-trimethyltridecyl) chroman.

Fig. 3.

GC-QQQ-MS reaction chromatogram comparisons of steranes, hopanes, and methylated MTTCS in the NP fraction (before Raney nickel) and new NP fractions released upon Raney nickel desulfurization from the polar fraction (after Raney nickel) of the solvent extract. (A) C₂₇ steranes, (B) C₂₈ steranes, (C) C₂₉ steranes, (D) C₂₉ hopanes, (E) C₃₀ hopanes and gammacerane, (F) C₃₁ homohopanes, (G–I) Mono-, di-, and trimethylated 2-methyl-2-(4,8,12-trimethyltridecyl) chromans (methylated MTTCs). Before Raney nickel: NP fractions from free lipid extracts of Benguela sediments in station 4; after Raney nickel: new NP fractions after Raney nickel reactions of polar fractions from free lipid extracts. The peak height shows the relative quantity of the specified biomarker. For steranes, βαS = βα(20S)-diasteranes, βαR = βα(20R)-diasteranes, αββR = 5α14β17β(20R)-steranes, αββS = 5α14β17β(20S)-steranes, βαα = 5β14α17α-steranes, αααR = 5α14α17α(20R)-steranes, αααS = 5α14α17α(20S)-steranes. For hopanes, αβ = 17α21β-hopanes, βαS = 17β21α(22S)-hopanes, βαR = 17β21α(22R)-hopanes. ββ = 17β21β-hopanes, Ga = gammacerane. For MTTCs, 8-Me-MTTC = 2,8-dimethyl-2-(4,8,12-trimethyltridecyl) chroman, 5,8-diMe-MTTC = 2,5,8-trimethyl-2-(4,8,12-trimethyltridecyl) chroman; 7,8-diMe-MTTC = 2,7,8-trimethyl-2-(4,8,12-trimethyltridecyl) chroman; 5,7,8-triMe-MTTC = 2,5,7,8-tetramethyl-2-(4,8,12-trimethyltridecyl) chroman. Methanol (MeOH) was used as the solvent during the Raney nickel desulfurization reactions (SI Appendix, Fig. S1).

Biomarker distribution patterns of the RN-NP fractions were dramatically different because the compounds released by the Raney nickel treatment came to dominate the assemblages (Figs. 2 and 3). The most abundant components were phytane, together with some isomeric phytenes, which are diagenetic products of phytol. The C₂₅ highly branched isoprenoid (HBI) alkane, hazlane, derived from C₂₅ haslene precursors that are prevalent in Haslea spp. diatoms (67, 68), was also a prominent hydrocarbon. Among the polycyclic biomarkers, the compounds released by desulfurization showed a major shift to biological isomer predominance in all compound classes. For example, the geological diasterane isomers were totally absent. The steranes largely comprise just two isomers, βαα 20R and ααα 20R, with the former being absent from the free hydrocarbon fractions. Moreover, the proportion of C₂₇ steranes among C₂₇–C₂₉ counterparts increased (0.44 vs. 0.57 on average, P < 0.01) (Table 1) after the treatment, while C₂₉ steranes decreased (0.33 vs. 0.17, P < 0.01) (Table 1). For the hopanes, the ββ biological stereoisomers were dominant for both the free and bound homohopanes. In addition, the sterane/hopane ratios shifted from values less than 1 in the NP hydrocarbons to values higher than 200 in the RN-NP inventory (Table 1), indicating that sterols were preferentially sulfurized compared with hopanoids and subsequently released during desulfurization.

Trapped and hidden by natural sulfurization, S-bound biomarkers are often overlooked in paleoenvironmental assessments. In addition to steranes and hopanes, gammacerane, the diagenetic product of tetrahymanol and a widely applied stratification indicator (69–72), was not observed in most of the NP hydrocarbon fractions, but it was released from all samples upon desulfurization (Fig. 3E). The elevated gammacerane index (0.74 vs. 0.02, P = 0.02) (Table 1), following the release of sequestered gammacerane from the macromolecular fractions, reveals a strong redox stratification. Methylated methyltrimethyltridecylchromans (MTTCs), despite elusive biological sources, have been proposed as salinity proxies in the marine surface layer (73–75). Mono-, di-, and trimethyl MTTCs were present in both NP and RN-NP fractions with very similar relative abundances overall (Fig. 3G). Differences in MTTC Index (MTTCI) were minor (0.9 vs. 0.89, P = 0.58) (Table 1), which is consistent with this being a normal salinity marine environment. The C₃₅ homohopanes, which are derived from C₃₅ bacteriohopanetetrol and other extended BHPs, are generally preserved as thiophene hopanoids and hopane sulfides in anoxic conditions (61, 76, 77). The fraction of C₃₅ homohopanes among C₃₁–C₃₅ counterparts is known as the homohopane index (HHI). High values (6.28 on average, up to 13.87) (Table 1) for this proxy are consistent with the availability of hydrogen sulfide in the water column and sediment porewaters of the BUS (47), leading to selective preservation of the hydrocarbon cores of BHP precursors.

Sulfurized Pigments Reveal the Existence of GSB in an Open-Ocean Setting.

Despite intense green colors and organic richness, few unaltered pigments were evident in the high-performance liquid chromatography mass spectrometry (HPLC-MS) analyses of Namibian shelf sediment extracts compared with a suite of reference samples, each of which contained complex mixtures of intact carotenoids and chlorophylls together with their degradation products (see SI Appendix, Table S2 for details). The only identifiable carotenoid in the Namibian shelf sediment extracts was β-carotene, which cooccurs with degradation products of chlorophyll a, including pheophytins a and b and pyropheophytins a and b, in most samples and with small amounts of chlorophyll a and its epimer in the surface-grab sample from station 6 at 26°S. These remnants of algal and cyanobacterial pigments were the only compounds that could be positively identified in the analyses of tetrapyrroles and intact carotenoids. The analyses of pigments in the reference microbial mat, sediment, and biomass samples verified the effectiveness of the analytical protocol, which was based on previously published methods (78–80) and shown to be able to identify bacterial and algal carotenoids, as well as chlorophylls and a variety of their degradation products. Thus, HPLC-MS analysis of intact pigments and their polar degradation products in the free (i.e., nonsulfur-bound) lipid extracts revealed only pigments diagnostic for oxygenic photosynthesis in the BUS, while failing to provide evidence of anoxygenic photosynthetic GSB or PSB.

In GC-tandem mass spectrometry (GC-MS/MS) analyses, carotenoid diagenetic products—including isorenieratane, β-isorenieratane, and paleorenieratane attributable to GSB—were detectable at very low abundances in the NP fractions of free lipid extracts (Fig. 4 and Table 2). The traces of paleorenieratane in the 546→134 reaction chromatograms, particularly from stations 2 and 9, suggest a fossil hydrocarbon origin for these compounds. Paleorenieratane is most commonly reported in Paleozoic and Mesozoic samples (4), and its biological precursor has not yet been identified in any modern environment. Notably, the peak for paleorenieratane is symmetrical while peaks for isorenieratane and β-isorenieratane are asymmetric, consisting of a major peak with a shoulder on the trailing edge.

Table 2.

A comparison of free and sulfurized biomarker ratios of carotenoids for the different Benguela sampling stations

	Samples	pal	iso	ren	rnp	β-pal	β-iso	β-ren	C38	C39	β-ren /β-iso	ren+rnp /iso
NP before Raney nickel	Station 2	0.30	0.41	0.00	0.00	0.62	0.30	0.00	0.00	0.00	0.00	0.00
	Station 4	0.10	0.41	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
	Station 5	0.15	0.26	0.00	0.00	0.00	0.20	0.00	0.00	0.00	0.00	0.00
	Station 6	0.16	0.24	0.00	0.00	0.00	0.19	0.00	0.00	0.00	0.00	0.00
	Station 7	0.14	0.29	0.00	0.00	0.00	0.24	0.00	0.00	0.00	0.00	0.00
	Station 9	0.26	0.33	0.00	0.00	0.14	0.23	0.00	0.00	0.00	0.00	0.00
	Station 10	0.18	0.24	0.00	0.00	0.00	0.23	0.00	0.00	0.00	0.00	0.00
	Station 10-grab	0.00	1.65	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
	Station X	0.22	0.23	0.00	0.00	0.00	0.30	0.00	0.00	0.00	0.00	0.00
NP after Raney nickel	Station 2	0.00	0.80	0.00	0.17	0.00	0.18	0.03	0.00	0.00	0.12	0.17
	Station 4	0.00	5.40	0.53	0.34	0.00	3.14	0.24	0.02	0.02	0.07	0.14
	Station 5	0.00	13.26	0.73	0.60	0.00	6.78	0.17	0.02	0.11	0.02	0.09
	Station 6	0.12	2.01	0.23	0.20	0.00	1.17	0.17	0.03	0.03	0.12	0.18
	Station 7	0.21	0.50	0.00	0.04	0.00	0.34	0.10	0.00	0.00	0.16	0.12
	Station 9	0.13	0.48	0.02	0.05	0.00	0.38	0.05	0.00	0.00	0.11	0.14
	Station 10	0.09	1.06	0.05	0.20	0.00	0.32	0.12	0.00	0.00	0.23	0.22
	Station 10-grab	0.00	1.10	0.03	0.60	0.00	0.18	0.06	0.00	0.00	0.24	0.37
	Station X	0.00	3.24	0.80	0.22	0.00	1.70	0.30	0.00	0.03	0.15	0.24
Average	Before Raney Ni	0.15	0.46	0.00	0.00	0.02	0.17	0.00	0.00	0.00	0.00	0.00
	After Raney Ni	0.06	3.10	0.27	0.27	0.00	1.58	0.14	0.01	0.05	0.14	0.18
	P values	0.01	0.07	0.03	<0.01	0.24	0.07	<0.01	0.06	0.19	<0.01	<0.01

All carotenoids were normalized to β-carotane. Abbreviations: C₃₈, C₃₈ diaromatic carotenoids; C₃₉, C₃₉ diaromatic carotenoids; iso, isorenieratane; pal, paleorenieratane; ren, renieratane; rnp, renierapurpurane; β-iso, β-isorenieratane; β-pal, β-paleorenieratane; β-ren, β-renierapurpurane.

However, following treatment of the polar fractions with Raney nickel, diverse and distinctive inventories of hydrogenated carotenoids were released. The predominant compound in all samples was still β-carotane, along with some partially hydrogenated counterparts (Fig. 3 D and E and Table 2), with the aromatic carotenoids isorenieratane and β-isorenieratane also prominent (Fig. 4). Less common C₄₀ carotenoids—including renieratane, renierapurpurane, and β-renierapurpurane—were also detected in lower abundances together with minor C₃₉ and C₃₈ carotenoid-related structures at several stations (Table 2). However, the monoaromatic C₄₀ carotenoid chlorobactane from green-colored GSB, which typically are found in shallower aquatic niches or in mats, compared with the isorenieratene-producing, brown-colored GSB, was not detected. Similarly, okenane derived from the C₄₀ carotenoid of some PSB was not detected. The release of unsaturated carotenes after the Raney nickel treatment, attributable to the presence of double bonds that had been shielded from sulfurization and reduction, strongly suggests that this hydrocarbon assemblage is recent and indigenous as opposed to reworked fossil material. Renieratane, renierapurpurane, and β-renierapurpurane, which were identified in the 546→134 and 552→134 reaction chromatograms, were released from the sulfur-bound fraction of samples from all stations but were most prominent at stations 2 and 10. Furthermore, a late-eluting shoulder on the chromatographic peak for all C₄₀ aromatic carotenoids distinguishes them from fossil C₄₀ aromatic carotenoids that have experienced geologic burial. The presence of this anomaly in chromatographic behavior contrasts with the smooth and symmetrical chromatographic peaks that are typically observed for the C₄₀ aromatic carotenoids in thermally mature ancient sediments and oils. The late-eluting shoulders on the C₄₀ aromatic carotenoids, which were also observed in the very immature lacustrine sediments of the Green River Formation (81), are likely due to later-eluting stereoisomers generated during reduction of double bonds in the polyunsaturated isoprenoidal chain by abiogenic reaction with sulfide (40), which only isomerize and merge into the primary peak over long timescales or at higher thermal maturities. Their presence, therefore, is a compelling signal of their contemporary origin as opposed to being carotenoids from ancient and reworked organic matter. In contrast, the symmetrical peak for paleorenieratane in the NP fractions suggests that this component is derived from ancient reworked organic matter.

The contrasting compositions of sulfur-bound carotenoids present in the Namibian shelf sediment samples provide additional insights to their contemporary biological origins. The strong correlation of S-bound β-isorenieratane and isorenieratane (R² > 0.99) (SI Appendix, Fig. S2A) across all samples indicates a common origin from low-light-adapted, brown-colored GSB, which is also consistent with their cooccurrence in cultured taxa (82) and a shared pathway to their biosynthesis (13). Similarly, there is also positive but weaker correlation of renieratane, renierapurpurane, and β-renierapurpurane (R² = 0.65) (SI Appendix, Fig. S2B) in the desulfurized polar lipids. Although renieratene and renierapurpurin were originally isolated from sponges (83, 84), these carotenes were suggested to originate from PSB (30) because of their 1-alkyl-2,3,4-trimethyl substitution pattern that is similar to that of okenone and the cooccurrence of their fossil hydrocarbon skeletons with okenane in the 1.65-Ga sediments of the Barney Creek Formation (85). However, renierapurpurin is also an intermediate in the biosynthetic pathway to the polar carotenoid synechoxanthin, which is prevalent among cyanobacteria (14, 15). A significant peak for renierapurpurane in the data for the RN-NP fraction from station 10 (Fig. 4F), compared with the sample data from the other stations, indicates that cyanobacteria are a possible source in this case (86). While synechoxanthin is outside our analytical window due to its dual carboxyl functions, traces of C₃₈ and C₃₉ carotenoid diagenetic products are present in stations 4, 5, 6, and X. These compounds are the reduction and decarboxylation products of synechoxanthin and its monocarboxylic acid counterpart (86), and their low relative abundances are a result of the Namibian shelf sediments having not experienced the geothermal heating necessary to drive the decarboxylation reaction (Table 2 and SI Appendix, Fig. S3). Overall, the aromatic carotenoids released in different relative abundances among the sampling stations by the Raney nickel treatment of these sediments is consistent with natural sulfurization and diagenesis of pigments predominantly derived from GSB with secondary input from cyanobacteria.

The “no modern analog” problem, which refers to the absence of modern, unrestricted marine environments that host GSB and PSB and their associated pigments in the water column, contributes uncertainty to the interpretation of these aromatic carotenoids in ancient marine sediments (4, 58). Because hydrogen sulfide rarely accumulates in the photic zone of contemporary, open-ocean water columns, the interpretation of these fossilized pigments in marine sedimentary archives is largely based on modern lacustrine systems and enclosed euxinic basins, such as the Black Sea (33, 87, 88). Without a modern analog, there has been no direct evidence that GSB and PSB could thrive in an unrestricted marine water column when the atmosphere is well oxygenated, as it has been for much of the Phanerozoic. This is particularly problematic for interpreting chlorobactane and okenane, which are often encountered in ancient marine sediments. Without modern examples of GSB and PSB proliferating in an open marine water column, other factors, such as basin restriction, microbial mat sources, or sediment transport, have been invoked to explain their occurrence in Phanerozoic marine sedimentary archives.

The data at hand are insufficient to establish whether the GSB pigments identified here originated from planktonic GSB that were active in the water column, as opposed to originating from microbial mats at the sediment–water interface. Although both are feasible, we favor the possibility of a water column source. First, low-light–adapted planktonic GSB proliferate in the water column at the ∼100-m redoxcline of the Black Sea (21, 87, 89, 90) and at 115 to 120 m in the deep, ferruginous waters of Lake Matano, Indonesia (91), both of which are significantly deeper than the anoxic, sulfidic water column depths observed on the Namibian shelf (43, 46, 57). Second, the recurrent sulfide plumes that occur on the Namibian inner shelf provide the essential ingredients of anoxia and reduced sulfur species to support the growth of GSB in the water column. Third, highest abundances of GSB carotenoids were observed at the inner-shelf stations 4 and 10, where satellite data indicate sulfide plumes have been intense and seasonally recurrent and where they correspond to periods when ventilation of bottom waters is reduced (47).

Although there are multiple lines of evidence for active sulfide metabolism in the strong upwelling zone off West Africa and the oxygen minimum zone offshore Peru and Chile (92, 93), current studies provide limited evidence for sulfide-dependent anoxygenic photosynthesis in these systems. However, molecular data provide some tangible clues. A metaproteomic study of waters in the coastal BUS revealed sequences attributed to Bacteroidetes/Chlorobi among identifiable taxa (94), although this is a very broad grouping and lacking in the detail needed to categorically identify isorenieratene-producing Chlorobi. Additionally, metagenomic and metatranscriptomic studies suggest that Chlorobi may be associated with the Peruvian oxygen minimum zone, where sulfide plumes have also been observed (95, 96). On the Namibian shelf, metagenomic analysis of a sulfur-oxidizing mat community did not reveal the presence of any GSB (97). On the other hand, chemolithotrophic bacterioplankton, comprising populations of γ- and ε-proteobacteria, have been shown to oxidize sulfide with nitrate, thereby detoxifying subsurface waters and creating a buffer zone between these and surface waters populated by vulnerable metazoa (50). Further resolution of these issues will require samples that capture an active sulfide eruption event.

Here, we have demonstrated that although pigments originating from oxygenic phototrophs, including algae and cyanobacteria, are predominant in the sediments (Fig. 2), carotenoid biomarkers derived from the low-light-adapted, brown-colored GSB, namely isorenieratane and β-isorenieratane, have been sequestered in sulfurized sedimentary organic matter underlying the BUS. These results imply that anoxygenic photosynthetic bacteria are important contemporary participants in the sulfur cycle of the Namibian shelf and that this environment is an appropriate yet previously unrecognized analog for the occurrence of GSB biomarkers in ancient open-ocean settings. The short-lived sulfide eruptions on the Namibian shelf may promote growth of GSB in the water column, suggesting that these transient events may be integral, if not required, for interpreting GSB biomarkers in ancient marine sediments according to an open-ocean BUS model. Finally, the transient and seasonal nature of the sulfide eruptions and growth of GSB in the water column may explain conflicting evidence for photic zone euxinia in the sedimentary record that cooccurs with evidence of oxic conditions (98).

Summary

A limited array of hydrocarbon biomarkers was detected in the free lipid extracts of sediments from the BUS, some of which could be ascribed to fossil hydrocarbon sources. In contrast, biomarkers were released in abundance from the polar fraction after Raney nickel desulfurization. These hydrocarbons—which included n-alkanes, acyclic isoprenoids, steranes, triterpanes, and carotenoids—were evidently from contemporaneous biogenic sources and were sequestered in polar lipids as organic sulfur compounds.

The strong correlations of sulfur-bound β-isorenieratane and isorenieratane (R² > 0.9) and a positive correlation of sulfur-bound renieratane, renierapurpurane and β-renierapurpurane (R² > 0.6) in Benguela sediments indicate carotenoid inputs from GSB and cyanobacteria, respectively.

The occurrence of sulfurized isorenieratane and β-isorenieratane, originating from GSB, suggests that sulfide released episodically into the water column from sediments fuels anoxygenic photosynthesis in the sunlit surface waters. The detection of biomarkers for GSB in a contemporary open-ocean upwelling system identifies an analog environment that accounts for the prevalence of phototrophic sulfur bacteria and photic zone euxinia in marine geologic records. Transient sulfide eruptions, therefore, may be integral to interpreting GSB biomarkers in the rock record and may explain apparently conflicting evidence for oxia and extreme euxinia in the same sedimentary horizons.

Materials and Methods

Lipid Extraction and Fractionation.

In 2014, a total of 14 samples from the regularly monitored stations were lyophilized and extracted under low-light conditions for pigment analysis by HPLC-MS (SI Appendix, Table S2) (58). Freeze-dried sediment was weighed into a clean centrifuge tube and extracted by sonication with acetone in an ice bath while simultaneously protecting the extract from light. The samples were sonicated for 15-min intervals, after which they were centrifuged. The supernatant was decanted and collected as the pigment extract. The process was repeated seven times for each sample after which the extract was still colored, showing that complete extraction was not possible with this solvent alone. The combined acetone extracts were concentrated at room temperature under a gentle stream of N₂ and passed through a filter to remove particulates, evaporated to dryness, and weighed, after which a 4- to 6-mg aliquot was transferred to an amber-combusted glass insert and was brought up to 200 μL with acetone for injection onto the HPLC-MS.

Subsequently, in 2019, a subset of eight samples was selected for detailed hydrocarbon biomarker analysis. For these, fresh extracts were made of the archived and freeze-dried sediment. Approximately 7 g of sample was extracted with dichloromethane:methanol (DCM:MeOH; 9:1 [v:v]), using a Dionex Accelerator Solvent Extractor 350 (ASE), operated at 100 °C. Extracts were subsequently redissolved in ∼1 mL hexane (Hex) to precipitate asphaltenes, which were removed after centrifugation. Elemental sulfur was trapped by the addition of activated copper shot, which was then removed by filtration. The resultant extracts were dried to <100 µL and fractionated using a silica gel column. The solvents Hex:DCM (4:1 [v:v]) and DCM:MeOH (4:1 [v:v]) were used to elute the NP and polar fractions, sequentially. The NP fractions were subjected to GC-MS and GC-MS/MS analyses while the polar fractions were desulfurized by reaction with Raney nickel. In this process, polar lipids dissolved in 1 mL methanol were mixed with 20 mL of a Raney nickel slurry in MeOH and heated to 80 °C under N₂ reflux for 2 h. MeOH was used as the solvent during the Raney nickel reactions (SI Appendix, Fig. S1) since tests of alternative solvents revealed that MeOH was devoid of hydrocarbon contaminants. The products were extracted into DCM with sonication (10 mL, four times) and the residual nickel sedimented by centrifugation. The Raney nickel-treated hydrocarbons were then separated (as above) on a silica gel column to prepare the RN-NP and polar components with the former then being analyzed by GC-MS and GC-MS/MS.

Instrument Measurements and Compound Identification.

HPLC-MS analyses, conducted using published methods with minor modifications (58, 78–80), were performed on an Agilent 1200 series HPLC coupled to an Agilent 6520 Q-TOF MS, which was operated in auto MS/MS mode where a maximum of three precursors were selected for MS/MS fragmentation per cycle. The HPLC was equipped with an Agilent Poroshell 120 EC-C18 column (2.1 × 150 mm, 2.7 AM) column with a precolumn (EC-C18 2.1 × 5 mm; 2.7 Vm) using a solvent flow rate of 0.4 mL/min and gradient of 20% solvent B (0 min) to 100% solvent B (28.5 min), hold 100% solvent B (48.5 min), return to 20% B (50 min), finally reconditioning at 20% B until a final time of 60 min where A was 80:20 (v:v) methanol:0.05 M ammonium acetate and B was 80:10:10 (v:v:v) methanol:acetonitrile:ethyl acetate. The MS was operated in APCI positive-ion mode, the nebulizer was set to 40 psig with gas and vaporizer temperatures at 300 °C and 400 °C, respectively. The drying gas flow rate was 6 L/min. The corona current was 5 VA with a capillary voltage of 2,000 V.

The NP and RN-NP fractions were analyzed by GC-MS and GC-MS/MS after addition of an aliquot of 3-methylheneicosane (ai-C₂₂; 2.2 μg) internal standard. For GC-MS, samples were injected into an Agilent 6890A GC interfaced with an Agilent 5975C (MS). The inlet of the GC-MS was equipped with a split/splitless injector, used in splitless mode and held at a constant temperature of 300 °C. The GC was fitted with a J&W DB-5MS fused silica capillary column (60 m × 0.25 mm × 0.25 µm) programmed with the following parameters: 60 °C injection and hold for 2 min, ramp at 10 °C/min to 140 °C, followed directly by a ramp at 3.5 °C/min to 320 °C, and a final isothermal hold at 320 °C for 33.6 min. The MS source was set at 230 °C, the quadrupole was set to 150 °C, and the MS was operated in electron impact mode at 70 eV. The scan range was set from 50 to 580 Da.

GC-MS/MS was conducted for high sensitivity, targeted-compound analysis using a 7890B Agilent GC coupled to a 7010A Agilent triple quadrupole MS (GC-QQQ-MS) operated in multiple reaction monitoring modes. The GC was equipped with a multimode injector at an initial injection temperature of 45 °C, which was ramped at a rate of 720 °C/min to 340 °C. Chromatography was conducted on a column similar to that described above with the GC oven temperature held isothermally at 40 °C for 2 min, ramped to 320 °C at a rate of 4°/min, and then held at this temperature for 22 min. The transfer line and source temperatures were set at 300 °C and 250 °C, respectively. The electron energy was set at 70 eV. All biomarker data were processed using MassHunter QQQ software. Each compound was identified and integrated under multiple reaction monitoring mode within a narrow retention time window (0.5 min). NP analytes, including n-alkanes, acyclic isoprenoids, steranes, hopanes, MTTCs, and carotenoids, were identified on the basis of their full scan spectra, diagnostic reaction chromatograms, and relative retention times in comparison with rock standards and previously reported chromatograms (4, 30, 81, 85).

Data Availability

All study data are included in the article and SI Appendix.