Report

The Rise of the Rhizosolenid Diatoms

See allHide authors and affiliations

Science  23 Apr 2004:
Vol. 304, Issue 5670, pp. 584-587
DOI: 10.1126/science.1096806

Abstract

The 18S ribosomal DNA molecular phylogeny and lipid composition of over 120 marine diatoms showed that the capability to biosynthesize highly branched isoprenoid (HBI) alkenes is restricted to two specific phylogenetic clusters, which independently evolved in centric and pennate diatoms. The molecular record of C25 HBI chemical fossils in a large suite of well-dated marine sediments and petroleum revealed that the older cluster, composed of rhizosolenid diatoms, evolved 91.5 ± 1.5 million years ago (Upper Turonian), enabling an accurate dating of the pace of diatom evolution that is unprecedented. The rapid rise of the rhizosolenid diatoms probably resulted from a major reorganization of the nutrient budget in the mid-Cretaceous oceans, triggered by plate tectonics.

In modern oceans, diatoms are the most important group of phytoplankton, responsible for almost half of the marine primary productivity (1, 2), but their ecological dominance occurred relatively recently in geological time (3). The fossil record of marine diatom tests reveals that diatoms only evolved in the Jurassic and became more common in the mid-Cretaceous (4). This indicates that in the past 100 million years (Ma) or so there has been a substantial shift in the relative importance of the different phytoplankton groups, with enormous consequences for the biogeochemical cycling of elements (e.g., carbon, nitrogen, phosphorus, and silicate) in the ocean (3). Unfortunately, the fossil record of diatoms is punctuated and not very reliable, because the silica tests of diatoms are prone to dissolution during early diagenesis. Therefore, our present knowledge of the advent of the diatoms and the consequences for biogeochemical cycling is rather limited. Conservative 18S rDNA gene sequences of extant diatom species can be used as an alternative tool to reconstruct the rise of the diatoms (5). However, for proper interpretation, the molecular clock rate needs to be calibrated against the often-incomplete fossil diatom record.

An alternative approach to reconstructing the evolution of the diatoms is to use molecular fossils: characteristic organic components that may survive for hundred of millions of years in sediments and even in petroleum. Probably the most specific group of diatom markers are the highly branched isoprenoid (HBI) alkenes. These components differ in terms of biosynthesis from most common acyclic and cyclic isoprenoid natural products because their skeletons are characterized by a distinctive “T branch” (Fig. 1). In the C25 HBI alkenes, the T branch is formed by attachment of a C10 isoprenoid unit to a C15 isoprenoid unit at C-7, most likely requiring an unusual set of enzymes. In the C30 HBI alkenes, a C15 instead of a C10 isoprenoid unit is used. C25 and C30 HBI alkenes have been identified in the diatom genera Rhizosolenia, Haslea, Navicula, and Pleurosigma (611), but not in any other organisms.

Fig. 1.

Neighbor-joining phylogenetic tree based on nearly complete 18S rRNA sequences of diatoms. Some of the sequences were published before (5); 86 others (see table S1 for details) were determined in this study. The sequences of Coccoid haptophyte and Emiliania huxleyi were used as outgroups but were pruned from the tree. Bolidomonas mediterranea is a sister group of the diatoms. The tree was created with the use of the Jukes Cantor model. HBI-biosynthesizing strains are indicated in red. Diatoms in green were tested but did not contain HBI alkenes; diatoms in black were not tested for the presence of HBI alkenes. The scale bar indicates 10% sequence variation. The inset shows the structure of C25 HBI alkane (27) and parent skeleton of C25 HBI unsaturated alkenes (711) produced by diatoms. Note that the odd non HBI-biosynthesizing Rhizosolenia strain, R. robusta, falls completely out of the Rhizosolenia phylogenetic cluster, indicating that its morphological classification as a Rhizosolenia diatom is probably wrong.

We examined over 120 marine diatom species, representing all major orders, for lipid composition and 18S rDNA phylogeny (12). Our study shows that only members of the four diatom genera listed above are capable of HBI alkene biosynthesis (table S1). Importantly, it also shows that phylogenetically these genera fall into in two different clusters: the so-called HBI-1 and HBI-2 clusters (Fig. 1). The HBI-1 cluster in the group of centric diatoms comprises Rhizosolenia species. These species produce either C25 or C30 HBI alkenes, or both (table S1). It has been demonstrated that changes in life cycle and environmental conditions can lead to a switch from C30 to C25 HBI biosynthesis (9). It is presently unknown whether this also applies to the two species that only produce C30 HBI alkenes (R. shrubshrolei and R. fallax). The HBI-2 cluster falls in the group of pennate diatoms and is composed of the genera Haslea, Navicula, and Pleurosigma (Fig. 1). These genera only produce C25 and not C30 HBI alkenes (table S1). Both phylogenetic clusters also contain some species that apparently do not biosynthesize HBIs. Species that share the capacity to biosynthesize specific lipids are generally also closely related phylogenetically on the basis of rDNA sequences. Examples include crenarchaeol biosynthesized by mesophilic crenarchaeota (13), ladderane lipids biosynthesized by planktomycetes performing anaerobic ammonium oxidation (14), and haptophytes producing long-chain alkenones (15). In that respect, it is unexpected that HBI alkene biosynthesis occurs in two diverse phylogenetic clusters whose average sequence similarity is only 87%. However, recently some of us showed that HBI alkene biosynthesis in the marine diatoms R. setigera and H. ostrearia, representative for both HBI clusters (Fig. 1), proceeds via different biosynthetic pathways (16). Whereas the biosynthesis of the C25 HBI alkenes in R. setigera proceeds via the mevalonate pathway, these sesterterpenes are produced via the methylerythritol pathway in H. ostrearia (16). Together with the two distinct clusters in the phylogenetic tree, this strongly suggests that the capacity to biosynthesize HBIs evolved independently in the centric and pennate group of diatoms and was not inherited from a common ancestor. Although the biochemical function of HBI alkenes is presently still unknown, the fact that HBI biosynthetic pathways apparently evolved twice also suggests that HBI alkenes must have given an important evolutionary advantage.

To determine the timing of the origin of the HBI biosynthetic pathway, we studied the fossil record. The geological occurrence of C25 HBI alkenes and the products formed during sediment diagenesis revealed a precise and consistent picture of the evolution of HBI-biosynthesizing diatoms (Fig. 2). Our sedimentary data set shows that HBI biosynthesis evolved in the Upper Turonian [circa (ca.) 90 Ma ago] (table S2). The first occurrence is noted in sediments from the Guyana Basin (French Guyana) in the Canje Formation. Before this era no C25 HBIs occur, whereas after the Upper Turonian C25 HBIs have a widespread occurrence in marine sediments. C30 HBIs were only found in 6 out of over 400 sediment samples (table S2) from the last 3 Ma, suggesting that C30 HBI biosynthesis is a relatively young adaptation of the HBI biosynthetic pathway. This detailed evolutionary picture of the evolution of the HBI biosynthetic pathway is confirmed by the HBI occurrence in petroleum (Fig. 2B). The oldest petroleum containing the C25 HBI alkane above the detection limit is from the Maracaibo Basin (Venezuela), with an inferred source rock of Cenomanian/Turonian age (table S3). Petroleum can be dated less accurately than sediments and may contain contributions from source rocks of different ages. However, for this same reason (i.e., generation from multiple source rocks), molecular fossils in petroleum provide a more “integrated” picture of fossil occurrence. The close match between the sedimentary and petroleum data provides an accurate assessment for the timing of the evolution of the HBI biosynthetic pathway. Our data not only provide a detailed account of the evolutionary rise of these diatoms but also demonstrate the robust application of the concept of age-related biomarkers for determination of the age of petroleum source rocks [i.e., specific organic compounds in petroleum that can supply information on the age of its source rock (17)]. A clear inference is that petroleum containing the C25 HBI alkane must include sources from sedimentary rocks that are Upper Turonian or younger in age. C30 HBI alkanes were not encountered in any of the petroleum samples (table S3), consistent with the relatively recent appearance of the C30 HBI alkenes in the sedimentary fossil record (table S2).

Fig. 2.

The occurrence of the C25 HBI skeleton in marine sediments (A) and petroleum (B) through geological time based on the analysis of over 400 sediment and 81 petroleum samples. For marine sediments, the relative abundance of the C25 HBI alkane is indicated by the ratio of the HBI alkane to phytane (Ph) (HBI/Ph+1) plotted on a log scale. Phytane is derived from the side chain of chlorophyll, which can be derived from all photosynthetic algae and cyanobacteria. For petroleum, the C25 HBI alkane concentration is presented on a log scale in μg g–1 (parts per million) of the total saturated alkanes of each petroleum. The detection limit is 100 ppm, and all samples plotted on the 100 ppm line, thus, do not contain detectable amounts of C25 HBI alkane. Uncertainties in the ages of the petroleum samples are indicated by error bars. Both plots clearly show that the HBI biosynthetic pathway did not exist before ca. 90 Ma. (Inset) shows the more exact timing of this event.

The HBI alkene biosynthestic pathway probably evolved independently at least twice in the diatoms (i.e., HBI-1 and -2 clusters in Fig. 1); the Upper Turonian marks the evolution of one of these two groups of diatoms. Because the fossil record clearly indicates that centric diatoms evolved before pennate diatoms (4), it is likely that HBI biosynthesis first evolved in the older group of centric diatoms (i.e., the Rhizosolenia genus). This is confirmed by the phylogenetic tree (Fig. 1), which reveals much larger distances between individual species of the HBI-1 than between species of the HBI-2 cluster. The first occurrence of HBIs is likely related to the advent of the rhizosolenid diatoms about 70 Ma ago, based on evidence from fossil silica skeletons (5). Classical fossils will underestimate the branching points in phylogenetic trees, because paleontologists will never find the first member of a clade: Diagenesis, metamorphism, and erosion remove rocks, and the fossil record will always be undersampled (18). This seems certainly likely to occur for silica skeletons too, because they are prone to dissolution, especially in organic matter–rich sediments (19). The difference in the first occurrence of Rhizosolenia based on classical and chemical fossils of ca. 20 Ma can be understood in this way: Chemical fossils transcend morphology, which may undergo evolutionary changes that obscure recognition of direct connection between fossil taxa. Furthermore, biosynthetic pathways are more conservative than morphology. The upper age limit for the evolution of the Rhizosolenia genus can be fixed by our data: We have investigated >150 sediment samples from the Upper Cenomanian/Lower Turonian from a variety of marine settings and found no HBIs in any of them. This enables their evolution to be dated with minimal error: 91.5 ± 1.5 Ma. With this date, we can now accurately calculate the evolutionary rate of the 18S rDNA gene in diatoms. The HBI-1 cluster differs on average by ca. 13% from the closest related species, Corethron species (Fig. 1). This results in an evolutionary rate of 1% per 14 Ma for the Rhizosolenia diatoms. This is substantially faster than the 1% per 18 to 26 Ma reported previously for diatoms (5) in general, which was already fast compared with reported clock rates of other organisms. However, its importance lies in the unprecedented accuracy of the molecular clock calibration.

Why did this group of diatoms evolve in the Upper Turonian? In present-day oceans, Rhizosolenia species are widespread, especially in nutrient-rich waters. They can form extensive mats, which can even be observed by satellite (20). Because they can control their buoyancy, they are able, in case of nutrient N shortage, to sink to deeper waters to access nitrate, which they store in their vacuoles (21). In this way they are able to fuel the photic zone with additional nutrient N. However, because of their large silica skeletons they require relatively high amounts of silicate. During times of extensive organic carbon burial in the Quaternary, they were often predominant members of the phytoplankton community, as in the case of sapropel formation in the Mediterranean (22) or in paleo-upwelling regions (23), in line with the inferred high-nutrient conditions. It may, therefore, at first sight be surprising that these diatoms did not flourish during one of the most extensive organic carbon burial events in Earth history: the Upper Cenomanian/Lower Turonian Oceanic Anoxic Event (OAE-2) (24). However, it has been demonstrated that during this event, owing to the stratified nature of the proto-North Atlantic ocean (25), there was a substantial depletion of nutrients in the photic zone, resulting in a competitive advantage of dinitrogen-fixing cyanobacteria (26). Breakdown of the stratification and concomitant mixing after OAE-2, because of the further opening and deepening of the Mid-Atlantic Gateway connecting the proto-North and South Atlantic Oceans, most likely resulted in higher nutrient conditions in the surface waters of the North Atlantic in the Upper Turonian. This development, initiated by plate tectonics, probably induced the evolution of open-ocean diatoms such as the Rhizosolenia species, which require high amounts of silicate. This type of phytoplankton has subsequently taken over the marine world and now fixes almost half of all the inorganic carbon used for photosynthesis in the ocean.

Supporting Online Material

www.sciencemag.org/cgi/content/full/304/5670/584/DC1

Materials and Methods

Tables S1 to S3

References

References and Notes

View Abstract

Navigate This Article