- The Paleontological Society
A morphologically diverse assemblage of organic-walled fossils from the middle Neoproterozoic Svanbergfjellet Formation, Spitsbergen, is identified as a monospecific assemblage representing the Gongrosira-phase of a vaucheriacean xanthophyte alga. As such, it provides a range of additional criteria with which to identify fossil vaucheriaceans and confirms the identification of Palaeovaucheria in the Mesoproterozoic Lakhanda Formation. Pronounced taxonomic inflation, through the practice of form-taxonomy, suggests that overall estimates of eukaryotic diversity in the Proterozoic need to be adjusted downward. Combined with positive evidence for low levels of speciation and extended stasis, pre-Cambrian eukaryotes are seen to evolve at a fundamentally lower rate than their Phanerozoic counterparts. This slower turnover accounts for the “delayed” appearance of animals without appeal to external triggers or constraints. The Cambrian acceleration of evolutionary rates was a direct consequence of newly introduced animals, whereas the much slower overall rates of the Proterozoic imply an absence of earlier metazoans.
Perhaps the single overarching question of Proterozoic paleobiology is why the inordinate amount of time between the appearance of crown-group eukaryotes in the Mesoproterozoic and the Cambrian explosion of large animals. A gap of 600–700 million years simply does not accord with the known potential for developmentally sophisticated, sexually reproducing eukaryotes to undergo rapid, coevolutionary change (see, e.g., Butterfield 2000; Poole et al. 2003). Cavalier-Smith (2002) has famously dismissed the issue by arguing for a much delayed mid-Neoproterozoic appearance of eukaryotes—elegant yes, but easily disproved (Javaux et al. 2003). More realistic explanations have focused on external constraints that might have frustrated the full evolutionary potential of early eukaryotes, such as limiting nutrients (Brasier 1992; Vermeij 1995; Anbar and Knoll 2002) and/or limiting oxygen (Knoll and Carroll 1999), though here, too, the arguments are often less than compelling (e.g., Butterfield 1997; Budd and Jensen 2000). The “snowball earth” hypothesis fails to explain the relationship between climatic perturbation and evolutionary innovation (Hoffman et al. 1998; Butterfield 2001a), whereas arguments for delayed developmental or ecological innovation do little more than beg the question.
The answer, of course, lies in deciphering the Proterozoic fossil record. Unfortunately, most Proterozoic organisms were microscopic, non-biomineralizing and, crucially, lacking in taxonomically diagnostic morphology. There are, however, a number of exceptions, and it is these that provide the framework for reconstructing early evolutionary patterns. Thus, the Neoproterozoic recognition of most of the major clades of cyanobacteria points to their early (at least Mesoproterozoic) diversification, followed by marked evolutionary stasis (Knoll 1996). Likewise, the record of ornamented acritarchs, VSMs, and multicellular “seaweeds” documents the unambiguous presence of Meso-/Neoproterozoic eukaryotes (Javaux et al. 2003). The problem with this eukaryotic record, however, is the shortage of more detailed taxonomic resolution—to the extent that only four pre-Ediacaran forms have been convincingly assigned to extant lineages (Hermann 1981; Butterfield et al. 1994; Porter and Knoll 2000; Butterfield 2000). Certainly each of these provides a valuable phylogenetic datum point; however, the broader trends of eukaryotic evolution remain obscured by very large stratigraphic gaps (typically on the order 1 Gyr) and very small sample sizes (typically a single fossil occurrence).
To a certain degree, these shortcomings have been met by incorporating the more densely sampled record of Proterozoic acritarchs (e.g., Knoll 1994, 1996), but here, too, the data can be problematic. Apart from the issue of unambiguously identifying unicellular eukaryotes (Cavalier-Smith 2002; Javaux et al. 2003), it is important to appreciate that acritarch taxa are entirely artificial. In the absence of an ontogenetic and biological context, the number of “different looking” form-taxa provides only the crudest estimate of true diversity, particularly following the appearance of complex multicellular protists and their potential for contributing multiple “organ taxa.”
In this study I describe a diverse assemblage of acritarch form-taxa from the middle Neoproterozoic (ca. 750 Ma) Svanbergfjellet Formation, Spitsbergen, which are shown to derive from a single monospecific population of vaucheriacean xanthophyte algae. As such, they help to bridge the 1000+ million-year gap between late Mesoproterozoic Palaeovaucheria and its living counterpart, and point to a significant overestimation of Proterozoic paleodiversity. Combined with a pervasive signature of stasis, the dynamics of Proterozoic eukaryotic evolution would appear to differ fundamentally from those that characterize the Phanerozoic. This alone may account for the evolutionary patterns of Proterozoic-Cambrian eukaryotes.
GEOLOGICAL AND PALEOBIOLOGICAL SETTING
The fossils occur in the Svanbergfjellet Formation of northeastern Spitsbergen, a 100–625-m-thick succession of shallow marine carbonates and shales, and the second of four lithologically similar formations that constitute the roughly 2000-m-thick Akademikerbreen Group (Harland 1997: p. 119). Although not well constrained radiometrically, the age of the Svanbergfjellet Fm. is recognized as middle Neoproterozoic by its stratigraphic position some 2000 m below Early Cambrian strata of the Oslobreen Group, and approximately 1000 m below terminal Proterozoic tillites of the Polarisbreen Group. There is some debate as to whether the two tillite horizons of the Polarisbreen Group correspond to distinct Sturtian and Marinoan events, or whether they represent a double-barreled Marinoan or Marinoan/post-Marioan glaciation (Knoll 2000). Halverson (2000), for example, has interpreted a marked negative excursion in δ13C values in the lower Akademikerbreen Group as indirect evidence for an earlier, possibly Sturtian, glaciation, which would make the Svanbergfjellet Fm. no older than perhaps 700 Ma. On the other hand, Kennedy et al. (1998) present a range of sedimentological and stable isotope evidence for identifying the lower Polarisbreen tillite as a Sturtian equivalent, suggesting an earlier age for the Svanberfjellet Fm., closer to ca. 750 Ma (see, e.g., Walter et al. 2000: Fig. 18). This latter interpretation is weakly supported by acritarch biostratigraphy (Hill et al. 2000).
The Svanbergfjellet Fm. comprises four members—the Lower Dolomite Member, the Lower Limestone Member, the Algal Dolomite Member, and the Upper Limestone member—all of which are microfossiliferous (Butterfield et al. 1994); the two dolomitic members are also conspicuously stromatolitic. The fossils discussed here are from a remarkably continuous coastal exposure of the Algal Dolomite Member on the western side of Lomfjorden between Faksevågen and Geerabukta (79°35′N,17°44E; see Butterfield et al. 1994: Fig. 1G).
The Algal Dolomite Member is approximately 140 m thick at the Lomfjorden section, comprising 83 cumulative meters of stromatolitic carbonate, 56 m of laminated shale, and a single 1.2-m-thick sandstone bed. The carbonates and shales are typically interbedded on a meter scale, but there are several more-substantial biostromes—up to 15 m thick—associated with the base and top of the member, as well as a 24-m-thick shale/sandstone succession near the middle. Carbonate facies include a variety of domal and columnar stromatolites, usually developed as laterally continuous biostromes, but also as isolated bioherms 0.5–2.0 m in diameter and up to 20 cm high. The shales are conspicuously fissile throughout the section and, along with the associated carbonates, tend to become increasingly red-colored toward the top of the member. Of particular note is the sharp lithological contact between the carbonates and shales, such that the shales rarely contain any appreciable carbonate, a habit that appears to correlate closely with exceptional, “Burgess Shale-type” preservation (Butterfield 1995).
Body fossils have not been recognized in carbonate facies of the Algal Dolomite Member, primarily because of the absence of early diagenetic silicification. Unoxidized shales immediately adjacent to these stromatolites, however, are richly fossiliferous and include a diverse range of prokaryotes, spheromorphic and acanthomorphic acritarchs, multicellular algae, and various “problematica” (Butterfield et al. 1994). Some of these shale-hosted fossils are allochthonous, derived from the plankton or locally reworked benthos; however, the common occurrence of extensive filamentous mats and bedding-plane associations of unicells points clearly to the presence of in situ benthic, photosynthetizers (see Butterfield and Chandler 1992). Combined with the occurrence of stromatolitic bioherms “floating” within the shale facies, this suggests that most, probably all, deposition of the Algal Dolomite Member took place under shallow (photic zone) subtidal conditions. The principal control on carbonate production and stromatolite growth appears to have been the fluctuating input of fine-grained siliciclastics.
The present study focuses on a 5-cm-thick horizon some 80 m above the base of the member, and 18.4 m above the base of the thick shale/sandstone succession (field sample 99-L-18). This medium-gray shale with occasional darker, sapropel-like laminations is conspicuously fossiliferous, with some bedding surfaces preserving abundant populations of small botuliform (sausage-shaped) compressions (Fig. 1). The thick shale unit in which it occurs varies from black and sulphidic through to light gray and contains an abundance of Chuaria and Tawuia compressions, particularly in the more organic-rich levels. Neither of these macroscopic fossils occurs in the L-18 sample, however, and the overall diversity of this horizon is substantially lower than other dark-shale horizons in the member; e.g., the “86-G-62” assemblage of Butterfield et al. (1994).
Several batches of the L-18 shale chips (a total of about 100 g) were dissolved in 30% hydrofluoric acid for 48 hours and the resulting slurries passed through a 62-μm-mesh sieve. Screenings were recovered in aqueous suspension and organic-walled microfossils individually picked by pipette under a transmitted-light stereomicroscope (50–100× magnification). Suspended microfossils were transferred to glass cover slips where the water droplet was removed by pipette. The cover slips with adhering fossils were then fixed to microscope slides with a heat-setting resin (Petropoxy 154). By adhering to the cover slips during mounting, most of the isolated, two-dimensional fossils were usefully oriented into a single focal plane. Local, out-of-focus detail has been corrected by the merging of multiple images taken at various focal depths.
DESCRIPTION OF THE L-18 FOSSIL ASSEMBLAGE
Both on bedding surfaces (Fig. 1) and as isolated fossils, the assemblage is dominated by simple botuliform vesicles from 20 to 80 μm diameter (x̄ = 54 μm, SD = 12 μm, n = 50), and from 80 to 1000 μm long (x̄ = 246 μm, SD = 132 μm, n = 50); in other words, these fossils are of relatively uniform diameter but range from short, effectively spheroidal vesicles to elongate, effectively filamentous vesicles, in both cases with complete, rounded termini; some bear a circular, subterminal pore and possible operculum (Fig. 2H, arrow).
The walls of these fossils are robust and well defined, with the absence of collapse-related folds and fractures pointing to a relatively plastic constitution. Most are defined by a single wall-layer; however, a number of specimens are double walled, typically with the two walls intimately associated (e.g., Fig. 2B), but occasionally with more obvious separation (e.g., Fig. 2J). The presence of a distinctive microfractured texture on the inner, but not the outer, of the two walls suggests they were of different constitution. A few specimens also preserve a third layer, a conspicuously more degraded tubular envelope surrounding/containing the botuliform cells (e.g., Fig. 2D,I).
The botuliform fossils occasionally contain darker axial structures 22–60% the diameter and 50–80% the length of the outer wall (Figs. 2C,E,G, 3B,C, 4G,K, 5C). These axial structures lack discrete walls or deformation structures and are best interpreted as degraded cytoplasm that has condensed and pulled away from the cell wall, similar to the condensation of cellular contents observed during the degradation of sheathed cyanobacteria (e.g., Golubic and Hofmann 1976; Horodyski et al. 1977). Unlike the degraded trichomes of cyanobacteria, however, these axial structures show no evidence of an originally multicellular or multiseriate construction. Indeed, along with the complete termini, these residual axial structures stand as strong evidence for identifying a single, undivided cytoplasm.
A conspicuous proportion of the botuliform fossils occur in linear arrays up to three cells long (Figs. 3F, 4). Within an array, the cells are sometimes equidimensional (Figs. 3F, 4A,G), but more often with one of the cells two to three times longer than its neighbor(s); in some cases, the constituent cells can be indistinguishable from leiosphaerid acritarchs (e.g., Fig. 4B,C,F,L). Adjacent cells in these arrays tend to be of similar diameter, texture, and color, attesting to their clonal nature. Where the texture or color is distinct, the difference may relate to the presence or absence of a double-layered wall (e.g., Fig. 4B,E,L), or a differentiated cell type (see below). The connection between adjacent cells in an array is usually tenuous with no indication of cytoplasmic connection, a habit that is also suggested by the discontinuous nature of preserved cytoplasmic residues (Fig. 4G,K). By the same token, the limited length of the cellular arrays, and the predominance of isolated botuliform cells, might be considered an artifact of taphonomy rather than original habit.
Despite the predominance of complete cytoplasmic separation, many specimens have retained a more or less open connection between adjacent cells. And unlike the obviously centripetal division responsible for the fully isolated cells, these incomplete divisions are conspicuously asymmetrical, cytokinesis apparently having proceeded from only one side of the “mother cell” (Figs. 3A, 5). The length of these incompletely and laterally divided cells shows the same irregular distribution as the fully separated, centripetal arrays, and both habits can occur on the same “filament” (Fig. 5F), strong indication that they represent alternate expressions of the same phenomenon. One possibility is that the phenomenon is simply conventional cytokinesis that has been captured at various stages in the fossil record, such as has been argued for a range of cytoplasmically connected acritarchs from the Proterozoic of Siberia (Timofeev and Hermann 1974; Hermann and Timofeev 1974); however, the peculiar asymmetry of the incomplete divisions in the L-18 fossils points to a distinct cytokinetic process. Importantly, the similar distribution of complete and incomplete divisions in these fossils indicates that cytokinesis was not related to filament growth/elongation, but rather to a secondary partitioning of a preexisting coenocyte.
In addition to the linear arrays of cells there are several examples of branching; i.e., multicelled arrays oriented in two dimensions (Fig. 6A,C–F). As with the linear arrays, the connection between cells often appears to be tenuous, but similarity in diameter and detailed microstructure confirm the relatedness of adjacent cells. If the constituent cells of these branching forms are the product of the same partitioning process as observed in the linear chains, then they would have derived from a preexisting branched coenocyte.
A significant number of isolated botuliform cells bear various subterminal or lateral structures, apparently of similar constitution to the host cell. Most of these are single and spheroidal (Figs. 6B, 7A,C–F), but in some cases they are multiple (Fig. 7B,I), botuliform (Fig. 7H), vase-shaped (Fig. 7J,K), and/or attached by a constricted stalk (Fig. 7F,G,I). Unlike the various “multicellular” forms discussed above, the attached structures in these instances tend to be considerably smaller than their host cells. Combined with their occasionally distinctive morphologies, it would appear that they represent structures ontogenetically and functionally distinct from the botuliform cells.
There is another class of spheroidal structure associated with the botuliform cells that is much darker (probably thicker walled) than those just mentioned, and with a tendency to develop folds and medial splits (Fig. 8). In some specimens, the thick-walled spheroids are clearly external to the botuliform host cells (Fig. 8D–F,I,K,L) and occasionally occur in series with them (Fig. 8E,F). In others it is possible, if not probable, that they are situated within the host cells (Fig. 8A–C,G,H,J). The diameter of these spheroids is typically 50–80% that of their host cells, but the total range includes both larger (Fig. 8F) and smaller (Fig. 8D,L) outliers. The combination of thick walls, medial splits, and, in one instance, an association of two much smaller spheroids (Fig. 8I) strongly suggests that these were some sort of reproductive cyst. That they are conspecific with the botuliform cells is confirmed by instances of interserial occurrence (Fig. 8E,F), localized constrictions in the host cells (Fig. 8A,B), alignment of multiple spheroids on or within a single botuliform cell (Fig. 8C,G), and the dearth of comparable leiosphaerid acritarchs not associated with botuliform cells.
A further variant of the botuliform cells are those with a subterminal extension that retain cytoplasmic continuity with the host cell (Fig. 9A–J). These extensions are typically 50–70% the diameter of their host cell and vary from small budlike protrusions to substantial filamentous outgrowths up to 130% the length of the host cell. The undivided extensions may be straight or curved and are usually complete with a rounded terminus. Unlike the secondary division of a coenocyte that yields the linear (and branched) arrays of cells, this variant appears to represent a growing structure caught at various stages of its development.
Finally, there are spheroidal forms with a single, often complete, filamentous extension some 40–70% the diameter of the host cell (Fig. 10). Insofar as these show the same micro-fractured texture as the botuliform cells, and the filamentous extensions fall into the same size range and have the same rounded termini (Fig. 10D,F) as those borne by botuliform cells (Fig. 9), they are reliably identified as conspecific. Indeed, alongside a range of intermediate forms (Figs. 9E, 10G), these are best interpreted simply as equidimensional versions of the botuliform cells (cf. Fig. 4B,C,F,L) that have similarly germinated a single filamentous extension.
The range of forms expressed by this fossil population presents a major taxonomic challenge. Depending on their size and aspect ratio, the isolated botuliform cells are assignable to a variety of acritarch form-genera, including Leiovalia, Navifusa, Brevitrichoides, Digitus, Teopiolia, Torgia, Eosynococcus, and/or Archaeoellipsoides. For those specimens showing indications of linear attachment, the form-genus would be Jacutianema or Arctacellularia, but if the tenuously attached, variable-sized cells also “branched” (e.g., Fig. 6) then Archaeoclada or Variclada would be more appropriate. Both the thin-walled (Fig. 7) and thick-walled (Fig. 8) spheroids associated with the botuliform cells are assignable to Leiosphaeridia, as are those botuliform cells at the equidimensional end of the size distribution. In isolation, the associated vase-shaped cells (Fig. 7J,K) might conceivably be assigned to Melanocyrilium, with obvious implications for the interpretation of some VSMs (see Porter and Knoll 2000). I am not aware of any established form-taxon that might accommodate the filament-bearing botuliform cells (Fig. 9); however, there are several that would cover the spheroidal end of the spectrum, i.e., Germinosphaera, Caudosphaera, and/or Clavitrichoides. Finally, the unbranched filamentous envelope surrounding the botuliform cells (Fig. 2D,I) would presumably be classified as a larger species of Siphonophycus, as indeed would fragments of both the botuliform cells (e.g., Fig. 8C) or their filamentous extensions. If, as argued above, the “branched” botuliform cells (Fig. 6) derived from a branched filamentous coenocyte, then there is also a reasonable case for inferring the presence of the form-genus Palaeovaucheria.
The phylogenetic affiliation of this fossil population is not immediately obvious. Looking just at the isolated and weakly linked chains of cells, a case can be made for identifying them as the akinetes of heterocystous cyanobacteria, such as Golubic et al. (1995) have argued for Archaeoellipsoides, and Nagovitsin (2000) for Palaeoanabaena. Like the L-18 fossils, cyanobacterial akinetes can be spherical or elongate, and new vegetative filaments escape through discrete germination pores (e.g., Adams and Duggan 1999: Fig. 3E); in this light, it might also be worth considering whether some of the intercalated spheroids (e.g., Figs. 4B,F, 8E,F) might represent differentiated heterocysts. The comparison, however, fails under closer inspection: unlike the L-18 fossils, cyanobacterial akinetes are thick-walled resting cells that are not subject to secondary division (Fig. 5). Cyanobacterial akinetes are also substantially smaller than the L-18 fossils (see Golubic et al. 1995: Fig. 7B), do not bear secondary cysts (cf. Fig. 8), and do not share a continuous wall with the protruding, newly germinated filament (Fig. 9).
Combined with its large size, the morphological and ontogenetic complexity of this fossil organism places it clearly within the protistan eukaryotes. Further systematic resolution can be found through morphological and taphonomic analysis of particular characters and their comparison with extant groups. At the most basic level, it is important to appreciate that these well-defined, extractable fossils are preserved as carbonaceous compressions and thus represent the robust, extracellular walls of the original organism (Butterfield 2003); i.e., they represent a walled “plant protist.” By the same token, the absence of smaller-scale partitioning, and the presence of an undivided residual cytoplasm, reliably identifies each of the fossil compartments as a “unicell.” The conspicuously large size of some of these cells—up to 70 × 800 μm—translates to cytoplasmic volumes requiring multiple nuclei (or a single giant nucleus such as in Acetabularia). Among extant plant protists, this type of siphonous or coenocytic architecture appears to be limited to the chlorophyte and chrysophyte algae (Niklas 2000: Table 1).
Of the several characters expressed by the L-18 fossils, the most distinctive is the nature of the connections between linked cells. Most show only tenuous axial attachment and almost certainly lacked any cytoplasmic connection via plasmodesmata; i.e., this is not a case of “true” multicellularity, in the sense that constituent cells intercommunicated and coordinated their activities (Ding et al. 1999; Niklas 2000). Certainly there is a clear conduit for intercommunication between cells showing “incomplete” division (Figs. 3A, 5), but the large size and lateral position of these constrictions are unlike any conventional cell division, where septum formation is typically centripetal and/or associated with a cell plate. Such constrictions are, however, a conspicuous feature of the so-called Gongrosira-phase of extant Vaucheria.
Vaucheria is a xanthophyte (yellow-green) alga characterized by branched nonseptate filaments and distinctively differentiated reproductive structures (Fritsch 1935; Entwisle 1987, 1988), and exhibits little resemblance to the L-18 fossils. Under certain conditions, however, the vegetative thallus of Vaucheria is found to be contiguous with filaments partitioned into more or less discrete segments, usually separated by thick mucilaginous intervals (Fig. 11) (Stahl 1879; Wittrock and Nordstedt 1882: no. 455; Puymaly 1922; Fritsch 1935: p. 433) and broadly comparable to the pattern found in Gongrosira, a branched filamentous chlorophyte. Notably, the division between adjacent segments of the Vaucheria “Gongrosira-phase” is not always completed, leaving either a tenuous axial attachment (Fig. 11.3), or a more open, often laterally positioned, cytoplasmic connection (Fig. 11.1, 11.2). Stahl (1879) also notes a gradational transition between Vaucheria and “Gongrosira,” with proximal portions of the Gongrosira-phase characterized by irregular and unusually long segments, and distal parts by more regular, equidimensional segments (e.g., Fig. 11.1). The Gongrosira-phase segments are multilayered and collectively surrounded by a thin continuous cuticle (Fig. 11.1–11.3, 11.8).
The Gongrosira-phase of Vaucheria appears to be an estivating/reproductive response to desiccation, in which case the resulting cells are reasonably described as akinetes (Stahl 1879; Fritsch 1935); presumably they represent the unidentified “seed bank” discussed by Dunphy et al. (2001). With the addition of water, the connecting mucilaginous components of Gongrosira-phase filaments tend to dissolve away preferentially, leaving the more resistant, layered walls of the individual akinetes. These may then (1) release the entire cytoplasm via a circular pore to establish a new vegetative Vaucheria thallus (Fig. 11.8); (2) germinate to produce a new vegetative Vaucheria thallus via an outgrowth of the inner akinete wall (Fig. 11.7); or (3) subdivide the contained cytoplasm to produce multiple amoebae, which may or may not be released (Fig. 11.9). These may then develop directly into new vegetative Vaucheria thalli (Fig. 11.10) or, under more unfavorable conditions, produce thick-walled resting cysts (Fig. 11.4, 11.5). These unusually active secondary cysts may remain spheroidal or take on a variety irregular shapes and/or divide; not uncommonly this division is incomplete, resulting in two cells connected by an hourglass-like neck (Fig. 11.5). The secondary cysts eventually germinate via a medial split (Fig. 11.6).
The similarity between the Gongrosira-phase of Vaucheria and the L-18 fossils is striking. Both, for example, are represented by linear and branched arrays of multiple-walled segments that vary in length from equidimensional spheroids to extended, undivided filaments. Both show evidence of circular escape structures in individual cells, and both may occur within a continuous outer envelope. In both cases adjacent segments typically retain little or no direct connection, giving rise to extensive disarticulation; and in both cases the only significant intercellular connection occurs in the form of a conspicuous, laterally positioned conduit. Like Vaucheria akinites, each of the “botuliform” elements in the L-18 assemblage appears to have retained the capacity to germinate a single filamentous primordium, and to produce one or more thick-walled cysts with a medial split-type of excystment.
Further comparison might be made between the stalked and vase-shaped vesicles found on some fossil cells (Fig. 7G,I–K) with the oogonia of certain living Vaucheria species. It is unlikely, however, that reproductive structures would have grown directly from akinetes, and these are probably better interpreted as variants of recently germinated filamentous primordia (cf. Figs. 9, 10), secondary cysts (cf. Fig. 8), or simply branched akinetes (Wittrock and Nordstedt 1882: no. 455; personal observation). The spheroidal structures illustrated in Figure 7A–F are also problematic with respect to Gongrosira-phase morphology: one reasonable possibility is that they are simply variants of the apparently thicker-walled spheroids illustrated in Figure 8, in which case they can be interpreted as variably sized secondary cysts (cf. Fig. 11.4, 11.5). Other discrepancies between the fossils and modern Gongrosira-phase Vaucheria—e.g., the interserial occurrence of thick-walled cells (Fig. 8E,F)—may also have fairly straightforward explanations, particularly given the considerable species diversity of modern Vaucheria (Entwisle 1987, 1988) and the lack of recent study on the Gongrosira-phase.
I have not detected any vegetative Vaucheria-like thalli in the L-18 assemblage, although there is a possible candidate elsewhere in the Algal Dolomite Member. Proterocladus hermannae is represented by a filamentous semi-coenocytic thallus with occasional club-shaped branches and a localized axial swelling, not unlike that of Palaeovaucheria (see below); however, the presence of septa on the main axis rather than the clublike branches suggests its affiliation is more likely with the siphonocladalean chlorophytes (Butterfield et al. 1994). In any event, the occasional, fragmentary preservation of an outer envelope (Fig. 2D,I) points to the likely existence of a precursor filamentous thallus, and the presence of “branched” arrays of akinete-like cells (Fig. 6) suggests that this thallus was originally branched and acellular. Combined with the considerable morphologic and ontogenetic similarities, and the expected taphonomic bias in favor of robust akinete-like structures, the L-18 fossils are convincingly interpreted as representing the Gongrosira-phase of a ca. 750 Ma vaucheriacean xanthophyte alga. The key to the identification lies in the large, laterally positioned connections between contiguous akinetes, which, at least among extant plant protists, are unique to Vaucheria. In other words, apart from the possibility of morphological convergence in an extinct filamentous, coenocytic plant protist, this character is diagnostic of vaucheriacean xanthophytes.
The xanthophytes are one of a number of chromophyte (heterokont, chlorophyll c-containing) algae most closely related to the phaeophytes (Potter et al. 1997; Ben Ali et al. 2002). Molecular analyses identify Vaucheria as the most deeply divergent member of the xanthophytes, and only distantly related to the other common siphonous form, Botrydium (Potter et al. 1997; Bailey and Andersen 1998).
OTHER FOSSIL VAUCHERIACEANS
The L-18 fossils are neither the only, nor the oldest, record of Vaucheria-like algae. Palaeovaucheria from the late Mesoproterozoic (Rainbird et al. 1998) Neryenskoi suite, Lakhanda series, of Siberia is reliably identified as the vegetative phase of a vaucheriacean on the basis of (1) a sparsely branched siphonous thallus, 18–50 μm diameter; (2) sparse septa, typically limited to bulbous filament terminations; (3) terminal circular openings in those isolated terminations, interpreted as release structures for asexual zoospores; (4) axial swellings, interpreted as developmentally arrested reproductive structures; and (5) two distinct size classes, interpreted as above- and within-sediment variants of the vegetative thallus (Hermann 1981; Woods et al. 1998). Faizullin (1998) has also identified fragmentary Palaeovaucheria material from the Baikalian of Siberia, which, on the basis of its fossils, is suggested as broadly age-equivalent to the Lakhanda.
Insofar as Palaeovaucheria is a vaucheriacean xanthophyte, it is worth considering whether it, too, might have had a Gongrosira-phase. A survey of the Neryenskoi (Lakhanda) assemblage reveals a number of likely candidates, of which the most obvious is Jacutianema, a loosely connected series of spheroidal, oval, and cylindrical cells 12–50 μm in diameter and 50–110 μm long (Timofeev and Hermann 1979; Jankauskas 1989: p. 110). Jacutianema cells have been interpreted as the akinetes and heterocysts of a filamentous cyanobacterium; however, their relatively large size, co-occurrence with Palaeovaucheria, and marked similarity to the L-18 fossils suggest they are better interpreted as the Gongrosira-phase of a vaucheriacean xanthophyte, probably Palaeovaucheria.
Also known only from the Neryenskoi suite are the branched, irregularly divided thalli of Archaeoclada and its probable taphonomic variant Variclada (in Jankauskas 1989). The compound branches of Archaeoclada are constructed of spheroidal, oval, and cylindrical cells of relatively constant diameter (25–37.5 μm) but conspicuously variable length (25–200 μm). Contiguous co-occurrence of spheroidal and botuliform cells is a notable feature of Archaeoclada, just as it is in the L-18 assemblage (compare Fig. 4C with Jankauskas 1989: Fig. 35.7), and indeed Jacutianema. Hermann (in Jankauskas 1989) has interpreted Arachaeoclada and Variclada as dasyclad algae on the basis of their branching habits and more or less coenocytic construction, but they may be better viewed as taphonomic variants of a Gongrosira-phase vaucheriacean.
The Neryenskoi suite also preserves specimens of Caudosphaera, a spheroidal vesicle bearing a single unobstructed tubular extension, which is not obviously distinguishable from Clavitrichoides or Germinosphaera tadasii (all established in Jankauskas 1989). These in turn are not obviously distinguishable from the filament-bearing spheroids of the L-18 assemblage (Fig. 10) and, like them, might reasonably be interpreted as germinating vaucheriacean akinetes (see also Butterfield et al. 1994: p. 14).
The tendency of the vaucheriacean Gongrosira-phase to disarticulate means it will regularly occur in the guise of various nondescript form-taxa, including as many as eight (putative) form-genera of botuliform microfossils (see above) and multiple species of Leiosphaeridia. Of these, the most obvious candidates for vaucheriacean affinity are the isolated botuliform cells of late Riphean Brevitrichoides bashkiricus which, like the L-18 fossils, are notable for their marked size range (50–270 μm long) and the occasional occurrence of an undivided, cytoplasmic residue (e.g., Jankauskas 1989: Fig. 21.2, 21.3, 21.6). Similarly variable but generally smaller cells (10–100 μm long) are also characteristic of Arctacellularia, which occurs both as isolated cells and in loosely connected chains, again with a condensed cytoplasmic residue. Apart from the terminal folding of individual Arctacellularia cells (its putative diagnostic feature), there is no obvious distinction to be made between B. bashkiricus and A. ellipsoidea recovered from the same Bashkirian borehole sample (Jankauskas 1989; compare Fig. 21.2, 21.3, 21.6 with Fig. 38.6–9); these, in turn, are not obviously distinguishable from Jacutianema or the akinetes of Vaucheria.
Finally, it is worth noting the occurrence of cytoplasmically connected spheroidal cells in the L-18 assemblage (Fig. 12) and in the Lakhanda and Miroedikha series of Siberia. The lack of such “intermediate” forms in large clonal populations of cells such as Eosaccharomyces, Ostiana, Eoentophyalis, or Myxococcoides suggests that these are not simply instances of arrested but otherwise conventional mitosis (contra Timofeev and Hermann 1974; Hermann and Timofeev 1974; Knoll 1996). In light of the present discussion, one reasonable alternative is that they represent the secondary cysts of a Gongrosira-phase vaucheriacean, which, as discussed above, are known to proceed through various degrees of “division,” sometimes terminating at the stage where the “daughter cells” remain connected via an hourglass-like neck (Fig. 11.5) (Stahl 1879). Interpretation of these cells as vaucheriacean is speculative at this point, but such structures are certainly expected if a modern-style Gongrosira-phase was present. Interestingly, this habit has also been documented in Arctacellularia (e.g., Jankauskas 1989: Figs. 8.17, 38.10).
Identification of the L-18 fossil assemblage as vaucheriacean xanthophytes includes it with just four other taxonomically resolved eukaryotes of pre-Vendian age: (1) Bangiomorpha Butterfield—bangiacean rhodophytes from ca. the 1200 Ma Hunting Fm., arctic Canada; (2) Palaeovaucheria, Hermann—vaucheriacean xanthophytes from the ca. 1000+ Ma Lahkanda series, Siberia; (3) Melanocyrillium Bloesser—“testate amoebae” from the ca. 750 Ma Chuar Group, Arizona; and (4) Proterocladus Butterfield—cladophoracean chlorophytes from the ca. 750 Ma Svanbergfjellet Fm., Spitsbergen. Together, these data provide sparse, but key, tie-points for reconstructing the early evolutionary history of the eukaryotes, not least as a control for molecular phylogenetic hypotheses (e.g., Yoon et al. 2002). As the younger of the two well-documented fossil xanthophytes, the Svanbergfjellet occurrence clearly does not extend the geological range of the lineage. It does, however, add considerable weight to the earlier Lakhanda identification, reducing the gap to the next known occurrence from 1000+ to just 250+ million years. More importantly, the identification of diverse Gongrosira-phase elements in the L-18 assemblage provides a host of new characters with which to identify fossil vaucheriaceans. The recognition of such elements in the Lakhanda biota corroborates earlier assignments of Palaeovaucheria to the xanthophyte algae (Hermann 1981; Woods et al. 1998) and distinguishes it from superficially similar forms such as the oomycte Saprolegnia (Potter et al. 1997).
The recognition of diverse “organ taxa” in the L-18 and Lakhanda assemblages presents a taxonomic dilemma: to “split” and thus recognize the overall range of morphotypes, or to “lump” in an attempt to approximate true paleodiversity. Documentation of “true” taxa and paleodiversity is presumably preferred, but such an approach is limited to studies of large, exceptionally preserved fossil populations that reveal the range of intraspecific variation, taphonomic, ontogenetic, or otherwise. In the case of the fossil rhodophyte Bangiomorpha pubescens, for example, the large in situ population provided a relatively complete ontogenetic series that, as isolated specimens, would have constituted perhaps a dozen form-taxa (Butterfield 2000); indeed, the co-occurring diad, Bicamera stigmata, conceivably represents a complementary diploid generation of B. pubescens (Butterfield 2001b). Likewise, large populations of the acritarch Trachyhystrichosphaera aimika have revealed a marked range of intraspecific variation, to the extent that it has subsumed at least eight junior synonyms (Butterfield et al. 1994). And among Proterozoic macrofossils, Kumar (2001) has made a strong case for recognizing a monospecific Chuaria-Tawuia-Tilsoia-Suketea association in the Vindhyan Supergroup (along with the more speculative suggestion that it may represent a vaucheriacean xanthophyte or ulotrichacean chlorophyte on the basis of its apparent coenocytic construction).
One possible conclusion to take from the Kumar (2001) study is that all “bona fide” occurrences of Chuaria or Tawuia or Tilsoia or Suketea (or indeed Longfengshania—see Hofmann 1985) represent various expressions of a single, extraordinarily long lived, globally distributed, relatively complex protist; a situation not dissimilar to that now appreciated for the Bangiaceae, Vaucheriaceae, and Cladophoraceae. Such an approach of course has limitations: the recognition that some leiosphaerid acritarchs are vaucheriacean, for example, does not imply that all leiosphaerids are vaucheriacean (though it does serve as a useful corrective to the assumption that all, or even most, Proterozoic acritarchs represent unicellular protists [Knoll 1994: p. 6743]). Even so, it is clear that taphonomic processes cannot help but inflate estimates of true fossil diversity, particularly in the case of multicellular protists with their potential for producing a broad range of relatively nondescript “organ-taxa.”
Accurate estimation of true fossil diversity is of course a prerequisite for determining true paleodiversity, and thus any meaningful account of evolutionary tempo and mode. Current estimates of early eukaryotic diversity and trends are acknowledged as volatile (e.g., Knoll 1994; Vidal and Moczydlowska Vidal 1997), but with the clear assumption that documented fossil diversity provides a representative underestimate of true diversity. At one level this is obviously true, with many taxa effectively non-preservable even under the most exceptional preservational circumstances (Butterfield 2003), and any number of taxa potentially represented by morphologically simple structures such as leiosphaerids. However, it is also the case that preservable fossils can stand as a proxy for total diversity and overall evolutionary dynamics (e.g., Sepkoski 1993; Butterfield 1997, 2001c, 2003)—provided that the collated taxa represent natural groups. Unfortunately, the binning of Proterozoic form-taxa through time cannot be interpreted as a Sepkoski-type curve, simply because, unlike the majority of Phanerozoic data, they do not represent “real,” or even paraphyletic (Sepkoski and Kendrick 1993), taxa.
Owing to their exceptional preservation, the Lakhanda and Svanbergfjellet assemblages are considered exemplars of late Mesoproterozoic and middle Neoproterozoic life, respectively. Both record a rich variety of microfossils, but the majority have been documented without accompanying biological or taphonomic analysis. In both cases, a significant percentage of the purported diversity has now been identified as variable expressions of vaucheriacean algae, probably the same species. By the same token, it is probably premature to recognize more than a few additional eukaryotic taxa from the remaining form-taxa. Yes, the Svanbergfjellet preserves more acritarch taxa than the Lakhanda (Knoll 1996), but its additional range of multicellular protists—a source of multiple form-taxa—suggests a much more modest difference in true diversity.
The case for recognizing a more limited diversity of Proterozoic eukaryotes—relative to what has conventionally been assumed—is supported by their extraordinary evolutionary conservatism. Knoll (1994), for example, has shown the turnover of Proterozoic acritarchs to be at least an order of magnitude lower than their Cambrian counterparts, a habit exemplified by the roughly 500-Myr range of Trachyhystrichosphaera aimika (Butterfield et al. 1994; Samuelsson and Butterfield 2001), the roughly 700-Myr range of Valeria (Hofmann 1999), and the approximately 1000-Myr range of Chuaria/Tawuia (Kumar 2001; Samuelsson and Butterfield 2001). A similar degree of stasis is seen among taxonomically resolved Proterozoic eukaryotes: bangiacean rhodophytes with a minimum 1200-Myr stratigraphic range (Butterfield 2000), vaucheriacean xanthophytes with a minimum 1000-Myr range (Hermann 1981; Woods 1998), cladophoracean chlorophytes with a minimum 750-Myr range (Butterfield et al. 1994), and testate amoebae with a minimum 750-Myr range (Porter and Knoll 2000). And, at a more modest scale of 20–30 Myr, it is becoming clear that terminal Proterozoic Ediacaran biotas likewise show limited, if any, evolutionary turnover (Narbonne and Gehling 2003; Grazhdankin 2004). Indeed, excluding forms known only from single occurrences (i.e., with unknown stratigraphic ranges), the most striking aspect of early eukaryotic biotas is how very similar they all are—to the extent that a convincing case of within-sequence biostratigraphy (i.e., with a uniquely older fossil succeeded by a uniquely younger one) has yet to be reported for the Proterozoic. To a first approximation, Proterozoic diversity is represented overwhelmingly by the equivalent of living fossils, i.e., extinction-resistant evolutionary lineages characterized by a conspicuously low levels of speciation (Stanley 1979). Almost by definition, a biosphere dominated by nonspeciating lineages is a biosphere of deeply limited diversity.
One exception to this generalization is the recently published account of 12 species, in nine genera, of vase-shaped microfossils (VSMs) from a single stratigraphic horizon in the middle Neoproterozoic Chuar Group (Porter et al. 2003). This monographic spike derives directly from the form-taxonomic approach applied to extant testate amoebae (Porter et al. 2003), which may or may not reflect true biological diversity (vs. ontogenetic/ecophenotypic variability), and may or may not be applicable to Proterozoic VSMs. If such protozoans were indeed diverse through the Neoproterozoic, then the absence of evolutionary change in contemporaneous taxa is particularly notable: whatever their overall ecological/evolutionary role, unicellular heterotrophs were clearly not contributing to morphological diversification.
More general support for the minimal impact of Proterozoic eukaryotes can be found in the stratigraphic record of 2-methylhopanoids, a biomarker molecule thought to be diagnostic for cyanobacteria (Summons et al. 1999). The prevalence of 2-methylhopanoids in offshore shales of Proterozoic, but not Phanerozoic, age suggests that eukaryotic phytoplankton only acquired its current prominence in the early Paleozoic. Proterozoic primary productivity appears to have been contributed almost entirely by cyanobacteria, which, like their eukaryotic counterparts, are noted for their deep evolutionary conservatism (Knoll 1996).
In summary, there is a strong case for reconsidering the nature of the Proterozoic biosphere. Rather than a cryptic equivalent of the Phanerozoic, it reflects a qualitatively different evolutionary context, represented by fundamentally lower diversity and fundamentally slower turnover. Whether or not a significant proportion of early eukaryotic diversity is obscured by simple morphology or non-preservation, those forms that did express distinctive, preservable morphology were conspicuously immune to both extinction and cladogenesis. Contrary to expectation (e.g., Butterfield 2000), escalatory morphological coevolution was not a feature of the Proterozoic biosphere.
The Cambrian Explosion
The evolutionary lethargy of the Proterozoic stands in stark contrast to high rates of morphological diversification seen in the early Cambrian and helps to explain the 600–700-Myr gap separating the Mesoproterozoic appearance of crown-group eukaryotes and the Cambrian explosion of large animals. Insofar as adaptations are the product of natural selection, the proximal causes of the Cambrian explosion are easily recognized in those features that distinguish the Cambrian biosphere. The sudden and polyphyletic appearance of jaws and claws, armor and eyes, large size and biomineralization, muscles and movement is clearly the product of interactions with (other) animals. The relative rapidity with which metazoans diversified is impressive, but hardly surprising—such rates are expected given the potential of eukaryotic evolution in the throes of a coevolutionary adaptive radiation (Schluter 2000; Cavalier-Smith 2002). The real question is why that potential was not realized earlier.
The answer, in part, lies in recognizing the fundamentally slower rates of evolutionary change in the Proterozoic. The seemingly inordinate length of time between the appearance of crown-group eukaryotes in the Mesoproterozoic and the Cambrian explosion only seems inordinate if Phanerozoic evolutionary rates are (unjustifiably) extrapolated back into the non-uniformitarian world of the Proterozoic. As much as anything, the Cambrian explosion marks the change from an archaic system of minimal evolutionary turnover to the conspicuously accelerated turnover of the Phanerozoic. Such increases might, of themselves, account for the evolutionary “discovery” of animals, but, given the nature of adaptations in Cambrian organisms, the reverse is more likely to be the case: the cause of the increased rates of evolution, and thus the Cambrian explosion, was the evolution of animals.
In a series of papers through the 1970s, Stanley (e.g., 1973, 1976) argued that, in the absence of sexually reproducing eukaryotic heterotrophs, marine ecosystems would have been dominated by low-diversity primary producers and consequent evolutionary stasis. Moreover, the introduction of such heterotrophy would have induced a “self-propagating feedback system of diversification,” otherwise known as the Cambrian explosion. This ecologically modeled “cropping hypothesis” has been marginalized over the years by the discovery of a significant disparity of pre-Cambrian crown-group eukaryotes, the recognition of sexual reproduction by at least the Mesoproterozoic, and the realization that eukaryotic organelles were acquired through a process of heterotrophic activity (phagocytosis), early in the history of the group. The scenario I have outlined above is effectively an updated version of the cropping hypothesis, incorporating these new data and arguing that it was the ecological impact of animals, in particular, that fueled the Cambrian explosion.
It is, of course, still possible that the belated appearance of animals derives from external constraints such as limiting nutrients (Brasier 1992; Vermeij 1995; Anbar and Knoll 2002) or oxygen (Knoll and Carroll 1999), but a convincing case has yet to be made for these additions (see, e.g., Butterfield 1997; Budd and Jensen 2000). Moreover, with a rate of evolutionary turnover at least an order of magnitude lower than those seen in the Phanerozoic (e.g., Knoll 1994), the 600–700-Myr gap between the appearance of crown-group eukaryotes and the Cambrian Explosion converts to no more than perhaps 50 Myr of equivalent Phanerozoic evolutionary time. There is, in fact, no real evolutionary gap and thus no a priori reason to invoke special constraints, or triggers (e.g., Hoffman et al. 1998; Knoll and Carroll 1999; Amthor et al. 2003; Bloh et al. 2003), for a delayed Cambrian Explosion. The cause of the Cambrian explosion was simply the evolution of animals, which, when adjusted for rate, occurred soon after the appearance of sexually reproducing, crown-group eukaryotes.
Not all gaps are as easily accounted for. The evolution of land plants, for example, was “delayed” by some 750 Myr, of which 150 were running at accelerated Cambrian and Paleozoic rates. Curiously, there has been less of an effort to invoke physical constraints in this instance, possibly because the most obvious hypothesis—oxygen/ozone-layer limitation—is easily tested, and dismissed, by reference to the contemporaneous animal record. If, as I have argued here, animals have been the prime motors of evolutionary change among eukaryotic organisms, then it may be worth considering an ecological explanation for the delayed appearance of vascular land plants—particularly in light of the derived and delayed appearance of marine macroherbivory (see Vermeij and Lindberg 2000).
Finally, the fossil record of Proterozoic plant protists can be used to assess claims of deep metazoan roots as inferred from molecular and phylogenetic analyses (e.g., Wray et al. 1996; Fortey et al. 1996) and/or putative Palaeoproterozoic/Mesoproterozoic trace fossils (e.g., Seilacher et al. 1998; Rasmussen et al. 2002a). With evolutionary rates running at least an order of magnitude slower than in the Cambrian, the interval separating Holland's (2002) 2200 Ma “Great Oxygenation Event” and the ca. 1600 Ma Vindhyan/Stirling “bilaterians” converts to approximately 60 Myr of equivalent Phanerozoic evolutionary time, broadly equivalent to estimates for the evolution of metazoan development suggested by the terminal Proterozoic/early Cambrian fossil record (Budd and Jensen 2000). What fails to ring true, however, is the billion additional years of undiminished stasis, given the capacity for animals, in particular, to drive rapid, progressive, coevolutionary change (Stanley 1973, 1976; Butterfield 1997, 2000, 2001b). Rasmussen et al.‘s (2002b) appeal to limited oxygen levels as the brake on subsequent diversification is hoist on its own claim for early, energetic, macroscopic metazoans, and Seilacher et al.‘s (1998) suggestion for the dramatically accelerating influence of biomineralization ignores the fact that the vast majority of Cambrian organisms were non-biomineralizing. By contrast, an appreciation of the unique impact of metazoan ecology on evolutionary tempo and mode, derived from judicious evaluation of the Proterozoic fossil record, provides a relatively simple and sufficient account of “early” eukaryote evolution.
All illustrated specimens are housed in the Sedgwick Museum, University of Cambridge (acquisition numbers SM X.40993 through SM X.41077).
Genus Jacutianema Timofeev and Hermann, 1979, emend
Brevitrichoides bashkiricus Jankauskas, 1980; Plate 12, Fig. 4. Arctacellularia kelleri Hermann and Jankauskas, 1989 (in Jankauskas 1989); Plate 38, Figs. 3–5. Arctacellularia varia Jankauskas and Hermann, 1989 (in Jankauskas 1989); Plate 38, Figs. 6–8, 10–13.
Gongrosira-phase of fossil vaucheriacean alga. Isolated or loosely connected series of botuliform cells of variable length. Sometimes with trace of condensed cytoplasm; sometimes double-walled; sometimes with circular subterminal pore; sometimes with outer filamentous envelope; sometimes with laterally positioned cytoplasmic connection between cells; sometimes branched; sometimes with associated, variably shaped, thin- or thick-walled vesicles, the thick-walled vesicles commonly showing medial-split excystment structures; sometimes with unobstructed subterminal filamentous extensions of variable length; sometimes in the form of incompletely divided thick-walled vesicles.
Finding a name and diagnosis for the morphologically diverse but monospecific L-18 assemblage shows up the basic conflict between natural and form taxonomy. I have chosen to use the form-genus Jacutianema because it is the most senior taxon whose type-material can be reasonably interpreted as homologous with the L-18 material. In other words, there is a good case—based on the range of morphology and co-occurring form-taxa—for identifying Jacutianema as the Gongrosira-phase akinetes of a vaucheriacean alga. Moreover, there is no obvious distinction to be made between the Lakhanda and L-18 populations of Jacutianema; nor is there any compelling morphological distinction to be made with other botuliform form-taxa, hence the identification of B. bashkiricus, A. kelleri, and A. varia as junior synonyms of Jacutianema. In all likelihood, Palaeovaucheria Hermann, 1981 is also a junior synonym, although this is unlikely to be accepted in the absence of directly contiguous fossils (cf. Stahl 1879 in the case of extant Gongrosira-phase Vaucheria). The retention of a separate name for this distinct life-history phase may be useful in any event, though at the cost of inflating true Proterozoic diversity. Form-species of Leiosphaeridia, Siphonophycus, Germinosphaera/Caudosphaera/Clavitrichoides are also represented in Gongrosira-phase vaucheriaceans but cannot be identified reliably in isolation.
I thank T. Dunkley for assistance with fieldwork, Norsk Polarinstitutt for logistical support, N. Baumgarten for translation of German literature, and D. Grazhdankin for translation of Russian literature. I am particularly grateful to R. Edgar at the Farlow Herbarium (Harvard), who tracked down the Wittrock and Nordstedt (1882) reference of “published” Gongrosira-phase material; the University of Cambridge Herbarium provided specimens from this reference collection. A. Poole and L. Graham reviewed the original manuscript and made several useful suggestions. Fieldwork for this study was funded by the Natural Environment Research Council (RG27173). This is Cambridge Earth Science Contribution 7560.
- Accepted 23 October 2003.