The Eocene Green River Formation in the Sand Wash Basin, Colorado (United States), contains the largest known lacustrine columnar stromatolites. Spectacular columns, as much as 5.5 m high with diameters of 7 m, occur over a broad area. The stromatolites are composed of laterally continuous centimeter-thick layers that can be traced from the base to the top of the column (synoptic relief), thus these stromatolites stood as much as 5.5 m above the lake floor. The layers consist of one to several different kinds of microbialites, which makes these large columns even more unusual. The giant columns were the result of a combination of factors including the lake transgressing a flooded woodland, in situ tree stumps providing elevated substrates above the lake floor, an abundant supply of calcium-rich spring and surface water that mixed with saline-alkaline lake water, and the subsidence rate of the nearshore lake environment not exceeding the rate at which the stromatolites grew.


The study of lacustrine microbialites (stromatolites) is undergoing a renaissance. The discovery of large petroleum reserves in reservoirs composed of microbialites in lacustrine successions in the Lower Cretaceous “pre-salt” of offshore Brazil and Angola has motivated a quest for modern and ancient microbialite analogs.

The Eocene Green River Formation in the Sand Wash Basin, Colorado (United States) (Fig. 1) contains the most abundant and diverse record of lacustrine microbialites. Some studies use these microbialites in making interpretations about the depositional system; however, few studies have described and analyzed the microbialites in detail (Bradley, 1929; Frantz et al., 2014).

The LaClede Bed, Laney Member, Green River Formation, contains a unique occurrence of microbialites: giant, multi-meter size, columnar stromatolites (Fig. 2A). Columns are up to 5.5 m high, are as much as 7 m wide, and have laterally continuous layers that can be traced from the base to the top of the column. These giant columns were noted by Kornegay and Surdam (1980) and Roehler (1993), but despite their impressive size, had not been studied. These are the largest lacustrine columnar stromatolites known.

Our study describes the giant stromatolites and attributes their large size to a combination of factors: (1) the lake transgressing a nearshore woodland, (2) in situ tree stumps providing substrates for stromatolite growth, (3) high input of Ca2+-rich waters from surface inflow and fault-sourced spring water into a saline-alkaline lake system, and (4) subsidence not exceeding rate of stromatolite growth.


The Green River Formation is the most studied ancient lacustrine system and is found in Colorado, Utah, and Wyoming. It was deposited in three, infrequently connected lakes (Lakes Gosiute and Uinta and Fossil Lake; Fig. 1) from 53.5 to 48.5 Ma (Smith et al., 2003). Four depositional basins were involved: the Green River Basin (Lake Gosiute), the Uinta and Piceance Creek Basins (Lake Uinta), and Fossil Basin (Fossil Lake). More than 2000 m of sediments accumulated in these basins, with ∼80% of it carbonate.

The Sand Wash Basin, a present-day sub-basin of the Green River Basin, is contiguous with the Washakie Basin to the north. The Green River Formation in these two sub-basins is ∼450 m thick and interfingers with the Wasatch Formation’s Cathedral Bluffs Tongue, which consists of floodplain deposits (Roehler, 1993). The Green River Formation here is composed of the basal Luman Tongue, followed upsection by the Tipton Member and the Laney Member (LaClede Bed) (Roehler, 1992). The depositional axis of the Green River Basin (Lake Gosiute) during LaClede Bed time was in the southern part of Lake Gosiute and paralleled the Uinta Mountains, which were being uplifted at this time (Roehler, 1993).

The LaClede Bed is laterally extensive and can be traced throughout most of the Green River Basin. In general, the lower part of the LaClede Bed was deposited in a brackish to saline-alkaline lake and the upper part in a freshwater phase (Surdam and Stanley, 1979).

The exact location of the stromatolites is available to qualified researchers from the Regional Paleontologist at the U.S. Bureau of Land Management, Colorado State Office, Denver. Three localities were studied (Fig. 1). Locality 1 is the main locality, where a section was measured (Fig. DR1 in the GSA Data Repository1) and the giant stromatolites occur in a complex biostrome, up to 5.5 m thick, associated with coarse-grained carbonate and quartz sand. The base of the biostrome has numerous stromatolite-coated logs (Fig. DR3A). Locality 2, ∼1 km to the northeast, has isolated stromatolites interbedded with a laminated aragonite and calcite mudstone. Individual beds can be traced from locality 1 to locality 2. Near locality 2, at least three stromatolite beds occur separated by meter-thick sequences of aragonitic laminated mudstone.

Locality 3, ∼2 km northeast of locality 2, is a deeper-water, basinward location and consists of primarily laminated aragonite and calcite mudstone with at least one thin, 20-cm-thick stromatolitic biostrome. This biostrome is not exactly time equivalent to the main stromatolite unit at locality 1 (it occurs ∼5 m below).


The giant stromatolites are found at locality 1 (Fig. 1) in the lower part of the LaClede Bed. The outcrops at locality 1 cover an area <1 km2, and the stromatolites occur in a 5.5-m-thick carbonate bed of domical and columnar stromatolites; in essence, a biostrome. Five major types of stromatolites have been observed at the macrostructural level: (1) large, multi-meter-size, stromatolite columns with hemispherical tops and hollow, cylindrical centers that are not open at the top (Figs. 2A and 2B); (2) stromatolite columns several centimeters to ∼1 m in diameter and height with conical tops (Fig. DR4A); (3) bulbous columnar stromatolites >1 m in size (Fig. 2A); (4) laterally linked bulbous stromatolites <1 m in diameter (Fig. DR4B), and (5) prostate logs encrusted by stromatolites (Fig. DR3).

The large, multi-meter-size columns are by far the most unusual stromatolites. They are “true” stromatolites; i.e., they are laminated. A number of the giant stromatolites have weathered out from outcrop and rolled down the slope, exposing cross sections of their hollow basal portions (Fig. 2B). Each one examined had a cylindrical hollow center.

An unusual feature in the giant stromatolites is the layering (Figs. 2A and 2B; Fig. DR2). Columns are constructed, from base to top, of successive layers, one to a few centimeters thick, which are laterally continuous. Each layer is composed of one to several different types of microbialites. Layers with a single type include centimeter-size columnar stromatolites, cumulate (laterally linked) stromatolites, ooidal boundstone, and shrubs (Fig. DR2; Figs. 2A and 2B). Layers composed of multiple types of microbialites include planar to wavy laminated stromatolites, cumulate (laterally linked domical stromatolites, <1 cm diameter), stubby stromatolite columns, ooidal boundstone, and shrubs. In thin section, radial (crystal) fans have not been observed, and laminae within the stromatolite portions of the layers are irregular and not isopachous, all suggestive of microbially influenced growth. Cumulate stromatolites and layers with shrubs are commonly silicified, although the shrubs themselves remain mostly carbonate.

Shrubs are an important but enigmatic (are they biogenic?) component of the lacustrine Cretaceous pre-salt hydrocarbon reservoirs off the coast of Brazil (Wright and Barnett, 2015; pre-salt refers to hydrocarbon-rich strata below thick salt deposits). Shrubs of locality 1 stromatolites (Fig. 2C) are arborescent structures, ∼1–2 mm in diameter and up to 1 cm tall, and occur in both calcareous and silicified layers of the giant stromatolites. Shrubs are composed of peloids, suggesting that these are biogenic rather abiogenic (crystal shrubs).

The giant stromatolites gradually decrease in spatial density, height, and diameter north of locality 1. At locality 2, stromatolites were only observed as isolated, cylindrical, prostrate stromatolites, up to ∼50 cm in diameter and up to 4 m long, most with hollow centers. The inner hollow surfaces commonly have impressions of bark. Although rare, silicified wood can be found in the stromatolite’s center (Figs. DR3C and DR3D). The 20-cm-thick biostrome at locality 3 is composed of bulbous stromatolites, up to 10 cm in diameter, initiating on clasts in a flat-pebble conglomerate. No coated logs were observed here.


A 20 m stratigraphic section containing the giant stromatolites was measured at locality 1 to establish the stratigraphic context (Figs. DR1, DR5A, and DR5B). The lower half of the section is dominated by fine-grained sediments: claystone and carbonate mudstone (laminated calcitic micrite), with one ostracodal wackestone bed. The upper half of the section contains ∼6 m of coarse-grained units (calcareous quartz sandstone, oolite, and a mixed quartz sand–oolite bed with coated ostracods). The stromatolites, including the giant ones, occur within this interval and are overlain by quartz sand mixed with ooids.

The interspace area between the giant stromatolites contains fine quartz sand and ooids. In addition to the giant stromatolites, thinner, 0.4–1.7-m-thick beds of smaller stromatolites occur. Laterally a few hundred meters to the west-northwest of the measured section, the stromatolites are more densely packed. To the northwest, within a few kilometers of the measured section, the lithology grades rapidly into fine-grained, laminated carbonate mudstones (Figs. DR5C and DR5D).


Table DR1 in the Data Repository contains carbon and oxygen isotopic analyses (versus Vienna Peedee belemnite) of stromatolites and non-stromatolitic carbonates. A 48-cm-tall stromatolite from locality 1 (sample VC4) was slabbed and sequentially sampled from bottom to top for isotopic analyses (Fig. DR4D). Results indicate a predominately negative δ18O, ranging from −9.43‰ to −7.07‰, averaging −7.95‰ (N = 18). The δ13C for this stromatolite ranged from +0.57‰ to +2.44‰ and averaged +1.76‰ (N = 18).

Basinward, at locality 3, the succession is replaced by thick successions of laminated carbonate mudstone, some with fish bones indicating deeper water. Aragonite (see below) in this mudstone has a δ18O value of −5.90‰ and δ13C value of +4.94. The aragonite is enriched in 18O in contrast to depleted values found shoreward at locality 1.

Table DR1 also provides X-ray diffraction (XRD) data from the stromatolites and associated facies. Aragonite occurs in five samples, ranging from 5% to 97%. The 97% sample came from the laminated carbonate mudstone at locality 3 (sample VC13-A). The shrub (sample 5 of 8-20-13; Figs. 2C and 2D) with the acicular aragonite contains 23% aragonite (the sample is mostly chert). At locality 3, monomineralic laminae of aragonite alternate with laminae that are 70% clay-size siliciclastics (45% quartz, 25% feldspar, trace of clay) and 30% carbonate (17% calcite, 13% dolomite). Analysis of microbialites indicates that dolomite increases relative to calcite in a basinward direction.

XRD analysis of shrubs in silicified layers indicate that they are 23% aragonite (Table DR1). Petrographic analyses of these shrubs revealed acicular aragonite on the shrubs’ surface extending into the inter-shrub space (Fig. 2D).


The growth and development of the giant stromatolites are related to a combination of factors, including (1) supply of calcium-rich ground and surface waters, (2) mixing of those waters with saline-alkaline, carbonate-rich lake water, (3) high accommodation space (deeper water), (4) nearshore, shallow, energetic water (waves), (5) clear, non-turbid water, and (6) episodic input of siliciclastics. The great height of the columns (>5 m) was facilitated by growth on in situ tree stumps that provided elevated substrates. A flooded woodland in this area was postulated by Kornegay and Surdam (1980). Microbialite growth was terminated by the input of voluminous siliciclastics and the lake deepening faster than microbialite growth rates.

Critical among factors controlling lacustrine microbialite development here is an abundant supply of calcium-rich spring and surface waters (Fig. 3). During stromatolite growth, the area was within a few kilometers of the ancestral Uinta Mountains. Basin-margin fanglomerates and active faults that parallel the mountain front were likely conduits of calcium-rich spring water. Spring water and faults were viewed as important in the development of tufa mounds and stromatolites along the south shore of Lake Gosiute near Manila, Utah (Surdam et al., 1980). High accommodation space created by rapid subsidence along the faulted north flank of the Uinta Mountains front provided deeper water. The lake-bottom gradient steepened toward the Uinta Mountains front.

The presence of aragonite within the stromatolites (Fig. 2D) and mudstone laminae composed of aragonite crystals (Figs. DR5D–DR5F) deposited basinward from the stromatolites provides another important element to understanding the environment. Smith (2009) described aragonite precipitation in the lacustrine late Pleistocene to mid-Holocene Searles Lake Formation, California: The lake was saline-alkaline (pH > 9), HCO3 rich, and enriched with Si, Mg, and Na. Carbonates rapidly precipitated during mixing of inflow waters with ambient lake waters, resulting in whitings that were deposited on the lake bottom as aragonitic laminae. The occurrence of aragonite within the giant stromatolites suggests that infiltrating spring water mixed with saline-alkaline pore waters within the stromatolites (layers with shrubs were very porous) resulted in silica cementation as the pH of the pore water dropped (Surdam and Stanley, 1979). Fresh inflow also diluted the saline lake water, resulting in a lateral chemical gradient within the lake, as indicated by an increase of dolomite in microbialites in a basinward direction.

The conclusion that the growth and development of the giant stromatolites was facilitated by the input of both spring and surface fresh water is consistent with the isotopic data. The depleted δ 18O values (average of −8.00‰) of the nearshore stromatolites at locality 1 are consistent with an interpretation of freshwater input. The primarily calcitic signature of the stromatolites is also consistent with freshwater influence. In contrast, the basinward increase in salinity-alkalinity (lateral chemical gradient) at locality 3 is supported by the enriched isotopic values (as well as increase in dolomite versus calcite) observed in the aragonite lamina (δ18O value of −5.90‰ and δ13C value of +4.94‰). Higher salinities are also supported by the dominance of dolomite and aragonite as the primary carbonate mineral in the carbonate mudstone at locality 3. However, lakes are dynamic chemical systems with evaporation and precipitation ratios changing with seasons and longer-term climatic cycles. At the nearshore, giant stromatolite location, lake chemistry was particularly dynamic and fluctuated depending on the amount of surface inflow and spring water entering the saline-alkaline lake. The aragonite in the shrubs provides evidence that this was the case. In addition, the presence of significant amounts of dolomite within other layers of stromatolites is interpreted as indicative of periods of higher salinity.

An alternate hypothesis comes from Frantz et al. (2014) who proposed that negative δ18O excursions reflected freshening of the lake and lake-level increase, while positive δ18O excursions reflected decreased recharge and lake-level drop. This is in agreement with the conclusions presented here; both models relate δ18O to lake freshening (becoming hydrologically open). However, the Frantz model assumes homogeneous lake chemistry. The model proposed here (Fig. 3) has a lateral chemical gradient (heterogeneous conditions), which is supported by the presence of aragonite in the study area, but its absence basinward (see Roehler, 1991). Aragonite forms by the contact of Ca-rich, fresh water with saline lake water (Reeves, 1968).


The world’s largest known lacustrine columnar stromatolites are not only worthy of attention as unique fossils, but they provide insight into critical paleoenvironmental and chemical factors conducive to the development of microbialites in lacustrine environments. It is unusual that these stromatolites were able to reach the size they did in such close proximity to the ancestral Uinta Mountains, which were being uplifted at the time. Large amounts of siliciclastics and fanglomerates (the Cathedral Bluffs Tongue) were being shed into the lake during Laney Member time (Roehler, 1992). This suggests that despite the close proximity of the Uintas, a period of relatively stable lake conditions and accommodation space existed for the growth of the stromatolites. Tree stumps provided an elevated substrate, with spring and surface waters rich in Ca2+ entering the nearshore environment resulting in the enhanced carbonate precipitation. The termination of stromatolite growth was brought about by abundant siliciclastic input (the stromatolites were buried by quartz sand), followed by rapid lake deepening resulting in deposition of profundal lacustrine mudstones on top of the quartz sandstone unit.

A research grant from the University of California–Santa Barbara’s Academic Senate funded the isotopic analyses. Research on this project was done in conjunction with a Paleontological Resources Use Permit (#COC76610) from the U.S. Bureau of Land Management. We thank Carie Frantz, Pedro Marenco, Hank Chafetz, and one anonymous reviewer for helpful comments and suggestions.

1GSA Data Repository item 2015241, Table DR1 and Figures DR1–DR5, is available online at www.geosociety.org/pubs/ft2015.htm, or on request from editing@geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.