Oxygen isotope composition of the Phanerozoic ocean and a possible solution to the dolomite problem

Oxygen isotope composition of the Phanerozoic

ocean and a possible solution to the

dolomite problem

Uri Ryb

a,1

and John M. Eiler

Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA 91125

Edited by Donald E. Canfield, University of Southern Denmark, Odense, Denmark, and approved May 16, 2018 (received for review November 10, 2017)

The

O of calcite fossils increased by

∼

‰

between the Cam-

brian and present. It has long been controversial whether this

change reflects evolution in the

δ

O of seawater, or a decrease in

ocean temperatures, or greater extents of diagenesis of older strata.

Here, we present measurements of the oxygen and

‟

clumped

”

iso-

tope compositions of Phanerozoic dolomites and compare these

data with published oxygen isotope studies of carbonate rocks.

We show that the

δ

O values of dolomites and calcite fossils of

similar age overlap one another, suggesting they are controlled

by similar processes. Clumped isotope measurements of Cambrian

to Pleistocene dolomites imply crystallization temperatures of 15

–

158 °C and parent waters having

δ

VSMOW

values from

−

2to

+

‰

. These data are consistent with dolomitization through sed-

iment/rock reaction with seawater and diagenetically modified sea-

water, over timescales of 100 My, and suggest that, like dolomite,

temporal variations of the calcite fossil

δ

O record are largely

driven by diagenetic alteration. We find no evidence that Phanero-

zoic seawater was significantly lower in

δ

O than preglacial Ceno-

zoic seawater. Thus, the fluxes of oxygen

–

isotope exchange

associated with weathering and hydrothermal alteration reactions

have remained stable throughout the Phanerozoic, despite major

tectonic, climatic and biologic perturbations. This stability implies

that a long-term feedback exists between the global rates of sea-

floor spreading and weathering. We note that massive dolomites

have crystallized in pre-Cenozoic units at temperatures

>

40 °C. Since

Cenozoic platforms generally have not reached such conditions,

their thermal immaturity could explain their paucity of dolomites.

clumped isotope

|

dolomite

|

Phanerozoic

|

sea water

|

oxygen isotope

he isotopic composition of seawater, averaged over time-

scales longer than glacial

–

interglacial cycles, is controlled by

surface and seafloor weathering and hydrothermal alteration (1).

Silicate weathering reactions and authigenic mineral pre-

cipitation occurring at or near Earth-surface temperatures are

associated with large (typically

∼

–

‰

) oxygen isotope frac-

tionations between

O-rich minerals and

O-poor waters (2).

If the ocean

’

s oxygen isotope budget was controlled only by

weathering of mantle-derived igneous rocks, the

VSMOW

(Vienna Standard Mean Ocean Water) of seawater would be

approximately

−

‰

(2). Hydrothermal alteration of the oceanic

crust generally occurs at elevated temperatures (up to

∼

300

–

350 °C) (3) where the oxygen isotope fractionation between

minerals and water is small (

∼

‰

) (2); if hydrothermal alter-

ation of mantle-derived magmas were the only process influ-

encing the oxygen isotope budget of the oceans, the

VSMOW

of seawater would be

∼

‰

. The fact that the

VSMOW

of the

ice-free ocean has been approximately

−

‰

throughout the

Cenozoic (4) implies that the ocean has been near steady state

with a balance of

∼

44% weathering and 56% hydrothermal al-

teration (as fractions of the oxygen isotope exchange fluxes) (5).

It has been estimated that the residence time of the ocean with

respect to oxygen isotope exchange with the lithosphere is 250 My

(6). This is short enough that we cannot assume the balance that

has prevailed over Cenozoic times was true deeper in Earth his-

tory. Conversely, any record of the

O of seawater at earlier

times could be interpreted as a reflection of changes in the relative

rates and/or conditions of weathering and hydrothermal alter-

ation. Several studies have attempted to reconstruct the isotopic

composition of seawater using marine carbonates (7

–

10), phos-

phorites and cherts (11

–

13), kerogens (14), and altered oceanic

crust and iron ores (1). The interpretation of all of these records

depends on assumptions regarding the temperatures at which the

studied materials last exchanged oxygen with water. A common

approach to this problem is to identify samples that grew at Earth-

surface conditions; however, this method is only useful if we can

reliably recognize materials that have escaped diagenesis.

Arguably the most abundant and influential record of this kind

has examined calcite fossils (mostly brachiopods) that are

inferred to have escaped diagenetic modification (7

–

10). Mate-

rials that pass these studies

’

criteria for preservation record an

‰

increase in

O between the Cambrian and the present. If

this is a primary depositional signal, it implies a similarly large

increase in the

O value of seawater, suggesting a decrease

over time in the relative importance of weathering and an in-

crease in the relative importance of hydrothermal alteration (2,

7, 10, 15). Alternatively, one might interpret these same mate-

rials as primary (free of diagenetic overprints) but conclude that

their change in

O reflects a decrease in surface temperatures,

Significance

The elemental and isotopic compositions of seawater have

evolved throughout Earth

’

s history, in tandem with major cli-

matic, tectonic and biologic events, including the emergence

and diversification of life. Over geological timescales, the ox-

ygen isotope composition of seawater reflects a global balance

between mineral

–

rock reactions occurring at the Earth

’

s sur-

face (weathering and sedimentation) and crustal (hydrother-

mal alteration) environments. We put constraints on the

oxygen isotope composition of seawater throughout the

Phanerozoic and demonstrate that this value has remained

stable. This stability suggests that the fluxes of globally aver-

aged oxygen isotope exchange, associated with weathering

and hydrothermal alteration reactions, have remained pro-

portional through time and is consistent with the hypothesis

that a steady-state balance exists between seafloor hydro-

thermal activity and surface weathering.

Author contributions: U.R. and J.M.E. designed research, performed research, analyzed

data, and wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the

PNAS license

To whom correspondence should be addressed. Email: uriryb@caltech.edu.

This article contains supporting information online at

www.pnas.org/lookup/suppl/doi:10.

1073/pnas.1719681115/-/DCSupplemental

Published online June 11, 2018.

6602

–

6607

PNAS

June 26, 2018

vol. 115

no. 26

www.pnas.org/cgi/doi/10.1073/pnas.1719681115

from as high as 80 °C during the early Paleozoic, rather than a

change in the

O of seawater (12, 13). This conclusion has been

criticized for its physical implausibility (16) but has been sup-

ported for the Archean and Proterozoic by studies of Si isotopes

in cherts (17),

O of kerogens (14), and thermal stabilities of

proteins that are conserved across diverse microbial taxa (18). It

has also been argued that records of the

O of ancient altered

oceanic crust, coupled with mass balance models, require that

the

VSMOW

of seawater has been buffered to values in the

narrow range 0

‰

throughout the Phanerozoic (19). One

solution to this debate suggests that the carbonate record reflects

neither varying ocean

O nor surface temperatures; rather, it can

be attributed to diagenetic altera

tion at elevated burial tempera-

tures (20) and/or in the presence of low-

O meteoric water (21).

Part of the reason why this debate has failed to reach a con-

sensus is that the measurements in question

—

O values of

minerals

—

depend on both temperature and the

O values of

water from which those minerals grew. Carbonate clumped iso-

tope composition provides a measure of the temperature of

carbonate mineral crystallization (at least, in cases where min-

erals form at or near isotopic equilibrium). This temperature

constraint is independent of mineral

O (or

C) value; in

fact, when a clumped isotope temperature is combined with a

mineral

O value, it allows one to independently calculate the

O of coexisting fluid, by assuming crystallization occurred at

thermodynamic equilibrium and was not followed by solid-state

isotopic alteration of the clumped isotope composition (22).

Several previous studies have applied clumped isotope ther-

mometry to pre-Cenozoic fossil carbonates; briefly, these studies

have demonstrated that (

) fossils that were previously suggested

to be geochemically pristine were in fact diagenetically altered at

elevated burial temperatures (23) and (

) the least altered

(coldest clumped isotope temperatures) samples last equili-

brated at or near surface temperatures of 15

–

37 °C with waters

having

VSMOW

values of

−

1.6

–

‰

during the Ordovician

and Silurian (23

–

26), late Carboniferous (25), and late Creta-

ceous (27). These results suggest that between the late Ordovi-

cian and late Cretaceous the ocean has been stable in

Oat

near Cenozoic values and broadly similar to the range of modern

Earth surface temperatures; these studies also suggest that the

large ranges in

O of pre-Cenozoic carbonates are mostly the

result of diagenetic alteration. These interpretations were chal-

lenged (7) by a suggestion that the fossil-record

O values are

primary and that clumped isotope-based temperatures were al-

tered through closed-system solid-state isotopic reordering at

elevated burial temperatures, and therefore did not reflect the

original crystallization temperatures of calcite. Although this

argument is not consistent with observations of the coupled

variation in

O and clumped isotope temperature of some

materials, it is generally plausible (28, 29). For this reason, we

seek to generate a clumped isotope record that examines

Phanerozoic carbonate strata but focuses on a phase that is more

resistant to solid-state reordering than calcite.

We present measurements of the

C, and clumped

isotope compositions of Phanerozoic dolomites. Our dataset in-

cludes measurements of dolostone samples from the Paleozoic

stratigraphy of the Colorado Plateau (southwestern United States)

as well as published measurements of Paleozoic (30, 31), Mesozoic

(30, 32

–

34), and Cenozoic (32, 34, 35) dolostone units from var-

ious locations in North America, Europe, and the Bahamas (

Appendix

,TableS1

and

Datasets S1

and

). Dolomite in platform

carbonate sequences of the type we consider largely forms by

diagenetic reaction with seawater that circulates through sedi-

mentary basins (36). Both experiments and studies of exhumed

marbles indicate that dolomite is considerably more refractory

than calcite with respect to solid-state isotopic reordering (37, 38).

Controlled heating experiments suggest that during burial dolo-

mite retains the clumped isotope composition it had at its

formation up to burial temperatures of

∼

180 °C (39), at which

point it undergoes partial reordering; full reordering to local

equilibrium does not occur below temperatures of

∼

300 °C (39).

In contrast, calcite begins partial reordering during burial to

∼

100 °C and fully reequilibrates at temperatures of

∼

160

–

200 °C

(28, 29). Using independent constraints on peak burial tempera-

tures of the rock units in our dataset, we exclude samples that may

have experienced peak burial temperatures

180 °C (

Dataset S2

On this basis, we expect that the clumped isotope temperatures

of dolomites included in this study reflect their crystallization

temperatures. We also note that disequilibrium clumped isotope

compositions are recognized in experiments, speleothems, and

a subset of corals and deep-sea vent carbonates (40

–

43). Such

compositions appear to be the product of rapid carbonate growth

where CO

outgassing or hydration/hydroxylation are rate-limiting

processes (40, 41, 44, 45). These phenomena are unlikely to be

factors in subsurface dolomitization. This inference is supported

by recent studies that compare clumped isotope constraints on

temperatures of dolomitization to independent constraints, such

as from fluid inclusion thermometry (31, 46).

Values of

VPDB

(Vienna Pee Dee Belemnite) for dolomites

considered in this study vary from

−

13.1 to 3.9

‰

.Valuesof

for this suite range from 0.759 to 0.428

‰

(absolute reference

frame) and correspond to crystallization temperatures of 15

–

158 °C. Crystallization temperatures of dolomites are inversely

correlated with

Ovalues(Fig.1

). This trend is consistent with

dolomite crystallization at a range of burial temperatures from

water with

VSMOW

values of

−

2to12

‰

(but dominantly

−

‰

). Calculated

VSMOW

values for water in equilibrium

with dolomites are correlated with crystallization temperatures

(Fig. 1

). At near-Earth-surface temperatures (

30 °C),

water

is 0

‰

(VSMOW), consistent with the known ice-free

VSMOW

value of Cenozoic seawater (4). At higher tempera-

tures,

water

values are in excess of 2

‰

(VSMOW), consistent

with basinal waters that evolve by progressive modification of sea-

water through water

–

rock reaction at increasingly warmer burial

temperatures (47). This trend is st

atistically robust for the tightly

grouped body of 109 samples that consists mostly of Cenozoic

dolomites and makes up the densest population in Fig. 1

;the

outliers to this trend may sample

portions of basinal hydrologic

systems that have unusually high water/rock ratios (i.e., water-

buffered) or involve parent waters that were

O-depleted by mix-

ing with meteoric waters (36). However, it is important to note that

these processes alone cannot generate the observed trends between

dolomite crystallization temperature and either

VPDB

of dolo-

mite or

VSMOW

of water (

SI Appendix

,Fig.S1

). Based on

stratigraphic relationships, several of the dolomite samples

in the dataset are known to be associated with local mixing

of seawater and meteoric water (33); these data are shown in

Figs. 1 and 2 (gray filled symbols) but do not figure in our

interpretations.

Fig. 2

plots the

VPDB

values of dolomites from this study

against their stratigraphic ages, which represent the upper limit on

the age of dolomitization. The

VPDB

of dolomites increases by

∼

‰

between the Cambrian and the Pleistocene and overlaps the

calcite record (Fig. 2

). We acknowledge that our data compilation

includes a limited number of sampl

es and sampling sites. Never-

theless, a similar overlap has been observed in a comprehensive

compilation of Precambrian dolomite and calcite

Ovalues(10).

We therefore consider the overlap in

O between coexisting calcite

anddolomitetobeacommonfeatureofmarinecarbonaterocks

and suggest that the observed temporal trends in Fig. 2

reflect a

common process affecting the

O value of both mineral phases.

Keepinginmindthatthe

O value of dolomite is controlled by the

temperature of diagenetic alteration and the

O of seawater and

basinal brines (Fig. 1

and

), we suggest the temporal trend in Fig.

is the result of older rock units

’

having experienced, on average, a

greater range of burial depths/temperatures for longer time spans,

Ryb and Eiler

PNAS

June 26, 2018

vol. 115

no. 26

6603

EARTH, ATMOSPHERIC,

AND PLANETARY SCIENCES

and being more likely to have been preserved from erosion if they

were deeply buried. Consistent wit

h this interpretation, both the

average and SD of the dolomite cry

stallization temperatures de-

creases from the Paleozoic (89

33 °C) to Mesozoic (57

30 °C) to

Cenozoic (25

6 °C) rock units (Fig. 3).

The similarity between the calcite and dolomite Phanerozoic

O records in Fig. 2

suggests that calcite

O is also largely a

product of diagenetic alteration at elevated burial temperatures.

Diagenetic alteration of early Paleozoic calcite fossils was dem-

onstrated for Silurian calcite fossils from Gotland, Sweden (23).

There, samples have recorded a range of clumped isotope tem-

peratures of 33

–

62 °C while preserving the original microstruc-

tures and Mg, Mn, and Sr compositions that are found in modern

organisms and commonly used as indicators for diagenetic al-

teration. Importantly, these samples have experienced a common

thermal history during burial and exhumation, and brachiopods

and optical calcites have been shown to have common reordering

kinetics (28). Therefore, any variation in clumped isotope com-

position must derive from the calcite temperature of crystalli-

zation, while solid-state isotopic reordering, if it occurred, would

have shifted all clumped isotope temperatures toward higher

values while slightly contracting the distribution of temperatures.

A range of 29 °C is beyond the plausible variability of seawater

temperatures and therefore must include elevated burial tem-

peratures recorded by diagenetic alteration.

While the crystallization ages of dolomite we have studied are

unknown, when one considers the range of depositional ages,

dolomite crystallization temperatures, and the temperature his-

tories of the basins from which these samples come, it is clear

that the dolomites in our dataset have formed throughout the

Phanerozoic (e.g., dolomites from the Paleozoic section at the

western Colorado Plateau have formed throughout the Paleozoic

and Mesozoic;

SI Appendix

, Fig. S2

Our findings suggest that the record of

O variations for

Phanerozoic calcite fossils is largely a product of subsurface

diagenetic alteration and not a record of Earth-surface envi-

ronments. This result is inconsistent with previous interpreta-

tions which have associated low Paleozoic carbonate

O values

with low

O seawater, ascribed to a higher proportion of

weathering to hydrothermal alteration reactions, driven by a

global increase in weathering rate (5) and/or a decrease in water

circulation through midocean ridges that followed their

“

blan-

keting

”

by pelagic sediments and/or a lower sea level (2, 48).

Clumped isotope and oxygen isotope constraints on

Oof

waters parental to both dolomite and calcite in the Phanerozoic

samples we considered indicate that all samples grew from waters

that were either within the range of Cenozoic ice-free seawater

O values or higher (which we interpret as a sign of water

–

rock

reaction in basins). We find no indication for seawater lower in

VSMOW

than

−

‰

(Fig. 2

). We conclude that the evidence

we presented for diagenetic alteration of the

O records of

calcite and dolomite as well as

the constraints we offer on the

O of Phanerozoic seawater are most consistent with the uni-

formitarian hypothesis (19), that is, that the budget of oxygen

isotope exchange fluxes associated with weathering and hydro-

thermal alteration had a balance of relative strengths similar to

today

’

s throughout all of the Phanerozoic. Mass balance model

results suggest that a persistent 20% decrease in the oxygen iso-

tope flux associated with hydrothermal alteration reactions or a

persistent 100% increase in the flux associated with weather-

ing reactions (relative to their estimated values for the present)

are required to drive seawater

O value below

−

‰

;such

-15

-13

-11

-9

-7

-5

-3

-1

100

150

200

dolomite

(‰

VPDB)

Crystallization temperature (

°C

)

-6

-4

-2

100

150

200

water

(‰

VSMOW)

Crystallization

temperature (°C)

-2‰

0‰

+2‰

+4‰

+6‰

+8‰

+10‰

+12‰

W/R=

∞

W/R=5

W/R=2

W/R=1

W/R=0.6

W/R=0.3

W/R=0

-2

-1.6

-1.2

-0.8

-0.4

0.4

0.8

1.2

1.6

Count

Log W/R (crystallization T >50

°C)

Porosity

∞

Dolomitization of

low Mg calcite

Cenozoic

Mesozoic

Paleozoic (mid-late)

Paleozoic (early)

Meteoric

Fig. 1.

(

) Dolomite

O values are negatively correlated with clumped

isotope-based crystallization temperatures (

0.421,

−

0.913). Gray

dots mark dolomites which were associated with local mixing with meteoric

water (33); all other data are constrained between modeled

dolomite

compositions at equilibrium with water

VSMOW

−

‰

and

‰

(dashed lines), and dominantly (90% of the data) to

−

1to

‰

. VSMOW

(gray band) calculated using Horita (66) equation. (

) Calculated

water

positively correlated with clumped isotope-based dolomite crystallization

temperatures (

0.406,

−

0.683).

water

is calculated from measured

dolomite

and clumped isotope temperatures using Horita (66) equation.

Most data points follow the expected trend for buffering of seawater by

dolomite with

VPDB

‰

. Other samples diverge from this trend to-

ward lower

O compositions and may reflect dolomite crystallization at

higher water/rock ratios (water-buffered) or mixing with meteoric water.

Dashed lines are predicted water to rock ratio (W/R) contours (see

Methods

for details). (

) A histogram of log W/R values calculated for dolomite that

have crystallized at temperatures

50 °C. These values reflect the minimum

actual water-to-rock ratio during dolomite crystallization. W/R values are

significantly higher than the range of pore water-to-rock ratio (67) and

mostly lower than the W/R values required to fully dolomitize a low-Mg

calcite (36).

6604

www.pnas.org/cgi/doi/10.1073/pnas.1719681115

Ryb and Eiler

perturbations are contraindicated by the findings of this study.

While it is possible that the older, high-temperature dolomites

grew from parental waters that were basinal brines derived from

much lower

O initial seawaters (i.e., much lower than

−

‰

VSMOW) there is no positive evidence for such waters, and any

such scenario would require paths through temperature

–

space significantly different from that documented by the main

trend of data; for example, early Paleozoic dolomites will have a

negative offset of

∼

‰

from the general trend (marked by the

gray trend in Fig. 1

), which is not the case.

It is significant that we find the proportions of weathering and

hydrothermal alteration have remained similar through time,

despite major tectonic, climatic, and biologic perturbations (e.g.,

the assembly and breakup of the Pangea supercontinent, tran-

sitions from icehouse to greenhouse Earth climate, and the

emergence of terrestrial plants) (Fig. 2

). These perturbations

may persist for several times 10

y and have the potential to drive

a several per-mille decrease in the

O of seawater (49), yet

their time-integrated effects are not detectable. The constant

proportionality of weathering and hydrothermal exchange

through these geological changes implies that a global feedback

exists between weathering and seafloor spreading rates over

timescales of 10s of millions of years. Such feedback was pre-

dicted by the

“

spreading rate hypothesis

”

(50), in which any in-

crease in seafloor spreading rate is accompanied by a higher flux

of CO

degassing from magmatic activity in spreading centers

and subduction zones, leading to global warming, higher water

acidity, and a global increase in weathering rates. A second

possible feedback mechanism is that higher seafloor spreading

rates will be accompanied by faster plate convergence, leading

to the buildup of relief and faster erosion and weathering

rates (51).

Dolomite is abundant in pre-Cenozoic strata but mostly absent

from Cenozoic strata. This observation was referred to as the

“

dolomite problem

”

and has been attributed to the experimental

finding that uptake of Mg

by Ca-carbonate minerals is kinet-

ically limited, with a rate that depends strongly on temperature,

and is prohibitively slow at average modern Earth-surface tem-

peratures (52). Because marine temperatures are believed to

have been higher during the Paleozoic and Mesozoic compared

with the Cenozoic, cooling of marine waters and sediments are

hypothesized to have driven a decrease in dolomite formation

rate (52). This interpretation views dolomitization as an early

(i.e., shallow sediment column) diagenetic process that was rapid

and widespread in pre-Cenozoic marine sediments and slow and

rare in Cenozoic and modern settings. The Bahamas platform is

an example of a rare modern setting where locally high tem-

peratures and salinities overcome these kinetic barriers and al-

low dolomite formation (53).

The findings presented here challenge this model, as they in-

dicate that pre-Cenozoic dolomites mostly formed at tempera-

tures significantly higher than plausible Phanerozoic ocean water

or shallow sediment column conditions and commonly grew from

isotopically evolved basinal fluids rather than unmodified sea-

water. This suggests that dolomite growth is promoted by pro-

tracted deep burial and that the increased proportion of dolomite

in pre-Cenozoic strata simply reflects the increase with age in

average temperature and time of burial. This finding suggests that

dolomite is sparse in Cenozoic sediments not because of any

peculiarity in their depositional conditions or compositions but

simply because they have not yet undergone burial to deep

Dispersal

Assembly

Pangea

Cenozoic Icehouse

Mesozoic-

Early Cenozoic

Greenhouse

Hirnanian Glaciation

Late Paleozoic

Icehouse

Mid-Paleozoic

Greenhouse

Early Paleozoic

Greenhouse

Assembly

-15

-10

-5

200

400

600

O (VDPB)

Stra

graphic age (Ma)

-5

200

400

600

wa ter

(VSMOW)

Time-invariant

wa ter

Stra

graphic age (Ma)

Calcite fossils

Dolomite

Dolomite (meteoric)

Perturba

Clima

c Tectonic

Fig. 2.

The Phanerozoic carbonate and water

O records. (

)

dolomite

overlaps the

calcite

record of well-preserved calcite fossils (7) and displays

a similar

∼

‰

increase between the early Paleozoic and the present. Calcite

and dolomite are different in

Oby

∼

–

‰

when both grow from the

same water at the same temperature, but this difference is subtle at the

scale plotted. (

)

water

record calculated from clumped and bulk isotope

compositions of dolomite and well-preserved calcite fossils (23

–

27). For the

calcite fossils we include only

water

that has been associated with least-

altered specimens (23

–

27). Except for dolomites that formed from local

mixing between sea and meteoric water (gray dots) all

water

values are

−

‰

VSMOW, consistent with time-invariant seawater

O composition

(gray rectangle) and inconsistent with the proposed time variation of sea-

water

O values [dashed black line (5)].

water

that is

‰

is explained

by isotopic modification of seawater through water

–

rock reactions at

elevated burial temperatures. The observed stability in seawater

overlaps with major climatic and tectonic perturbations including the as-

sembly and breakup of Pangea and several transitions from an icehouse to

greenhouse climate.

Ryb and Eiler

PNAS

June 26, 2018

vol. 115

no. 26

6605

EARTH, ATMOSPHERIC,

AND PLANETARY SCIENCES

diagenetic settings. Assuming a seafloor temperature of 20 °C, a

crustal thermal gradient of 25 °C

−

, and sedimentation rate in

carbonate platforms of 0.01 mm

−

[expectedwhenaveragedover

timescales of 10

My (54)], Cenozoic depos

its are expected to ac-

cumulate to a maximum thickness of 650 m and to reach a maxi-

mum temperature of 36 °C, which is in a range where dolomite

formation from seawater is slow (36, 55) and below the range of

temperatures we see associated with most dolomite formation. An

implication of this hypothesis is that while Cenozoic strata gener-

ally fail to reach efficiently dolomitizing environments, dolomiti-

zation of at least some older stra

ta took place during the Cenozoic

(i.e., because they only reached deep burial in the recent geological

past). This is a testable predictio

n, as it leads to the expectation

that quantitative dating [such as by U-Pb techniques (46)] of at least

some dolomites in pre-Cenozoic strata will yield Cenozoic ages.

An open question is how Mg

enters these rocks in the first

place. Most simply, the temperatures we measure might repre-

sent the conditions at which Mg-rich fluids first convert calcite to

dolomite. In this case, carbonate platforms have been commonly

characterized by hydrological systems capable of transporting

significant quantities of dissolved Mg

to deep diagenetic en-

vironments. Using the observed

O values and crystallization

temperatures of dolomites, we estimate the minimum water to

rock ratio at the sites of dolomite crystallization (Fig. 1

and

)

(see

Methods

for details). We find that the minimum ratio of low-

O water to primary sedimentary rock in which dolomites form

is generally significantly higher than the range of typical pore

volumes of carbonate rocks (Fig. 1

), indicating that these deep

diagenetic environments have been flushed with surface waters

(or basinal fluids derived from surface waters). Although our

quantification of this effect makes use of highly simplified ar-

guments regarding the mass balance of fluid rock reaction, the

first-order conclusion is not easily explained in any other way: As

the

O values of carbonate rocks decrease consistently and

progressively with increasing temperatures of last crystallization,

they must have undergone late-diagenetic reaction with sub-

stantial volumes of water that was derived from the surface (or

shallower depths, where water

–

rock reaction buffers fluids to low

O values). Indeed, reactive transport models suggest that the

circulation of seawater can provide sufficient Mg

for massive

dolomitization at burial depths of 1

–

2.5 km (56, 57). We con-

clude that our data are consistent with the hypothesis that do-

lomitization occurs in platform carbonate sequences where

seawater circulated to kilometer-scale depths. Importantly, if

these fluids were derived from seawater, the minimum water-to-

rock ratios required by

O values of most dolomites would

have been insufficient to deliver the Mg

required to fully

convert a low-Mg calcite to dolomite (36) (Fig. 1

). This ap-

parent insufficiency may simply reflect the underestimation of

significantly larger volumes of modified (isotopically heavier)

seawater that have interacted with the rock (58). Alternatively,

the data we present are also consistent with a more complex

scenario in which Mg

was first bound in a disordered dolomite

precursor during early diagenesis, and the temperatures we ob-

serve reflect the conditions where this precursor converted to

ordered dolomite late in diagenesis, with little or no further in-

put of Mg

from solution. In this case, fluids need not deliver

deep into platform carbonate sequences. However, the

relationship between

O and formation temperature of dolo-

mite still requires that final dolomite crystallization occurred at

elevated temperatures and during reaction with large volumes of

low-

O fluid, that is, this alternative allows that the Mg budget

of dolomite-rich strata could be set by early diagenesis, but the

data still require that such rocks typically undergo deep burial,

and change in

O by reaction with surface-derived fluids at

high water/rock ratios.

Methods

We collected 33 dolostone samples from carbonate rock units at the base,

middle, and top of the Paleozoic section at the Grand Wash and the Upper

Gorge of the Grand Canyon at the southwestern Colorado Plateau and from

borehole cores from the Paradox basin in the Plateau interior (

SI Appendix

Fig. S3

). Samples were cut, cleaned, and crushed using a mortar and pestle.

Samples that contained multiple carbonate fabrics (i.e., cements, concre-

tions, and veins) were selectively sampled using a microdrill, resulting in a

total of 40 subsamples.

We analyzed the proportions of carbonate minerals (calcite, dolomite, and

aragonite) in the powdered samples using a Bruker 2D Phaser XRD system. All

samples presented here consisted of

95% dolomite.

We analyzed bulk (

C and

O) and clumped (

) isotope compositions

following the procedures described in Ghosh et al. (59), Huntington et al.

(60), and Passey et al. (61). In short, we dissolved

∼

10 mg of sample at 90 °C

in 103% phosphoric acid. Evolved CO

was separated cryogenically and pu-

rified on a gas-chromatography column. We measured masses 44

–

49 of the

purified CO

gas using a Thermo MAT253 isotope ratio mass spectrometer.

Measurements were replicated up to five times during different sessions and

on two different mass spectrometers. Heated (1,000 °C) and equilibrium

(25 °C) CO

standard gases of variable

O and

C were measured rou-

tinely to characterize and correct for the pressure baseline effect and iso-

topic

“

scrambling

”

in the ion source (62). We routinely measured in-house

carbonate standards with long-term average

values and SDs of 0.408

0.02 (CIT Carrara) and 0.655

0.02

‰

(TV04).

C, and

values were calculated following Huntington et al.

(60) and Dennis et al. (62) and assuming the Brand isotopic ratios for oxygen

(17/16 and 18/16) and carbon (13/12) in VPDB standard and slope of the

triple oxygen isotope line (

) (63).

values were calculated in the absolute

reference frame using the ClumpyCrunch v1.0 online calculator (63) and

corrected for acid fractionation by addition of 0.092

‰

(64). When the av-

erage

of standards measured in a session deviated from either of the

above values, we corrected our data to the standard using a linear transfer

function,

47,

corrected

47,

measured

, where

and

are the slope and

intercept of the standards measured versus known values,

47,

measured

is the

uncorrected value, and

47,

corrected

is the

value after standard correction.

Errors are reported as 1 SE of replicates, or for singly measured samples, the

internal measurement SE. Clumped isotope temperatures were calculated

using Bonifacie et al. (65) calibration.

Using calculated and reported clumped isotope temperatures and

values we have calculated the

O composition of water in thermodynamic

equilibrium using the equation of Horita (66). Water-to-rock ratios have

been calculated for a closed system (47) assuming an initial water

O value

of 0

‰

(VSMOW) and initial rock

O value of 3.6

‰

(VPDB) which corresponds

to dolomite in equilibrium with the assumed initial water at 15 °C

—

the

minimum temperature observed in our dataset. Low-temperature dolomites

have undergone relatively minor diagenetic modification of

Ovalues(Fig.

). The calculation of water-to-rock ratio for these samples is very sensitive to

the assumption of initial water and rock compositions and therefore highly

uncertain (as implied by the convergence of W/R contours toward the assumed

initial water

O value and temperature in Fig. 1

). To avoid this uncertainty,

we exclude from this analysis dolomites that have crystallized at tem-

peratures

50 °C. Importantly, the results of this zero-dimension analysis

should be regarded as minimum constrains on the actual water-to-rock

ratio in the dolomite crystallization sites (58). Samples with

water

0.0

0.1

0.2

0.3

0.4

0.5

100

110

120

130

140

150

160

Normalized frequency

alliza

on temperature (°C)

Cenozoic (n=82)

Mesozoic (n=15)

Paleozoic (n=57)

Fig. 3.

Distributions of dolomite crystallization temperatures by eras. Av-

erage and SD of crystallization temperatures decrease from the Paleozoic to

the Mesozoic and Cenozoic.

6606

www.pnas.org/cgi/doi/10.1073/pnas.1719681115

Ryb and Eiler

values lower than the minimum value permitted under the assumptions

of this analysis (

‰

VSMOW) are considered water-buffered (W/R

∞

ACKNOWLEDGMENTS.

We thank Yael Kiro, Max K. Lloyd, and Alex Lipp for

assisting in the field and with sample p

reparation procedures; two anonymous

reviewers for their detailed and constructive comments; and the Grand

Canyon National Park and the US Geological Survey Core Research Center for

facilitating sample collection. This work was supported by NSF Grant EAR-

1624827 (to J.M.E.). U.R. was supported by an O. K. Earl fellowship during

this study.

1. Muehlenbachs K (1986) Alteration of the oceanic-crust and the O-18 history of sea-

water.

Rev Miner

16:425

–

444.

2. Kasting JF, et al. (2006) Paleoclimates, ocean depth, and the oxygen isotopic com-

position of seawater.

Earth Planet Sci Lett

252:82

–

93.

3. Bowers TS, Taylor HP (1985) An integrated chemical and stable-isotope model of the

origin of midocean ridge hot spring systems.

JGeophysResSolidEarth

90:12583

–

12606.

4. Zachos J, Pagani M, Sloan L, Thomas E, Billups K (2001) Trends, rhythms, and aber-

rations in global climate 65 Ma to present.

Science

292:686

–

693.

5. Wallmann K (2001) The geological water cycle and the evolution of marine

18O

values.

Geochim Cosmochim Acta

65:2469

–

2485.

6. Muehlenbachs K, Clayton RN (1976) Oxygen isotope composition of oceanic-crust and

its bearing on seawater.

J Geophys Res

81:4365

–

4369.

7. Veizer J, Prokoph A (2015) Temperatures and oxygen isotopic composition of Phan-

erozoic oceans.

Earth Sci Rev

146:92

–

104.

8. Veizer J, et al. (1999) Sr-87/Sr-86, delta C-13 and delta O-18 evolution of Phanerozoic

seawater.

Chem Geol

161:59

–

88.

9. Prokoph A, Shields GA, Veizer J (2008) Compilation and time-series analysis of a

marine carbonate delta(18)O, delta(13)C, (87)Sr/(86)Sr and delta(34)S database

through Earth history.

Earth Sci Rev

87:113

–

133.

10. Jaffrés JBD, Shields GA, Wallmann K (2007) The oxygen isotope evolution of seawa-

ter: A critical review of a long-standing controversy and an improved geological

water cycle model for the past 3.4 billion years.

Earth Sci Rev

83:83

–

122.

11. Kolodny Y, Epstein S (1976) Stable isotope geochemistry of deep-sea cherts.

Geochim

Cosmochim Acta

40:1195

–

1209.

12. Knauth LP, Epstein S (1976) Hydrogen and oxygen isotope ratios in nodular and

bedded cherts.

Geochim Cosmochim Acta

40:1095

–

1108.

13. Karhu J, Epstein S (1986) The implication of the oxygen isotope records in coexisting

cherts and phosphates.

Geochim Cosmochim Acta

50:1745

–

1756.

14. Tartèse R, Chaussidon M, Gurenko A, Delarue F, Robert F (2017) Warm Archean

oceans reconstructed from oxygen isotope composition of early-life remnants.

Geochem Perspect Lett

3:55

–

65.

15. Veizer J, Godderis Y, François LM (2000) Evidence for decoupling of atmospheric CO2

and global climate during the Phanerozoic eon.

Nature

408:698

–

701.

16. Raymond A, Metz C (2004) Ice and its consequences: Glaciation in the Late Ordovician,

Late Devonian, Pennsylvanian-Permian, and Cenozoic compared.

J Geol

112:655

–

670.

17. Robert F, Chaussidon M (2006) A palaeotemperature curve for the Precambrian

oceans based on silicon isotopes in cherts.

Nature

443:969

–

972.

18. Gaucher EA, Govindarajan S, Ganesh OK (2008) Palaeotemperature trend for Pre-

cambrian life inferred from resurrected proteins.

Nature

451:704

–

707.

19. Muehlenbachs K (1998) The oxygen isotopic composition of the oceans, sediments

and the seafloor.

Chem Geol

145:263

–

273.

20. Land LS (1995) Oxygen and carbon isotopic composition of Ordovician brachiopods

–

Implications for coeval seawater

–

Comment.

Geochim Cosmochim Acta

59:2843

–

2844.

21. Dyer B, Maloof AC, Higgins JA (2015) Glacioeustasy, meteoric diagenesis, and the carbon

cycle during the Middle Carboniferous.

Geochem Geophys Geosyst

16:3383

–

3399.

22. Eiler JM (2007)

“

Clumped-isotope

”

geochemistry

–

The study of naturally-occurring,

multiply-substituted isotopologues.

Earth Planet Sci Lett

262:309

–

327.

23. Cummins RC, Finnegan S, Fike DA, Eiler JM, Fischer WW (2014) Carbonate clumped

isotope constraints on Silurian ocean temperature and seawater delta O-18.

Geochim

Cosmochim Acta

140:241

–

258.

24. Finnegan S, et al. (2011) The magnitude and duration of Late Ordovician-Early Silu-

rian glaciation.

Science

331:903

–

906.

25. Came RE, et al. (2007) Coupling of surface temperatures and atmospheric CO2 con-

centrations during the Palaeozoic era.

Nature

449:198

–

201.

26. Bergmann KD, et al. (2018) A paired apatite and calcite clumped isotope thermom-

etry approach to estimating Cambro-Ordovician seawater temperatures and isotopic

composition.

Geochim Cosmochim Acta

224:18

–

41.

27. Dennis KJ, Cochran JK, Landman NH, Schrag DP (2013) The climate of the Late Cre-

taceous: New insights from the application of the carbonate clumped isotope ther-

mometer to Western Interior Seaway macrofossil.

Earth Planet Sci Lett

362:51

–

65.

28. Stolper DA, Eiler JM (2015) The kinetics of solid-state isotope-exchange reactions for

clumped-isotopes: A study of inorganic calcites and apatites from natural and ex-

perimental samples.

Am J Sci

315:363

–

411.

29. Henkes GA, et al. (2014) Temperature limits for preservation of primary calcite

clumped isotope paleotemperatures.

Geochim Cosmochim Acta

139:362

–

382.

30. Geske A, et al. (2015) The magnesium isotope (delta Mg-26) signature of dolomites.

Geochim Cosmochim Acta

149:131

–

151.

31. Came RE, Azmy K, Tripati A, Olanipekun BJ (2017) Comparison of clumped isotope

signatures of dolomite cements to fluid inclusion thermometry in the temperature

range of 73-176 degrees C.

Geochim Cosmochim Acta

199:31

–

47.

32. Loyd SJ, Corsetti FA, Eiler JM, Tripati AK (2012) Determining the diagenetic conditions

of concretion formation: Assessing temperatures and pore waters using clumped

isotopes.

J Sediment Res

82:1006

–

1016.

33. Dale A, John CM, Mozley PS, Smalley PC, Muggeridge AH (2014) Time-capsule con-

cretions: Unlocking burial diagenetic processes in the Mancos Shale using carbonate

clumped isotopes.

Earth Planet Sci Lett

394:30

–

37.

34. Winkelstern IZ, Lohmann KC (2016) Shallow burial alteration of dolomite and lime-

stone clumped isotope geochemistry.

Geology

44:467

–

470.

35. Murray ST, Swart PK (2017) Evaluating formati

on fluid models and calibrations using clumped

isotope paleothermometry on Bahamian dolomites.

Geochim Cosmochim Acta

206:73

–

93.

36. Machel HG (2004)

The Geometry and Petrogenesis of Dolomite Hydrocarbon Reservoirs

,eds

Braithwaite CJR, Rizzi G, Darke G (Geol

ogical Society of London, London), Vol 235, pp 7

–

63.

37. Ryb U, Lloyd MK, Stolper DA, Eiler JM (2017) The clumped-isotope geochemistry of

exhumed marbles from Naxos, Greece.

Earth Planet Sci Lett

470:1

–

12.

38. Lloyd MK, Eiler JM, Nabelek PI (2017) Clu

mped isotope thermometry of calcite and do-

lomite in a contact metamorphic environment.

Geochim Cosmochim Acta

197:323

–

344.

39. Lloyd MK, Ryb U, Eiler JM (2017)

Experimental Determination of Dolomite

Δ

Solid-

State Reordering Rates and the Preservation Potential of the Dolomite

Δ

Ther-

mometer

, Sixth International Clumped Isotope Workshop. Available at www.ipgp.fr/

sites/default/files/6thiciw_program.pdf. Accessed May 31, 2018. Sixth International

Clumped Isotope Workshop (Paris).

40. Daëron M, et al. (2011) (CO)-C-13-O-18 clumping in speleothems: Observations from

natural caves and precipitation experiments.

Geochim Cosmochim Acta

75:3303

–

3317.

41. Affek HP, et al. (2014) Accounting for kinetic isotope effects in Soreq Cave (Israel)

speleothems.

Geochim Cosmochim Acta

143:303

–

318.

42. Saenger C, et al. (2012) Carbonate clumped isotope variability in shallow water corals:

Temperature dependence and growth-related vital effects.

Geochim Cosmochim Acta

99:224

–

242.

43. Loyd SJ, et al. (2016) Methane seep carbonates yield clumped isotope signatures out

of equilibrium with formation temperatures.

Nat Commun

7:12274.

44. Guo W, et al. (2008) C-13-O-18 bonds in dissolved inorganic carbon: Implications for

carbonate clumped isotope thermometry.

Geochim Cosmochim Acta

72:A336.

45. Tripati AK, et al. (2015) Beyond temperature: Clumped isotope signatures in dissolved

inorganic carbon species and the influence of solution chemistry on carbonate min-

eral composition.

Geochim Cosmochim Acta

166:344

–

371.

46. Mangenot X, Gasparrini M, Bonifacie M, Gerdes A, Rouchon V (2017)

An Emerging

Thermo-Chronometer to Resolve Longstanding Enigmas in Sedimentary Basin Anal-

ysis: D47/(U-Pb)

, Goldschmidt. Available at: https://goldschmidt.info/2017/abstracts/

abstractView?id

2017003775. Accessed May 31, 2018.

47. Taylor HP (1974) The application of oxygen and hydrogen isotope studies to problems

of hydrothermal alteration and ore deposition.

Econ Geol

69:843

–

883.

48. Wallmann K (2004) Impact of atmospheric CO2 and galactic cosmic radiation on Phaner-

ozoic climate change and the marine

18O record.

Geochem Geophys Geosyst

5:Q06004.

49. Walker JCG, Lohmann KC (1989) Why the oxygen isotopic composition of sea-water

changes with time.

Geophys Res Lett

16:323

–

326.

50. Brener RA, Lasaga AC, Garrels RM (1983) The carbonate-silicate geochemical cycle and its

effect on atmospheric carbon dioxide over the past 100 million years.

Am J Sci

283:641

–

683.

51. Raymo ME, Ruddiman WF (1992) Tectonic forcing of late Cenozoic climate.

Nature

359:117

–

122.

52. Holland HD, Zimmerman H (2000) The dolomite problem revisited.

Int Geol Rev

42:481

–

490.

53. Swart PK (2015) The geochemistry of carbonate diagenesis: The past, present and

future.

Sedimentology

62:1233

–

1304.

54. Kemp DB, Sadler PM (2014) Climatic and eustatic signals in a global compilation of

shallow marine carbonate accumulation rates.

Sedimentology

61:1286

–

1297.

55. Land LS (1998) Failure to precipitate dolomite at 25 degrees C from dilute solution

despite 1000-fold oversaturation after 32 years.

Aquat Geochem

4:361

–

368.

56. Wilson AM, Sanford V, Whitaker F, Smart P (2001) Spatial patterns of diagenesis

during geothermal circulation in carbonate platforms.

Am J Sci

301:727

–

752.

57. Whitaker FF, Xiao YT (2010) Reactive transport modeling of early burial dolomitiza-

tion of carbonate platforms by geothermal convection.

AAPG Bull

94:889

–

917.

58. Sharp Z (2007)

Principles of Stable Isotope Geochemistry

(Prentice Hall, Upper Saddle

River, NJ), 1st Ed.

59. Ghosh P, et al. (2006) (13)C-(18)O bonds in carbonate minerals: A new kind of pale-

othermometer.

Geochim Cosmochim Acta

70:1439

–

1456.

60. Huntington KW, et al. (2009) Methods and limitations of

‘

clumped

’

CO2 isotope (

47)

analysis by gas-source isotope ratio mass spectrometry.

J Mass Spectrom

44:

1318

–

1329.

61. Passey BH, Levin NE, Cerling TE, Brown FH, Eiler JM (2010) High-temperature envi-

ronments of human evolution in East Africa based on bond ordering in paleosol

carbonates.

Proc Natl Acad Sci USA

107:11245

–

11249.

62. Dennis KJ, Affek HP, Passey BH, Schrag DP, Eiler JM (2011) Defining an absolute reference

frame for

‘

clumped

’

isotope studies of CO2.

Geochim Cosmochim Acta

75:7117

–

7131.

63. Daëron M, Blamart D, Peral M, Affek HP (2016) Absolute isotopic abundance ratios

and the accuracy of Delta(47) measurements.

Chem Geol

442:83

–

96.

64. Henkes GA, et al. (2013) Carbonate clumped isotope compositions of modern marine

mollusk and brachiopod shells.

Geochim Cosmochim Acta

106:307

–

325.

65. Bonifacie M, et al. (2017) Calibration of the dolomite clumped isotope thermometer

from 25 to 350 °C, and implications for a universal calibration for all (Ca, Mg, Fe)CO3

carbonates.

Geochim Cosmochim Acta

200:255

–

279.

66. Horita J (2014) Oxygen and carbon isotope fractionation in the system dolomite-

water-CO2 to elevated temperatures.

Geochim Cosmochim Acta

129:111

–

124.

67. Moore CH, Wade WJ (2013)

Developments in Sedimentology

, eds Moore CH,

Wade WJ (Elsevier, New York), Vol 67, p 49.

Ryb and Eiler

PNAS

June 26, 2018

vol. 115

no. 26

6607

EARTH, ATMOSPHERIC,

AND PLANETARY SCIENCES