You are on page 1of 21

Earth-Science Reviews 208 (2020) 103293

Contents lists available at ScienceDirect

Earth-Science Reviews
journal homepage: www.elsevier.com/locate/earscirev

Sea-level fluctuations driven by changes in global ocean basin volume T


following supercontinent break-up
Nicky M. Wrighta, ,1, Maria Setona, Simon E. Williamsa,2, Joanne M. Whittakerb,

R. Dietmar Müllera
a
EarthByte Group, School of Geosciences, University of Sydney, NSW, Australia
b
Institute for Marine and Antarctic Studies, University of Tasmania, TAS, Australia

ARTICLE INFO ABSTRACT

Keywords: Long-term variations in eustatic sea level in an ice-free world, which existed through most of the Mesozoic and
Sea level early Cenozoic eras, are partly driven by changes in the volume of ocean basins. Previous studies have de-
Plate tectonics termined ocean basin volume changes from plate tectonic reconstructions since the Mesozoic; however, these
Ocean basins studies have not considered a number of important elements that contribute to ocean basin volume, such as
Seafloor spreading
regional differences in sedimentation, or uncertainties within the plate tectonic model itself, such as spreading
Paleobathymetry
asymmetries and the incomplete representation of back-arc basins in the Mesozoic. Additionally, studies on long-
term changes in sea level related to the extension and rifting of passive margins have not been performed on a
global-scale and likely significantly underestimated the influence of this process. In order to improve re-
constructions of sea level on geologic time scales and assess the uncertainty in deriving the volume of ocean
basins based on a global plate kinematic model, we investigate the influence of back-arc basins, spreading
asymmetry, large igneous provinces (LIPs), sediment thickness, and passive margins on ocean basin volume since
200 Ma. We find that less-constrained plate tectonic elements, such as the presence of back-arc basins or
spreading asymmetry, may contribute up to ~120 m or ~150 m to sea level respectively. Changes in the sea
level related to sedimentation and LIPs are respectively ~75–165 m and ~45 m. Changes in sea level associated
with passive margin formation are almost negligible at present day, though were much larger in the Cretaceous,
and the assumed sedimentation style strongly influences the rate and magnitude of sea-level change. We in-
corporate predictions for these components during times where ocean basins are predominantly synthetic re-
constructions and find that sea level driven by fluctuating ocean basin volume has changed by ~200 m since the
Jurassic, which is comparable to previous estimates. Our revised estimates will need to be combined with other
processes driving long-term sea-level change, including mantle convection-driven dynamic topography and
glacio-eustasy for constructing a complete eustatic sea-level curve. Understanding and quantifying the un-
certainties in the volume of ocean basins has implications for modelling subduction flux, the oceanic carbon
cycle, and heatflow, and is important for exploring Earth's evolutionary cycles, especially during times in the
geologic past where much of the ocean basin history has been lost.

1. Introduction Miller et al., 2005), which in turn has been linked with variations in
biodiversity and minor extinctions (Tennant et al., 2016), climate var-
Ocean basins comprise ~71% of Earth's surface at present day, iations (e.g., Mitrovica et al., 2001) and continental flooding (e.g.,
however, this value has fluctuated throughout supercontinent cycles. Miller et al., 2005; Harrison, 1990; Bond, 1979). Estimates for changes
Understanding how ocean basin volume has varied through time has in ocean basin volume through time can be combined with other pro-
both geodynamic and climatic implications. For example, changes in cesses driving eustasy, particularly dynamic topography and glacio-
mid-ocean ridge and ocean basin volume are thought to drive changes eustasy (see review by Müller et al., 2018b), to construct a global sea-
in long-term (106 to 108 years) eustatic sea level (e.g., Harrison, 1990; level curve which takes into account the major driving forces


Corresponding author.
E-mail address: nicky.wright@anu.edu.au (N.M. Wright).
1
Now at the ARC Centre of Excellence for Climate Extremes and Research School of Earth Sciences, the Australian National University, ACT, Australia.
2
Now at the State Key Laboratory of Continental Dynamics, Department of Geology, Northwest University, Xi'an, China.

https://doi.org/10.1016/j.earscirev.2020.103293
Received 24 February 2020; Received in revised form 3 July 2020; Accepted 16 July 2020
Available online 20 July 2020
0012-8252/ © 2020 Elsevier B.V. All rights reserved.
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

modulating regional and global sea level. In addition, paleogeographic reconstruction methods (for example, the timing of the peak of the
reconstructions that include paleobathymetry provide important con- Cretaceous highstand varies by 30–40 million years between different
textual information for understanding possible pathways for past ocean estimates; Müller et al., 2018b). It is widely accepted that first-order
circulation and species migration, and serve as important boundary fluctuations in global sea level are primarily driven by changes in global
conditions for deep-time paleoclimate models. ocean basin volume (Kominz, 2001, 2013) and can be modelled based
Variations in global sea level are due to the combined effects of: (1) on plate tectonic and seafloor isochron reconstructions (e.g., Müller
changes in the volume of ocean basins; and (2) changes in the volume of et al., 2008b; Xu et al., 2006; Vérard et al., 2015), resulting in paleo-
water in the oceans. Changes in the volume of ocean basins are largely oceanic age grids that can be converted to basement depth using an age-
driven by variations in the volume of the global ridge system, which is depth relationship, i.e., Stein and Stein (1992) (used in Müller et al.,
inherently connected to the formation and breakup of supercontinents 2008b; Xu et al., 2006) or Turcotte and Schubert (2002) (used in Vérard
via plate tectonics and seafloor spreading (e.g., Hays and Pitman, 1973; et al., 2015). Other factors that influence seafloor depth can ad-
Pitman, 1978; Kominz, 1984; Cogné et al., 2006; Xu et al., 2006; Müller ditionally be incorporated, such as sediment accumulation, and LIPs
et al., 2008b). Such processes drive long-term (106 to 108 years) (Müller et al., 2008b; Vérard et al., 2015). However, assumptions in the
changes in eustatic sea level, with variations of several hundred metres. plate tectonic model, including plate configuration, timing and extent
Other commonly considered processes that influence the volume of of back-arc basins, and symmetrical spreading, may greatly influence
ocean basins (but to a smaller extent) include marine sediment accu- the trends observed in such reconstructions of sea level, and partly
mulation (Harrison, 1990; Müller et al., 2008b; Dutkiewicz et al., explain why such curves do not match those based on sequence stra-
2017), seafloor volcanism including the emplacement of large igneous tigraphy (e.g., Haq et al., 1987). An opposing view has been put for-
provinces (LIPs) (Harrison, 1990; Müller et al., 2008b; Conrad, 2013), ward by Rowley (2002), who hypothesised that the presently observed
dynamic topography which modulates the elevation of the continents linear decrease in area versus age of the ocean floor (called a triangular
relative to the oceans (Spasojevic and Gurnis, 2012; Conrad, 2013; distribution) implies a steady-state of seafloor spreading and subduc-
Conrad and Husson, 2009) and thermal uplift associated with super- tion, and consequently of ocean basin volume. This interpretation of the
continent insulation and/or plume head arrival (Guillaume et al., preserved seafloor spreading record was debunked by Demicco (2004)
2016). Changes in the volume of water in the oceans are driven by who demonstrated that the decreasing area with increasing age of
glaciation (e.g., Abreu and Anderson, 1998; Miller et al., 2005), thermal preserved ocean floor does not necessitate a steady-state model of
expansion of water, and periodic water sequestration (Cloetingh and ocean-floor spreading and destruction through time. Both mantle con-
Haq, 2015). Variations in sea level driven by ice-volume change occurs vection models and tectonic ocean basin reconstructions result in major
on much shorter timescales (generally up to 105 years), and has been changes in the age-area distribution of ocean floor between rectangular,
explored in detail for more recent times (e.g., Lambeck et al., 2002). triangular and skewed distributions (Coltice et al., 2013), suggesting
Long-term changes in eustatic sea level have been explored using a that long-term global tectonic cycles result in significant fluctuations in
variety of approaches since the 1970s (Table 1). In general, studies tend mid-ocean ridge length, spreading rates and thus ocean basin volume
to focus on the contribution of one or several components that influence through time.
sea level (such as changes in mid-ocean ridge volume, or variations in Sequence stratigraphy has traditionally been used to determine
continental flooding) rather than combining many processes into one changes in sea level (e.g., Haq et al., 1987; Haq and Al-Qahtani, 2005),
analysis. Despite variations in approach, the broad, long-term trend in which broadly relies on changes in the accommodation space of a basin
the various sea-level reconstructions agree (for example, sea level was and, to a lesser extent, changes in the sediment supply. However, se-
higher in the Cretaceous relative to present day; Fig. 1), even though quence stratigraphy is restricted to constraining second- and third-order
the magnitude and timing can vary between the different relative sea-level changes (Kominz, 2001); therefore, first-order eustasy

Table 1
Summary of long-term eustatic sea-level curves since the Mesozoic.
Reference Age (Ma) Brief description

Vail et al. (1977) ~200–0 Based on seismic stratigraphy


Pitman (1978) 85–15 Based on changes in mid-ocean ridge volume
Watts and Steckler (1979) ~160–0 Based on determining the subsidence history and separating ‘tectonic’ and ‘eustatic’ effects for five wells along the Atlantic
continental margin
Hallam (1984) ~541–0 Based on areal plots of the changes in the temporal distribution of marine deposits, along with sequence stratigraphy and other factors
Kominz (1984) 80–0 Based on changes in mid-ocean ridge volume
Watts and Thorne (1984) ~160–0 Based on the analysis of tectonic subsidence from the US Atlantic margin
Haq et al. (1987) 255–0 Based on a synthesis of sequence stratigraphic data
Harrison (1990) 180–0 Determined from changes in ocean basin volume for 80–0 Ma; amount of continental volume from 180 to 0 Ma
Miller et al. (2005) 100–0 Relies on backstripping stratigraphic data from New Jersey coastal plain coreholes (100–7 Ma) and benthic foraminiferal oxygen
isotopes (δ18O; 9–0 Ma)
Xu et al. (2006) 62–0 Based on the reconstruction of seafloor age using two different tectonic models
Kominz et al. (2008) 108–0 Based on stratigraphic data from the New Jersey margin. Also includes a sea-level curve with corrections for subsidence from dynamic
topography
Müller et al. (2008b) 140–0 Based on a global plate tectonic reconstruction (Müller et al., 2008a), includes sediment thickness and LIP emplacement
Snedden and Liu (2010) 550–0 Based on sequence stratigraphic analysis
Spasojevic and Gurnis (2012) 90–0 Based on dynamic topography models and a global plate tectonic reconstruction
Conrad (2013) 140–0 Based on global plate tectonic reconstruction (Müller et al., 2008a), includes sediment thickness, seafloor volcanism, dynamic
topography, ocean area, and climatic (glaciation and seawater expansion) components
Haq (2014) 152–59 Based on a synthesis of global stratigraphic data
Vérard et al. (2015) 550–0 Based on synthetically created topography underlain by geodynamical maps created in 5–20 Myr intervals
Haq (2017) 205–140 Based on sequence stratigraphic data, with updated chronostratigraphy
van der Meer et al. (2017) 835–0 Based on 87Sr/86Sr ratios, with compensation for weathering estimates
Müller et al. (2018b) 140–0 Based on global plate tectonic reconstruction (Müller et al., 2016), includes modelled dynamic topography and glacio-eustasy
Karlsen et al. (2019) 230–0 Based on global plate tectonic reconstruction (Müller et al., 2016), includes influence of sediment thickness, seafloor volcanism,
dynamic topography, ocean area, climate (glaciation and seawater expansion) and global water fluxes between oceans and the mantle

2
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

Fig. 1. Summary of some key sea-level reconstructions. Note that no attempt to convert for various time-scales has been made. The sea-level curve from Müller et al.
(2018b) is model M7 which includes dynamic topography and glacio-eustasy. Kominz et al. (2008) includes corrections for dynamic topography. + Vérard et al.
(2015) relies on underlying geodynamic maps, however, specific details on the plate reconstruction are unavailable. * Karlsen et al. (2019) rely on the global plate
reconstruction of Müller et al. (2016), with additional components including deep water cycling and dynamic topography. Note that their sea-level curve does not
incorporate changes in ocean basin area.

is ill-constrained by sequence stratigraphic interpretations. A further volume of ocean basins and sea-level reconstructions. Many studies
drawback of this approach is that it assumes variations in sediment (e.g., Spasojevic and Gurnis, 2012; Conrad, 2013; Conrad and Husson,
supply play an insignificant role on fluctuations (van der Meer et al., 2009; Müller et al., 2018b) have explored how dynamic topography
2017). Stratigraphic data from passive margins have been used to re- may influence sea level, however, it is difficult to constrain the mag-
construct eustatic sea level (i.e., Miller et al., 2005; Kominz et al., nitude of this effect. Estimates based on different types of geodynamic
2008)—while this is a valuable approach and avoids the large un- models result in changes in mean oceanic dynamic topography from
certainties associated with past ridge volume estimates, regional tec- 100 to 150 m over the past 140 Ma (Müller et al., 2018b). Dynamic
tonic activity, especially time-dependent dynamic topography driven topography may also influence regions thought to be tectonically
by interaction with continents with the convecting mantle may have stable. Along passive margins the modelled dynamic topographic
influenced such locations (Müller et al., 2018b). Alternative approaches change over the last 140 Ma can reach 350 ( ± 150) m of dynamic
to deriving sea level have been explored, such as changes in strontium subsidence (Müller et al., 2018b).
isotopes ratios (87Sr/86Sr) (e.g., van der Meer et al., 2017), and inferred Here, we explore the variations in ocean basin volume since the
changes in the volume and depth of ocean basins based on paleogeo- Jurassic based on a recent plate tectonic reconstruction (Müller et al.,
graphic estimates of inundations and paleoshorelines (e.g., Rowley, 2019; Fig. 2), and assess the contribution of different components (such
2017). In particular, 87Sr/86Sr isotope ratios in seawater, which reflects as sediment thickness, LIPs, spreading asymmetry, and back-arc basins)
the mixing of input from continental runoff and hydrothermal altera- on long-term sea-level trends. We largely focus on the oceanic crustal
tion of seafloor basalt (i.e., from mid-ocean ridges and other volcanics) component of basins (i.e., Sections 2.1 to 2.3 focuses solely on the
(Spooner, 1976), can be used as an independent method to explore modelled oceanic crustal portion), although we briefly explore the in-
changes in sea level, especially during times where oceanic crust cannot fluence of changes in continental crust associated with passive margins
be directly reconstructed. Early work suggests a simple correlation (Section 2.4). For most of our analysis, we determine changes in ocean
between seawater 87Sr/86Sr isotope ratios and land area since ~70 Ma basin volume by incorporating both changes in the area and depth, i.e.,
(Spooner, 1976), although this relationship does not apply when con- we do not assume that the area of ocean basins has remained constant
sidering the entire Phanerozoic (Hallam, 1984). Estimates based on a through time, although we do assume that the Earth's radius has re-
Phanerozoic plate tectonic model and synthetically derived topography mained constant over long timescales. We produce paleobathymetry
and bathymetry shows a relationship with the 87Sr/86Sr isotope record grids (Fig. 3) from 230 Ma to present day in 1 Myr increments, and use
and volume of mountain belts, which in turn is driven by tectonics our derived paleobathymetry reconstruction to produce a 200 Myr sea-
(Vérard et al., 2015). Recent work by van der Meer et al. (2017) in- level curve (i.e., from supercontinent through to break-up and dis-
vestigates sea-level variations using the mantle-derived component of persal). We additionally explore elements that are less constrained in
87
Sr/86Sr isotope records, by removing the effect of the continental the plate kinematic model, such as the influence of back-arc basins and
weathering using runoff estimates from climate simulations. While this variations in seafloor spreading asymmetry, and investigate their in-
approach includes a number of assumptions and possible circulatory (as fluence on sea level.
the paleogeographies used in the climate simulations have a sea level This paper is organised as follows. In Section 2, we explore the in-
embedded in them), it provides a unique attempt at reconstructing fluence of each factor that contributes to ocean basin volume and sea
eustatic sea level over long time periods. level, where within each subsection we include both the methodology
Dynamic topography, the mantle's influence on Earth's surface to- for calculating the contribution for this factor, and the results from this
pography, is an important solid Earth process that can influence the factor in isolation and with some discussion of the limitations. In

3
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

Fig. 2. Global plate reconstruction and age-area distribution of oceanic crust (paleo-age grids) from Müller et al. (2019), shown in 40 Myr increments. Dark grey
areas correspond to passive margins and other submerged, non-oceanic crust such as microcontinents and island arcs. Light grey areas refer to present-day coastlines.

Section 3, we bring together the various results outlined in Section 2, conjugate spreading flanks which can be relatively well constrained by
computing a range of possible combined sea-level change curves and the preserved spreading flank in the Pacific Ocean basin). In particular,
comparing these to some of the previous estimates in the literature, and there are large uncertainties in the history of the pre-Cenozoic Pacific
further discuss limitations. Finally, Section 4 presents our conclusions. Ocean basin, and its precursor, Panthalassa, as well as along continental
margins where the history of back-arc basins prior to the Cenozoic is
limited.
2. Factors that influence the volume of ocean basins
Early attempts to calculate the influence of ocean basins on sea-level
change have focused on variations in mid-ocean ridge volume, notably
2.1. Plate tectonics and the volume of mid-ocean ridges
by Pitman (1978) and more extensively by Kominz (1984). These stu-
dies were based on the mid-ocean ridge itself by determining ridge
Changes in ridge volume through time due to plate tectonics forms a
lengths and spreading rates, and applying an age-depth relationship to
major constraint on ocean basin volume and sea level, as the depth of
calculate volume and consequently do not incorporate changes in the
the seafloor increases with distance from the ridge as the oceanic li-
ocean basin area. More recent studies (i.e., Müller et al., 2008b) have
thosphere cools and thickens. As such, plate tectonic reconstructions
relied on determining the oceanic-age distribution (commonly referred
are frequently used to construct models for the volume of ocean basin,
to as ‘age grids’) through time and applying an age-depth relationship
based on the reconstruction of seafloor ages due to seafloor spreading
to directly determine changes in the volume of ocean basins—notably,
and mid-ocean ridge geometry and the conversion of seafloor age to
this approach also accounts for changes in ocean area.
basement depth using a published age-depth relationship (including
Age-depth models
relationships outlined in Parsons and Sclater, 1977; Stein and Stein,
There are numerous published seafloor age-depth models, including
1992; Crosby et al., 2006, etc.). For ocean basins where both spreading
(but not limited to) Parsons and Sclater (1977), Stein and Stein (1992)
flanks are preserved (such as the Atlantic Ocean), it is relatively
(‘GDH1’), Crosby et al. (2006), Crosby and McKenzie (2009), Hasterok
straightforward to establish its spreading history using magnetic
(2013), Richards et al. (2018), and Rowley (2019). We compare se-
anomaly identifications (e.g., Seton et al., 2014) and the seafloor fabric
lected age-depth models (Fig. 5), and find that GDH1 and Crosby and
(e.g., Matthews et al., 2011; Wessel et al., 2015). In cases where only
McKenzie (2009) suggest very similar sea-level trends since the Triassic,
one spreading flank remains (such as portions of the Pacific Ocean
as does Parsons and Sclater (1977) and Richards et al. (2018), while
basin), the seafloor fabric and magnetic anomaly identifications from
there is a ~100 m difference in sea level during the Cretaceous between
the preserved tectonic plate are used to determine the seafloor
GDH1/ Crosby and McKenzie (2009) and Parsons and Sclater (1977)/
spreading history for both spreading flanks, and symmetrical spreading
Richards et al. (2018). Assessing the differences between the age-depth
is assumed. However, it is significantly more difficult to determine past
relationships themselves are beyond the scope of this study. However,
spreading histories when both spreading flanks have been sub-
we rely on the age-depth relationship from GDH1 (Stein and Stein,
ducted—plate boundaries are inferred from the onshore geological re-
1992), as it has previously been shown to be preferable for re-
cord (e.g., terrane boundaries, major sutures), and the spreading di-
constructing paleo-basement depths and thermally rejuvenated litho-
rection and rate are assumed to be similar to the preserved crust (Müller
sphere (Müller et al., 2008b) (which better describes the behaviour of
et al., 2016), resulting in large uncertainties in the past plate config-
the older oceanic crust that comprises much of ocean basins through
uration itself. World uncertainty, i.e., the fraction of Earth's lithosphere
time), and it also allows for a simpler comparison with many previously
that has been subducted since a given time (Torsvik et al., 2010), may
published sea-level curves (e.g., Müller et al., 2008b). We assume that
give an idea of how uncertain a plate tectonic reconstruction is at a
the relationship between age and depth of the seafloor has remained the
particular time in the past (Fig. 4). For investigating uncertainties in
same throughout time; however, we note that geodynamic influences
ocean basin volume, the percentage of present-day oceanic crust re-
such as the Pacific superplume (East et al., 2020) may have altered this
constructed through time provides an additional measure as to how
relationship during times when the paleo-Pacific/Pathalassa was the
much a particular time period may rely on synthetically reconstructed
dominant ocean basin, but this possible impact is difficult to constrain.
ocean crust (Fig. 4), with over 50% of oceans synthetically re-
Influence of different tectonic models
constructed prior to ~55 Ma (though, this excludes the now-subducted

4
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

Fig. 3. Our preferred paleobathymetry grids from this study, presented in 10 Myr increments. Paleobathymetry is based on the plate tectonic reconstruction of Müller
et al. (2019), the age-depth relationship of GDH1 (Stein and Stein, 1992), sediment model from Dutkiewicz et al. (2017), and the reconstruction of large igenous
provinces (LIPs) based on Cesile (2016).
5
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

Fig. 4. World uncertainty (black) and the percentage of presently preserved


oceanic crust (blue) through time, based on the plate tectonic reconstruction of
Müller et al. (2019). (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)

Here, we compare a number of plate kinematic models (Fig. 6):


specifically, Müller et al. (2008a), Müller et al. (2016), and Müller et al.
(2019), as they are the only publicly available models that also contain
a set of seafloor spreading isochrons. The differences between these
plate reconstruction models are generally limited to the incorporation
of different regional tectonic models, rather than independent plate
reconstructions and is largely because these tectonic models are all built
upon each other. The age-area uncertainty is determined in a similar
way to Müller et al. (2008a, 2008b), and is based on the mean misfit in
the age of both observed and reconstructed magnetic anomaly identi-
fications and lines approximating the location of the continent-ocean
transition (e.g., Williams et al., 2011) and the underlying age grid. To
determine the age uncertainty through time, we use the plate re-
construction model to reconstruct magnetic anomaly identifications to
their past position, and for times when only one spreading flank is
preserved, we synthetically reconstruct magnetic anomaly identifica-
tions onto the conjugate plate assuming symmetrical spreading. Our age
error does not explicitly include uncertainties associated with possible
long-term and sustained spreading asymmetry, however, shorter-lived
asymmetries associated with spreading complexities along sections of a
spreading system may be encompassed within our misfit between
magnetic anomalies and the age grid. Additionally, as our age un-
certainty primarily reflects the misfit between the age grid and mag-
netic anomaly identification and as prior to ~55 Ma the reconstruction
is dominated by synthetically modelled oceanic crust, we scale the
mean age error to be 20 Myr for now subducted ocean floor—an in-
crease from previous estimates (i.e., 10 Myr in Müller et al.,
2008b)—based on greater uncertainties in older reconstructions.
However, we note that it is not possible to fully capture the age un-
Fig. 5. Age-depth relationships through time and their influence on ocean ba-
certainty for oceanic crust where both spreading flanks are syntheti-
sins. Comparison of (a) the age-depth relationships from Parsons and Sclater
cally reconstructed. (1977), Stein and Stein (1992) (‘GDH1’), Crosby and McKenzie (2009) and
Since the Triassic, the mean ocean age has varied by ~30 Myr Richards et al. (2018) (complete plate model with pressure- and temperature-
(Fig. 6), with a large increase in the mean age of ocean crust since dependent conductivity in crustal layer); and its influence on (b) mean base-
~120 Ma (from ~40 Myr at 120 Ma to 64 Myr at present day), similar ment depth of oceanic crust (c) volume of oceanic crust, and (d) sea level, based
to the trends seen in Müller et al. (2008b) and Xu et al. (2006). Sea level on the global plate tectonic reconstructions from Müller et al. (2019). Sea level
increased significantly between 200 Ma and 120 Ma; this increase re- is calculated based on changes in the volume and area through time, relative to
flects the progressive development of mid-ocean ridges, especially from the present-day volume for that model. Uncertainties shown are based on the
the inception of Atlantic seafloor spreading around 160 Ma, and the age uncertainty associated with the paleo-age grids. Note that the axes on part
decrease in mid-ocean ridges from 120 Ma in the Pacific and Tethys. (a) vary from parts (b)–(d).
The difference in mean age, depth, and volume between Müller et al.
(2016) and Müller et al. (2019) arises from changes in the tectonic 2.1.1. Plate tectonic considerations
reconstruction itself, in particular, changes to the Caribbean and While the plate kinematic model from Müller et al. (2019) is our
eastern Tethys—both of these regions are now reconstructed as younger basis for investigating changes in ocean basin volume since the Jurassic,
oceanic crust in Müller et al. (2019) relative to Müller et al. (2016). changes in plate configuration and/or seafloor spreading rates may
greatly influence the age-area distribution of oceanic crust (and

6
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

~87 Ma, while the presently preserved oceanic crust is only ~30% at
87 Ma, and virtually all oceanic crust formed before ~180 Ma has now
been subducted (Fig. 4). In an attempt to explore some uncertainties in
the plate reconstruction itself, we investigate two elements of plate
reconstructions that may influence sea level: (1) the development and
destruction of back-arc basins and marginal seas; and (2) changes in
seafloor spreading asymmetry in the Pacific Ocean basin.

2.1.1.1. Back-arc basins. The evolutionary cycle of a typical back-arc


basin is intricately linked to the evolution of subduction and crustal
accretion, as they form episodically (either via extension or seafloor
spreading) behind an active subduction zone and are eventually
destroyed at continental margins as the final stage of back-arc
closure. Most back-arc basins active during the Cenozoic are
preserved in-situ, for example, the back-arc basins of the western
Pacific, and have yet to initiate the final stage of their life cycle (i.e.,
basin closure). This allows for a detailed reconstruction of their seafloor
spreading history using the traditional method of magnetic anomaly
and fracture zone interpretation. In contrast, evidence for back-arc
basins active during the Mesozoic is only preserved in the onshore
geological record (e.g., ophiolites) and provides an incomplete record
of back-arc opening. While the broad timing of back-arc activity and
back-arc closure can sometimes be deciphered, it is impossible to fully
uncover the complexity of its spreading regime. Back-arc basins are
strongly associated with subduction, where back-arc extension results
from rollback of the subducting plate (Taylor, 1995), resulting in young
and shallow oceanic crust ‘replacing’ the space previously
accommodated by the old and deep subducting seafloor. For example,
the Cenozoic back-arc basins of the Philippine Sea plate (West
Philippine, Parece Vela, Shikoku Basins and Mariana Trough) occupy
space that would otherwise be old Pacific Plate lithosphere.
Consequently, while back-arc basins cover a small portion of Earth's
surface, they may result in significant changes in ocean basin volume
and sea level due to this replacement of old oceanic crust with younger
crust.
To explore the possible influence of back-arc basins on paleo-
bathymetry and sea level, we create an alternative plate tectonic re-
construction of seafloor ages which excludes back-arc basins and mar-
ginal seas through all times (including present day; Fig. 7). This
approach provides us with a means to approximate the influence of
back-arc systems without detailed plate reconstructions of the back-arc
basins themselves. Specifically, for regions where back-arc basins and
marginal seas were incorporated into our plate kinematic model (here,
Müller et al., 2016), we remove the associated spreading systems and
seafloor isochrons and allow seafloor spreading isochrons associated
with the downgoing tectonic plate to accommodate these regions. This
alternative tectonic reconstruction is not meant to represent reality—as
evidence for past back-arc basins does exist—but rather represents an
Fig. 6. Comparison of selected plate tectonic reconstructions through time. The extreme scenario that allows us to explore the influence of back-arc
plate reconstructions include Müller et al. (2008b; green), Müller et al. (2016;
basins without having to explicitly model all possible back-arc basins
red), and Müller et al. (2019; blue). (a) mean age of oceanic crust, with un-
themselves, as the geologic record for all prior back-arc basins is likely
certainty indicated for the paleo-age grids; (b) mean depth of oceanic crust, (c)
volume of oceanic crust, and (d) changes in sea level through time. Basement incomplete.
depths are derived using GDH1 (Stein and Stein, 1992), and changes in volume Influence of back-arc basins/marginal seas on sea level
also incorporate changes in oceanic area (of the oceanic crustal portion only). The reconstruction with back-arc basins eliminated from the model
Sea level is calculated based on changes in the volume and area through time, largely deviates from the Müller et al. (2016) model during the Cen-
relative to the present-day volume for that model. The age uncertainty is based ozoic, where the absence of back-arc basins results in a ~6 Myr older
on the misfit between magnetic anomaly identifications and the location of the mean age of ocean crust, and up to 100 m deeper mean basement depth
continent-ocean transition, and does not capture the full range of uncertainties (Fig. 8). The elimination of back-arc basins results in a 50–120 m de-
inherent in the plate reconstructions. Uncertainty shown on sea-level curves are crease in sea level during the Cenozoic, with the biggest difference
based on the propagation of age uncertainty associated with the paleo-age
observed at 30 Ma. Based on this, we estimate that the presence of back-
grids. (For interpretation of the references to colour in this figure legend, the
arc basins had a maximum contribution of ~120 m increase on sea level
reader is referred to the web version of this article.)
through geologic time (Fig. 8).

therefore, the first-order control on sea level), especially during times 2.1.1.2. Asymmetries in crustal accretion. Asymmetries in seafloor
where they are less well-constrained such as the early Cenozoic and spreading can be determined when both spreading flanks are preserved;
prior. Based on Müller et al. (2019), world uncertainty is 50% at however, at times when the conjugate flank has been subducted,

7
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

Fig. 7. a. Paleo-age grids in 40 Myr increments showing a reconstruction with the supposed absence of back-arc basins. b. Difference between the reconstruction with
an absence of back-arc (from a), and the paleo-age grids from Müller et al. (2016) at corresponding times.

seafloor spreading is reconstructed based on symmetrical spreading. At the Pacific Ocean basin and do not extend this analysis to the Tethys, as
present day, oceanic crust in the Atlantic and Indian Ocean basins the Pacific Ocean is the dominant ocean basin during the Mesozoic, and
displays less than 4% cumulative spreading asymmetry (Müller et al., as past seafloor spreading systems in the Pacific Ocean can be
1998); however, asymmetries in crustal accretion are larger along the somewhat constrained by the preserved magnetic anomalies from one
fast-spreading East Pacific Rise since ~50 Ma (e.g., Müller et al., 1998; of the spreading flanks. We rely on a revised version of the plate
Rowan and Rowley, 2014). It is unclear if the asymmetric spreading reconstruction of Müller et al. (2016) to avoid complexities in plate
observed along the East Pacific Rise is a relatively recent development, deformation regions incorporated in Müller et al. (2019). We vary the
though it has been proposed that reconstructions of the East Pacific Rise asymmetry applied to the Pacific–Farallon (prior to 47.9 Ma only),
and the Pacific-Farallon ridge (precursor to the East Pacific Rise) should Pacific–Izanagi and Pacific–Phoenix spreading systems, so that the
include these long-term asymmetries (e.g., Rowan and Rowley, 2014). crustal accretion on the Pacific plate represents 25%, 37.5%, 44%
Considering the vast area of the Pacific Ocean basin, possible long-term (corresponding with observed Pacific–Nazca asymmetries; Rowan and
asymmetric crustal accretion may have a large influence on the global Rowley, 2014), 56%, 62.5%, and 75% of the total crust formed (Fig. 9).
mean age, depth, and volume of ocean basins. In an attempt to assess These scenarios encompass extreme endmembers compared to what
the uncertainties associated with the assumption of symmetrical might be considered likely values. In all cases, we do not modify the
spreading and its influence on the long-term evolution of ocean location of the mid-ocean ridge or the ‘half’ stage rotation applied to the
basins, we create suites of alternative plate kinematic reconstructions Pacific plate, as its history of crustal accretion can be determined using
in 5 Myr increments with varying asymmetry in crustal accretion in the preserved magnetic anomaly identifications, but rather, we calculate
Pacific Ocean basin. Here we use a workflow similar to that outlined in new finite rotations for each relative plate pair.
Williams et al. (2020) to test alternative scenarios for the velocities of Influence of crustal asymmetry on sea level
synthetic oceanic plates (and therefore the implied spreading As expected, variations in the asymmetry of crustal accretion in the
asymmetries) for the Pacific Ocean basin. We focus this analysis to Pacific Ocean basin results in similar long-term trends to the

8
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

symmetrical spreading embedded within Müller et al. (2016). All sce- alternative reconstructions but with spreading asymmetries ranging
narios give a present-day mean age of ~64 Myrs, which decreases to a from 37.5–67.5% varies crustal age by less than ± 10 Myr. When we
mean age between 25 and 52 Myrs by the Cretaceous (i.e., 120 Ma), and vary spreading asymmetries by 44% (corresponding with observed
increases to between 36 and 55 Myrs by 200 Ma. Notably, the series of Pacific–Nazca asymmetries), mean crustal age varies by 0.25–4.7 Myr

(caption on next page)

9
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

Fig. 8. Comparison of the global plate tectonic reconstruction to a reconstruction without back-arc basins, and to a reconstruction with modified spreading
asymmetry. (a–e): global plate tectonic reconstruction from Müller et al. (2016; black line) and age uncertainty (grey shading), compared to their reconstruction
modified to exclude any back-arc basins (green) for the: (a) mean age of ocean basins; (b) mean basement depth of ocean basins; (c) volume of oceanic crust; (d)
contribution of the lack of back-arc basins on mean depth of oceanic crust; and (e) contribution to sea level. Panel at the bottom of (e) indicates the number of back-
arc basins removed from the model through time, ranging from one (light grey) to nine (black) back-arc systems; (f–j) similar to (a)–(e), but for variations in the
percentage of Pacific crustal accretion (25% accretion: navy; 37.5% accretion: blue; 44% accretion: light blue; 56% accretion: light red; 67.5% accretion: red; 75%
accretion: dark red). Basement depths are derived using GDH-1 (Stein and Stein, 1992), and the contribution of components on mean depth of oceanic crust and
global sea level is determined from its difference to the Müller et al. (2016) reconstruction, where no difference between the models is 0 m. Note that the y-axis differs
between the back-arc and asymmetric crustal accretion comparison. (For interpretation of the references to colour in this figure legend, the reader is referred to the
web version of this article.)

to the symmetrically spreading scenario (Fig. 8f). The mean age, depth, crust from Müller et al. (2019), however, we do not find a major dif-
and volume from the reconstructions of Pacific plate asymmetry be- ference in the relationship compared to those derived from previous
tween 37.5% and 67.5% is within the age uncertainty envelope of the iterations of present-day oceanic age and sediment thickness grids (e.g.,
symmetrical spreading scenario (Fig. 8f). This implies that the long- Whittaker et al., 2013). We additionally correct for the isostatic com-
term changes in mean age, depth, and volume, are relatively robust, pensation of sediment thickness through time, based on Sykes (1996).
where the mean age of oceanic crust becomes younger during Pangea Influence of sediment thickness models on sea level
break-up and development of seafloor spreading in the Atlantic/Indian We find the inclusion of deep marine sedimentation affects the
oceans from ~160 Ma, and progressively older between ~120 Ma and global mean depth of ocean basins by up to 145 m (Fig. 10), with
present day, in both the symmetrical spreading and moderately asym- a ~100 m difference between the two sediment thickness reconstruc-
metric scenarios. Additionally, this suggests that a 20 Myr age-area tions during the late Jurassic. Consequently, there is an ~80 m differ-
uncertainty value (e.g., Fig. 6) may be appropriate for our plate re- ence in sea level between the two sediment thickness reconstructions in
construction. The Pacific Ocean basin, which dominates ocean basin the Mesozoic. The global sediment thickness model and the predictive
reconstructions since the Jurassic, are largely based on magnetic distance-based model (Dutkiewicz et al., 2017) are very similar be-
identifications from one spreading flank and assumed symmetrical tween present day and 80 Ma. However, prior to 80 Ma, these models
spreading. are significantly different—the global relationship overestimates mean
Based on asymmetries somewhat similar to presently observed va- sediment thickness (and subsequently, sea level) during times when the
lues (i.e., between 44 and 56%, with observable extremes between 37.5 paleo-Pacific/Panthalassa dominates reconstructions of ocean basin
and 67.5%), we find that asymmetry varies mean depth (relative to the volume (such as during the Mesozoic), while Dutkiewicz et al. (2017)
symmetrical spreading scenario) by −110 m (in the 67.5% accretion better incorporates the unique sedimentation histories of the individual
scenario) to 225 m (in the 37.5% scenario), with the biggest difference ocean basins, and ultimately results in a 70 m lower sea level compared
in mean depth observed around 120 Ma. This consequently influences to the global relationship in the Jurassic. Overall, the sediment thick-
sea level at 120 Ma by a ~76 m decrease (in the 67.5% scenario) to ness relationship from Dutkiewicz et al. (2017) suggests a much smaller
a ~150 m increase (in the 37.5% scenario) around 120 Ma, relative to contribution of sediments to sea level during the Mesozoic compared to
the symmetrical spreading scenario. Notably, when spreading asym- the global sediment thickness relationship, with respective contribu-
metries are varied by 44%, we find an overall increase in sea level re- tions of around 75 m and 150 m (Fig. 10).
lative to symmetrical spreading by up to 80 m (at 125 Ma). Extreme
values of asymmetry (e.g., 25% or 75% crustal accretion on the Pacific 2.3. Emplacement of large igneous provinces (LIPs)
plate, which is briefly observed along Pacific–Farallon ridge at
47.9–40.1 Ma; Wright et al., 2016) influences sea level by up to 300 m The eruption of LIPs have been found to occur at locations where
(e.g., in the 25% crustal accretion scenario in the early Cretaceous), mid-ocean ridges and mantle plumes interact (Whittaker et al., 2015),
relative to symmetrical spreading (Fig. 8). including recent work by Sager et al. (2019) suggesting Shatsky Rise
resulted from plume-ridge interactions. Syntheses of LIPs and their
2.2. Marine sedimentation volcanism history (e.g., Coffin and Eldholm, 1992) provide great insight
into their emplacement. The present-day mean elevation of individual
Marine sedimentation through time influences the global mean LIPs can be determined relative to the surrounding oceanic basement
depth and volume of the ocean but the history of sedimentation is and reconstructed through time by adding its mean elevation to paleo-
difficult to determine, especially prior to the Cenozoic where much of basement depth at its eruption age (Müller et al., 2008b). However, this
the record has been lost via subduction. A generic model predicting method does not account for the potential transient (Richards et al.,
sediment thickness through geologic time has been derived based on 1989) and/or permanent (Ito and Clift, 1998) surface uplift of the
the present-day relationship between oceanic crustal age, latitude, and seafloor associated with LIP emplacement. Additionally, the record of
sediment thickness (i.e., Müller et al., 2008b), However, observed se- LIP emplacement is incomplete, as plateaus may have been subducted
diment thickness is much thinner in the Pacific Ocean basin compared within the Pacific and Indian Ocean basins, and the incorporation of
to the Atlantic and Indian Ocean basins—this is attributed to the Pacific these subducted components may influence the mean depth and volume
basin's large area, resulting in less bioproductivity, and the subduction of ocean basins. On a similar note, those LIPs which form due to a mid-
of sediments along the circum-Pacific (Conrad, 2013). ocean ridge-plume interaction likely resulted in conjugate counterparts
Here we compare the influence of two sediment thickness models: on subducted oceanic plates and have been invoked in collisional
(1) a global sediment thickness model, equivalent to Müller et al. events, e.g., Hess and Shatsky conjugates and the Laramide orogeny
(2008b); and (2) a predictive model based on the mean distance to the (Liu et al., 2010). They might also cause flat slab subduction, as has
nearest passive margin, from Dutkiewicz et al. (2017). Additional been inferred for the Inca plateau (Gutscher et al., 1999), as well as for
comparison between these two sediment thickness models, a re- fragments of the Manihiki Plateau (Hochmuth and Gohl, 2017). We rely
gionalised relationship broadly based on Conrad (2013) and a cubic on the LIPs outlined in Whittaker et al. (2015), and reconstruct LIP
polynomial fit between present-day sediment thickness and oceanic age emplacement using two models: (1) the present-day elevation of LIPs
(Olson et al., 2016) can be found in Dutkiewicz et al. (2017). We derive from surrounding oceanic basement only (Müller et al., 2008b); and (2)
both sediment thickness relationships using the present-day sediment the elevation of LIPs based on surrounding oceanic basement, combined
thickness grid of Straume et al. (2019) and present-day age of oceanic with the additional elevation of the LIP and surrounding 5° radius of

10
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

Fig. 9. Paleo-age grids created by varying the spreading asymmetry in the Pacific Ocean basin, shown at 80 Ma, 120 Ma, and 160 Ma.

11
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

seafloor to account for the depth anomalies between observed and Fig. 10. Comparison of the influence of sediment thickness models through
predicted LIP depth, and for the transient uplift at the time of LIP time. The sediment thickness models include the global sediment thickness
emplacement (Cesile, 2016). We do not attempt to quantify the con- relationship (based on crustal age, latitude, and present-day sediment thick-
tribution of any possible conjugate counterparts of LIPs, however, we ness) used in Müller et al. (2008b; blue lines) and the predictive sediment
thickness model based on distance to the margins from Dutkiewicz et al., 2017
note this would increase the contribution of LIPs through time.
(red; lines). (a) mean depth of oceanic crust, when sediment thickness is (iso-
statically) added to basement depth (dark grey); (b) the volume of oceanic
crustal portion of ocean basins when sediment thickness is added to basement
depth (dark grey line); (c) the difference in the overall mean depth of ocean
basins with sediments, compared to the mean depth to basement only from
Müller et al. (2019); (d) variation in sea level when compared to paleo-base-
ment depths derived from Müller et al. (2019) only, and (e) changes in sea level
based on the combination of paleo-basement depth (dark grey: Müller et al.,
2019, and GDH1) and the two sediment thickness models (red: Dutkiewicz
et al., 2017; blue: global relationship used in Müller et al., 2008b), where the
sea-level curves incorporating sediments are calculated relative to the present-
day volume of paleo-basement depth only. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of this
article.)

Influence of LIPs
The different LIP emplacement models have a minor effect on the
mean depth and volume of ocean basins through time (Fig. 11), with
a ~20 m difference in mean depth between LIP models. LIPs may in-
fluence sea level up to 45 m, with a ~20 m difference in sea level de-
pending on the LIP model: the LIP reconstruction used within Müller
et al. (2008b) underestimates the contribution of LIP compared to the
reconstruction of Cesile (2016). It is important to note that since the
reconstruction history of LIPs is incomplete (i.e., due to subduction),
the overall contribution of LIPs to sea level is almost certainly much
larger. We also note that these reconstructions do not include other
seafloor volcanism, such as seamounts, that are included in Conrad
(2013), who suggest a ~100 m change in sea-level rise during the
Cretaceous period.

2.4. Passive margins and crustal extension

Crustal extension and rifting associated with the breakup of con-


tinents influences global ocean basin volumes in two broad ways. On
the one hand, continental rifting results in an increase in the total area
of continents requiring a decrease in the total area of oceanic crust to
compensate, resulting in a decrease in ocean basin volume. On the other
hand, the increase in continental area involves stretching and thinning
of the continental crust so that much of the newly-formed rifted mar-
gins lie significantly below sea level, resulting in an increase in ocean
basin volume. Previous estimates (Kirschner et al., 2010) suggest that
these two opposing components are similar in magnitude, such that the
overall influence on sea level is relatively small (a ~21 m increase in
sea level since the onset of Pangea rifting).
Plate tectonic reconstructions provide a simple means of estimating
the first of these two components, through the changing area of re-
constructed continental extents through time (e.g., Müller et al.,
2008b). The second component is more complicated, requiring a time-
dependent quantification of the volume of water held within evolving
rifts and passive margins, which in turn requires estimates of basement
subsidence and sedimentation (Kirschner et al., 2010). First-order es-
timates can be derived from knowledge of the present-day bathymetry
and basement depth, the time ranges of extension, and post-rift thermal
subsidence. The subsidence history within areas of extended con-
tinental lithosphere can be reconstructed through time using the uni-
form extension model of McKenzie (1978).
Here, we follow an approach similar in scope to that of Kirschner
et al. (2010) to estimate the combined influence of decreasing ocean
volume due to rifting, and increasing volume within passive margins by
reconstructing the paleodepth of present-day passive margins and rifted
continental regions based on their total tectonic subsidence and the

12
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

uniform extension model (McKenzie, 1978). In common with Kirschner Fig. 11. Comparison of two large igneous province (LIP) emplacement models
et al.'s (2010) study, we do not account for passive margins formed through time. The LIP emplacement models include Müller et al. (2008b; or-
during the last 200 Ma but not preserved intact at present day, such as ange lines), and Cesile (2016; blue lines). (a) mean depth of ocean basins when
those associated with rifting within the Tethys. We do however include LIP emplacement models (coloured lines) are included in the paleobathymetry
reconstruction, compared to paleobathymetry without LIPs (dark grey; based
on paleo-basement depth from Müller et al., 2019, and sediment accumulation
from Dutkiewicz et al., 2017), and (b) the corresponding volume of ocean ba-
sins; (c) the difference in the overall mean depth of ocean basins including LIPs
compared to the mean depth for paleobathymetry without LIPs (i.e., the dif-
ference between the coloured and grey line in part a); (d) the variation in sea
level based on the inclusion of LIP emplacement models, compared to paleo-
bathymetry without LIPs; and (e) changes in sea level based on the combination
of paleobathymetry without LIPs (dark grey; based on paleo-basement depth
using Müller et al., 2019, and GDH1, and sediment thickness using Dutkiewicz
et al., 2017), and the two LIP emplacement models (orange: Müller et al.,
2008b; blue: Cesile, 2016). The sea-level curves incorporating LIPs are calcu-
lated relative to the present-day volume of paleobathymetry without LIPS—the
sea-level curves are identical before 150 Ma, as no LIPs are incorporated prior
to this time. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)

all preserved passive margins within the Müller et al. (2019) re-
construction including some regions that Kirschner et al. (2010)
omitted.
To generate maps of paleobathymetry for passive margins preserved
at present day, we first generate a series of maps depicting the crustal
stretching factor and sediment thickness through time. For each passive
margin region, the age of rift onset is inferred from the initial timing of
relative motion between two rifting plates from Müller et al. (2019),
while the end of rifting and subsequent onset of thermal subsidence is
set to match the age of the oldest seafloor adjacent to the margin. Our
estimates of stretching factor and sediment thickness back in time are
all based on backstripping present-day bathymetry and basement
depth. The thickness of sediment across passive margins at present day
is available from the grids of Straume et al. (2019), which combined
with maps of present-day bathymetry (ETOPO1; Amante and Eakins,
2009) yields a map of present-day basement depth. Under the McKenzie
(1978) rifting model, the present-day seafloor depth at any point within
a rift is related to the timing and duration of rifting, the time elapsed
since rifting finished, the stretching factor, and the thickness of sedi-
ment deposited. In our case, for each grid point within the present-day
passive margins we have estimates for all these values except the
stretching factor. We thus determined the stretching factors which re-
produce present-day basement depth using the Constrained Optimiza-
tion BY Linear Approximation (COBYLA) algorithm (Powell, 1998); the
workflow to determine the stretching factor can be found as part of
‘pyBackTrack’ (Müller et al., 2018a). For simplicity, we only consider
extension in areas below sea level at present day (the nonmarine por-
tions of passive margins are likely to account for less than 4% of the
total budget according to Kirschner et al., 2010). We also make uniform
assumptions across all regions for parameters such as initial crustal
thickness (35 km), and the sediment density as a function of depth
(using values taken from Sawyer, 1985).
To reconstruct the seafloor depth through time for a given
stretching history, a further uncertainty comes from the rate at which
sediment is deposited as the basin subsides and new accomodation
space is created. Since the detailed history of sedimentation rates in
space and time is generally poorly known, we investigate two simple,
contrasting scenarios for the rate of sediment accumulation: (1) a
constant rate of sedimentation from the beginning of rifting to present
day; and (2) a sedimentation rate which keeps up with the subsidence
of the basin (Kirschner et al., 2010), until the sediment thickness is
equivalent to the present-day observed thickness (Fig. 12).
Following the steps outlined above, we computed paleobathymetry
at time increments of 1 Myr at each grid point (at a 0.1° resolution)
within present-day passive margins using the time span of rifting, the

13
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

Fig. 12. The depth evolution through time for different sedimentation scenarios for passive margins. Examples of basement depth (dashed) and paleobathymetry
(solid) using either a constant sedimentation rate (purple) or a sedimentation rate that keeps up with subsidence (orange). In this example, rifting is from 145 to
100 Ma, while the present-day bathymetry is 500 m, and present-day sediment thickness is 2000 m. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)

Fig. 13. Example of passive margin evolution through time for the South Atlantic. Prior to rifting and during the syn-rift phase, the margin extent overlaps (e.g., at
160 Ma and 120 Ma), and so the paleobathymetry is the average value for overlapping regions. Continental regions are shaded in grey, where the continent-ocean
boundary is indicated by a thick grey line. The reconstructed depth of passive margins are shown by the colours, and white denotes normal oceanic crust.
Reconstruction is based on Müller et al. (2019).

14
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

stretching factor, and the two alternative models of sediment accumu- Fig. 14. Changes related to passive margins. (a) The volume of ocean basins,
lation. These paleobathymetry grids were then reconstructed to the based on oceanic crustal components only (black line; paleo-basement depth
positions at the corresponding geological age using the reconstruction using GDH1; sedimentation model from Dutkiewicz et al., 2017; LIP elevations
parameters of Müller et al. (2019), which utilizes detailed analyses of from Cesile, 2016) and the volume of the oceanic components combined with
the reconstruction of present-day passive margins based on two sedimentation
crustal extension for several regions to constrain the plate
models: (1) constant sedimentation through time (purple), or (2) sedimentation
that keeps up with subsidence (yellow). The passive margin-based reconstruc-
tions alters volume through the inclusion of continental crust. (b) The change in
ocean area since 200 Ma. The change in oceanic crust (black line) reflects a
decrease or loss in the area of ocean crust, while the change in the area of
passive margin regions based on our study (blue line) reflects an increase in the
area of these regions. Also shown are the changes in passive margin regions
associated with Kirschner et al.'s (2010) late-onset rifting scenario without
seawater correction (green line). (c) The absolute change in volume since
200 Ma, based on the ‘container’ volume paleobathymetry change in passive
margins using the constant sedimentation scenario (purple) and sedimentation
that keeps up with subsidence scenario (yellow), and the change in the volume
of the lost deep oceanic crust based on our passive margin regions (that is, the
area from part b multiplied by 4.269 km, which is the average depth of deep
ocean from Kirschner et al., 2010; blue line), and the comparable change in
oceanic volume from Kirschner et al.'s (2010) uncorrected late-onset rifting
parameters (green line). (d) The net change in the volume of passive margins
caused by continental stretching since 200 Ma, which is the difference in the
volume of deep oceanic crust and the ‘container volume’ of passive margin
regions (that is, blue line − purple/yellow line in part c) for the different
passive margin sedimentation scenarios, and the comparable change in volume
caused by continental stretching from Kirschner et al. (2010) (corrected late-
onset rifting parameters; red line). (e) The difference in sea level from 200 Ma
based on changes in volume associated with extension of continental crust,
based on the constant sedimentation (purple line) and sedimentation that keeps
up with subsidence (yellow line) scenarios for passive margins, and the change
in sea level since 200 Ma based on Kirschner et al.'s (2010) corrected late-onset
rifting parameters (red line). (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)

configurations during rifting. In these reconstructed configurations,


during the syn-rift phase, the paleobathymetry maps from present-day
conjugate margin extents overlap one another (e.g., Fig. 13), so we
average the values in the overlapping regions. In this way, the re-
constructed maps incorporate both the change in area of continental
crust consistent with the global kinematic reconstruction model, and
the increase in volume of water contained within rifts and rifted mar-
gins.
Influence of passive margins on sea level
To assess the influence of continental extension associated with
passive margin formation, we explore its influence on the change in
ocean area and changes in volume (Fig. 14). Extension of continental
crust results in a corresponding decrease (or loss) in the deep ocean
area (i.e., the area associated with oceanic lithosphere only) by about
9 × 106 km2 in our reconstruction (Fig. 14b), which is slightly larger
than Kirschner et al. (2010) as we consider a more complete coverage of
present-day passive margins. Consequently, we find a slightly larger
decrease than Kirschner et al. (2010) in the change in deep ocean vo-
lume that has been lost due to continental extension (Fig. 14c; blue
line). In Fig. 14c we also show the ‘container’ volume change for the
continental crust associated with passive margins since 200 Ma, which
are of a similar magnitude to our decrease in the volume of deep
oceanic crust. Notably, passive margin regions in our plate re-
construction (Müller et al., 2019) only account for about half of the loss
of deep ocean area (Fig. 14b). Changes in the continental geometries
used within the plate reconstruction account for the other half of the
loss in deep ocean area—these include changes to terranes (i.e., the
appearance of terrane polygons in the Arctic at 120 Ma, causing a large
decrease in deep ocean area in Fig. 14b) and island arc formation (i.e.,
the presence of arcs in the Caribbean and SW Pacific at 85 Ma, causing
another large decrease in ocean area).
To illustrate the effect of continental extension on sea level, we use
an approach similar to Kirschner et al. (2010) and determine the net

15
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

Fig. 15. Changes in sea level through time. (a) The contribution of components to sea-level change, including basement depth (grey), sediments (peach), large
igneous provinces (LIPs; dark blue), the lack of back-arc basins (green), the inclusion of the area and volume associated with continental extension and other factors
(including but not limited to island arcs and microcontinents; purple), and the asymmetry scenarios with 44% crustal accretion onto the Pacific plate (light blue) and
56% crustal accretion onto the Pacific plate (red). Here our basement depth contribution is calculated based on changes in ocean volume, and incorporates both
changes in mean oceanic crustal depth and mean ocean area. (b) Changes in sea level based on: (1) the traditional model (red line), which is the combination of
basement depth (using Stein and Stein, 1992; grey), sediment thickness (using the relationship from Dutkiewicz et al., 2017; peach); and LIP emplacement (using the
model from Cesile, 2016; dark blue). This model is comparable to the reconstruction in Müller et al. (2008b). Here we have separated these components (basement
depth, sediments, and LIPs) to show how they each contribute to the overall sea level (i.e., red line) through time; (2) our refined model (black line), which is the
combination of the traditional model (basement depth, sediments, LIPs) with variations associated from the change in area and volume associated with continental
extension and other factors (shown in purple), plus the contribution from past (unmodelled) LIP emplacement (dotted dark blue) and an estimate for lost back-arc
basins (green). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

change in volume. As the decrease in the volume of normal (deep) change in volume between 200 Ma and present day is extremely small.
oceanic crust from continental extension is of a similar magnitude to There is potentially more significant variation during the past 200 Ma,
the container volume of passive margins (Fig. 14c), we find that the net though whether this influence would increase or decrease sea level

16
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

depends on the sedimentation style used. Overall, we find a ~10 m back-arc basins and LIPs that may not be accounted for within the
increase in sea level at present day related to the extension of con- traditional model as discussed previously (Fig. 15b). We do not in-
tinental crust, which is slightly smaller than the ~21 m increase found corporate long-term spreading asymmetry into this sea-level curve,
in Kirschner et al. (2010), but the change of sea level through time is though we show the possible contribution that asymmetry may have in
strongly dependent on the assumed sedimentation style: a constant Fig. 15a.
sedimentation rate results in an initial decrease in sea level by 30 m We estimate the contribution of lost back-arc basins or those that
between 200 and 120 Ma, followed by an increase in sea level by have yet to be incorporated in our plate model to be a ~113 m increase
~40 m to present day, while if sedimentation keeps up with subsidence in sea level prior to 150 Ma. This is based on the median influence of
then sea level broadly increases between 200 Ma and 20 Ma by 20 m the absence of back-arc basins during the past 50 Myr—a period where
(similar to Kirschner et al., 2010), followed by a slight decrease to the back-arc basin history can be relatively well resolved. Our estimate
present day (Fig. 14e). The most rapid change in sea level due to the for the influence varies through time, to account for the times in our
time-dependent influence of rifts is in the Early Cretaceous, when the preferred plate model (Müller et al., 2019) where back-arc basins have
total length of (ultimately successful) rifts is largest (Brune et al., 2017). already been implemented. Specifically, we do not include an addi-
An important note here is that in order to isolate the specific influence tional estimate for back-arc basins between 0 and 50 Ma; we apply our
of continental extension and make our results comparable to previous maximum estimate (~113 m increase in sea level) for times prior to
estimates, sea-level change due to continental extension is calculated 150 Ma; and we estimate the influence between 50 and 150 Ma using a
relative to the entire area of present-day oceans (including continental second order spline interpolation. While the influence of LIPs can be
shelves) and this area is fixed through time—this differs greatly from determined based on those preserved at present day, it is limited to the
Sections 2.1–2.3, which are calculated based on the area of oceanic past ~150 Myr because any LIPs that existed prior to this time have
crust only and incorporate changes in the area of continental crust either been subducted or obducted on to continental margins. For times
through time. This difference does not change the basic outcome of this before 150 Myr, we estimate the past influence of LIPs to be a ~43 m
section of our analysis, that the net effect of volume changes due to increase in sea level, based on the median influence of LIPs between 0
continental extension on global ocean basin volume is relatively small and 120 Myr.
compared to other factors. The incorporation of estimates for back-arc basin formation results
Our results using the sedimentation that keeps up with subsidence in a ~113 m difference in sea level in the Early Jurassic (200 Ma)
are consistent with the trends and magnitude of sea-level change for (Fig. 15). By including maximum estimates for the volume change as-
most of the past 200 Ma from Kirschner et al. (2010). The differences sociated with LIP emplacement (based on Cesile, 2016), we find an
we find are likely due to the combination of a number of factors: we increase in sea level between ~43 m (at 200 Ma) and 20 m (at 120 Ma)
consider all present-day passive margins, rather than the subset con- (Fig. 15). Sustained spreading asymmetry in the Pacific Ocean basin
sidered by Kirschner et al. (2010); we utilize more recent sediment may either result in an increase in sea level (i.e., if 37.5–44% accretion
thickness grids (Straume et al., 2019), which contain a number of sig- is on the Pacific plate), or decrease in sea level (i.e., if 56–67.5% ac-
nificant refinements such as better representing thick sediments along cretion is on the Pacific plate). The inclusion of continental crust, such
the Australian and Antarctic conjugate margins; and we use plate tec- as regions of continental extension associated with rifting along with
tonic reconstructions in which extension estimates account for addition island arc and microcontinent formation, changes both the present-day
of igneous material to the crust at volcanic margins, such as in the ocean basin area and volume compared to the other components which
North Atlantic, a factor acknowledged by Kirschner et al. (2010) as a focus solely on the oceanic crust portion of the ocean basins and only
potential source of error in their analysis. influence the volume; this change in area from the inclusion of con-
tinental components results in up to a ~100 m difference in sea level at
3. Long-term changes in sea level times during the past 200 Ma, compared to the reconstruction that are
based on the area of oceanic crust only (Fig. 15).
Based on the contribution of factors described in the previous sec-
tion (i.e., tectonic factors, marine sedimentation, LIPs, etc.), we de- 3.1. Comparison with sea level derived from alternative plate
termine changes in sea level since the Jurassic. Our sea-level calcula- reconstructions
tions incorporate both changes in the volume and area of ocean basins.
The changes in the area covered by oceanic crust are related to a Plate tectonic reconstructions have previously been used to explore
number of factors including the extension and rifting of continental changes in long-term sea level throughout the Cenozoic (Xu et al.,
crust, deformation of continents (e.g., India-Eurasia collision, Andean 2006), since the Cretaceous (Müller et al., 2008b), and throughout the
orogeny), and the formation and destruction of intra-oceanic arcs. We entire Phanerozoic (Vérard et al., 2015). These studies vary in the
additionally account for the isostatic correction of seawater by scaling global plate tectonic model used, in particular for the seafloor age
our sea-level changes by 0.69 (i.e., by (⍴m − ⍴w)/ ⍴m, where ⍴m is the distribution within regions of now-subducted oceanic lithosphere,
density of the mantle, 3330 kg/m3, and ⍴w is the density of seawater, yet all show similar trends of > 100 m decrease in sea level during the
1030 kg/m3). Our sea-level curve does not include changes associated early Cenozoic (Fig. 16).
with glaciation, such as the development of ice sheets since ~34 Ma. As There are some major differences between the sea-level curves de-
sea level would rise by 54 m (Miller et al., 2005) to 70 m (Alley et al., rived from plate tectonic reconstructions, such as between our tradi-
2005) if all present-day ice sheets melted, we consequently shift our tional model, and those derived in Müller et al. (2008b), Xu et al.
curve by 70 m, the latter of these estimates. We derive two sea-level (2006), and Vérard et al. (2015) (Fig. 16). All sea-level curves that have
curves: (1) a traditional model for sea level, which is based on the same been derived using an underlying plate tectonic model show a long-
factors considered by Müller et al. (2008b) (i.e., age-area distribution, term decrease in sea level during the Cenozoic. Müller et al. (2008b)
sediments, and LIPs) this provides a direct comparison between the new indicate relatively little sea-level change since ~35 Ma, while Xu et al.
and previously published results that depend solely on changes in the (2006) (based on Hall, 2002; herein referred to as Xu-Hall) show
plate reconstructions, derived sedimentation model, and LIP analysis; a ~50 m decrease in sea level during this time—considering all these
and (2) a refined sea-level curve, which additionally incorporates the sea-level curves were created in an ice-free world, this discrepancy is
change in area and volume due to the inclusion of continental crust due to the differences in the plate tectonic reconstruction. The magni-
(such as the volume and area associated with continental extension tude of sea-level change also differs between the curves: since 65 Ma,
along present-day passive margins, as well as changes in island arc and Müller et al. (2008b) suggests a 65 m decline in sea level, Xu et al.
microcontinent configuration), estimates for the presence of additional (2006) (based on Gordon and Jurdy, 1986; herein referred to as Xu-GJ)

17
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

Fig. 16. Comparison of some key sea-level curves with our two sea-level curves: (1) traditional model (red line), which is based on changes associated with basement
depth (determined from the age-area distribution), sediments, and LIPs and is comparable to the method used in Müller et al., 2008b), and is based on changes
associated with basement depth (determined from the age-area distribution), sediments, and LIPs; and (2) our refined model (black line), which is the combination of
the traditional model changes associated with the area and volume change associated with passive margins and other factors, along with estimates for the presence of
back-arc basins and past LIP influence. Our changes in sea level incorporate both changes in the area and depth of ocean basins. Note that no attempt to convert for
various time-scales have been made for the published sea-level curves. The sea-level curve from Müller et al. (2018b) is model M7 which includes dynamic
topography and glacio-eustasy. Kominz et al. (2008) includes corrections for dynamic topography. + Vérard et al. (2015) relies on underlying geodynamic maps,
however, specific details on the plate reconstruction are unavailable.* Karlsen et al. (2019) rely on the global plate reconstruction of Müller et al. (2016), with
additional components including deep water cycling and dynamic topography, however their sea-level curve does not incorporate changes in ocean basin area. (For
interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

suggest a 250 m decrease in sea level, while both Xu-Hall and our sea- (2008b), our sea-level curve (traditional model) indicates an extremely
level reconstruction show a 100–110 m decrease in sea level during this large (380 m) increase in sea level from 180 to 120 Ma—this increase
time. The sea-level curve derived from Vérard et al. (2015) shows a occurs during the main phase of Pangea break-up, such as the devel-
large (~120 m) decrease in sea level from 200 to 175 Ma, while most opment of seafloor spreading in the Atlantic and Indian ocean basins.
other sea-level curve indicates a minor change in sea level during this
time period. This difference may be attributed to the different under-
lying plate tectonic models and their spatio-temporal distribution of 3.2. Comparison with other models and implications for global sea level
seafloor ages; however, this is difficult to ascertain due to the ambiguity
in the model as presented in Vérard et al. (2015), as their plate tectonic Our preferred sea-level curve (our refined model) results in broadly
reconstruction is not shown. similar trends to published curves (Fig. 16); however, the timing of our
Our new sea-level curve (our traditional model) shows significant sea-level highstand occurs much earlier, at 120 Ma, compared to the
changes compared to Müller et al. (2008b), due to the refined plate highstand at ~100 Ma found in Haq et al. (1987). In particular, the
kinematic model (Müller et al., 2019) used within this study. These sea- incorporation of estimates (e.g., presence of back-arc basins, LIP vol-
level curves do not differ over the past 20 Myr. From 80 to 20 Ma, there canism prior to 120 Ma) reconciles the ~250 m discrepancy between
is up to a ~60 m difference between the sea-level curve from Müller the published sea-level curves and the traditionally derived curve based
et al. (2008b) and our two new sea-level curves, related largely to the on Müller et al. (2019) in the Jurassic (Fig. 15). While our sea-level
significant refinements in the plate tectonic history of SE Asia, and curve using the traditional model shows much larger changes in the
minor refinements in the plate configuration and seafloor spreading Mesozoic than other tectonic-based sea-level curves, our sea-level curve
history of the Pacific Ocean basin (Müller et al., 2016, 2019). Our new with estimates incorporated into it results in very similar sea-level
sea-level curve using the traditional method (i.e., similar to Müller trends to Müller et al., 2018b, which also incorporates dynamic topo-
et al., 2008b) shows a remarkably similar trend in sea level with a graphy, and Karlsen et al. (2019), which additionally incorporates dy-
plateau between 120 and 80 Ma (Fig. 16), while our refined sea-level namic topography and deep-water cycling. Our refined sea-level curve
curve (i.e., that includes additional estimates) shows a clear peak in sea also results in changes in sea level of a similar magnitude to published
level around 120 Ma. While 120 Ma coincides with major LIP volcanism stratigraphic-based curves such as Haq et al. (1987).
(e.g., Ontong Java and southern Kerguelen), it is important to re- While investigations of past sea level based on ocean basin re-
member that this sea-level peak may be exaggerated due to the esti- constructions consistently show ~200 m variations in sea level over the
mated addition of various components. Significant changes in plate past 100 Myr, long-term changes in sea level have also been modelled
configuration of the Pacific Ocean basin, and variations to the Tethys using alternative approaches which do not depend on a plate tectonic
including several back-arc basins along the southern Eurasian margin reconstruction, such as those based on well data from the New Jersey
also occurred during the Cretaceous. Prior to the record in Müller et al. margin (Miller et al., 2005), onshore proxies and/or flooding and pa-
leoshorelines (Rowley, 2017) and the 87Sr/86Sr record (van der Meer

18
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

et al., 2017). While both of these approaches do not require a plate through geological time, thereby providing a reference point from
tectonic model and instead rely on an alternative proxy, they each which to derive sea level, whether from well data (Miller et al., 2005),
imply very different sea-level histories; for example, Rowley (2017) or plate tectonic reconstructions (Müller et al., 2008b; Xu et al., 2006).
infers much smaller changes in long-term sea level, with only However, continents are not stable because of the vertical motions
25 ± 22 m since the Late Jurassic, while van der Meer et al. (2017) caused by mantle heterogeneity, known as dynamic topography (Hager
find 165 m variation in sea level since the Late Jurassic, and long-term et al., 1985; Braun, 2010). Dynamic topography can influence sea level
cyclicity which is more consistent with sea-level curves derived from significantly, but the magnitude of dynamic topography varies con-
plate reconstructions or sequence stratigraphy. One major assumption siderably between the models—since the Cretaceous, models suggest
that may influence van der Meer et al. (2017) sea-level curve is the that dynamic topography has increased sea level ~70 m (Conrad,
assumption that 87Sr/86Sr records are a suitable proxy for crustal pro- 2013), 90 m (Conrad and Husson, 2009), or 100–200 m (Spasojevic and
duction (van der Meer et al., 2017). The implied mean depth and age of Gurnis, 2012). The magnitude of mean oceanic dynamic topography
oceanic crust can be determined from both methods; Rowley (2017) itself since 140 Ma has recently been explored, and while it varies
finds very small variations in mean ocean depth (~15 ± 11 m), re- considerably between each model, dynamic topography influences sea
sulting in a mean age of oceanic crust of ~62.1 ± 2.4 Myr, while the level between 50 and 100 m (Müller et al., 2018b). Time-dependent
variations in mean age determined by van der Meer et al. (2017) are dynamic topography models have also been used to reconcile the dis-
broadly in line with Müller et al. (2019). The minor variation in the crepancy between stratigraphic-derived sea-level estimates and sea
mean age of global ocean basins determined by Rowley (2017) dis- level derived from ocean basin reconstructions (Müller et al., 2008b).
agrees with the preserved seafloor spreading record, where the mean For example, dynamic topography resulted in 105–180 m of subsidence
age of oceanic lithosphere has varied by ~20 Myr over the past 80 Ma, along the New Jersey margin, thereby influencing the sea-level curve
even when long-term spreading asymmetry (e.g., those presented by presented in Miller et al. (2005). Ideally, global time-dependent dy-
Rowan and Rowley, 2014) in the Pacific Ocean basin has been in- namic topography estimates should be incorporated into sea-level re-
corporated. constructions derived from ocean basin volume, however, as models of
dynamic topography vary greatly in their magnitude and have large
3.3. Model limitations uncertainties in the oceanic realm, we have not considered their in-
fluence on sea level here. The exchange of water between the oceans
A number of limitations related to the reconstruction of synthetic and the mantle (e.g., Cloetingh and Haq, 2015) is an additional factor
components arise in our model. While preserved oceanic crust de- that may influence sea level through time, though this is difficult to
monstrates that spreading asymmetry varies between segments of a constrain. Recently, Karlsen et al. (2019) investigated deep water re-
mid-ocean ridge, we assess the changes in ocean basin volume by cycling between the oceans and mantle, based on the parameterisation
uniformly modifying the percentage of crustal accretion for the three of deep subduction water flux and the tectonic model from Müller et al.
major spreading systems associated with the Pacific plate (i.e., (2016). They found that deep water cycling may influence sea level
Pacific–Izanagi, Pacific–Farallon, and Pacific–Phoenix). As it is ex- around 130 m (Karlsen et al., 2019), and may play a large role in sea-
tremely unlikely that all major spreading systems in the Pacific Ocean level changes around times of rapid subduction of oceanic lithosphere,
basin would undergo a similar variation in asymmetry for long time such as around 150 Ma.
periods, we are likely overestimating the actual effect of spreading
asymmetries. Additionally, a comparison of spreading asymmetries 4. Conclusions
along the East Pacific Rise reveals that crustal accretion along the
southern East Pacific Rise is unique in the Pacific Ocean basin, and the Global reconstructions of sea-level change based on a recent plate
higher asymmetries in certain segments can be attributed to mantle tectonic reconstruction (Müller et al., 2019) significantly overestimate
upwelling associated with the underlying Pacific superswell, causing the change in sea level during the Mesozoic compared to published
successive westward ridge jumps (Wright et al., 2016). Consequently, it curves (e.g., Haq et al., 1987). Considerable uncertainties are inherent
would be incorrect to extrapolate the ~48 Myr history of spreading in these plate tectonic reconstructions, such as the plate tectonic con-
asymmetry in the Pacific Ocean basin to the rest of the Pacific Ocean figuration and distribution of mid-ocean ridges and back-arc basins. We
basin through time, contrary to what Rowan and Rowley (2014) pro- find that less-constrained elements of plate tectonic reconstructions,
pose. We incorporate fluctuations in spreading asymmetry by changing such as back-arc basins, have influenced sea level by 120 m. Assuming
the spatial distribution of oceanic ages on the conjugate plates (Izanagi, long-lived spreading asymmetry in the Pacific Ocean basin does not
Phoenix, and Farallon), rather than the location of the mid-ocean ridge significantly alter the long-term trend in mean age, depth, and ocean
itself, as the location of the Pacific mid-ocean ridges are constrained basin volume since the Jurassic, though, may contribute up to 150 m to
based on magnetic anomaly identifications for the Pacific–Phoenix and sea level. The incorporation of LIP volcanism and sedimentation to the
Pacific–Farallon ridges, though the position of the Pacific–Izanagi ridge volume of ocean basins changes sea level by ~45 m and between 75
prior to M0 (120.6 Ma) is unknown. Regardless of our variations in and 165 m, and models used to predict sedimentation or LIP empla-
spreading asymmetry for the Pacific Ocean basin, we find a similar cement can vary by ~20 m. Sea-level variations are related to passive
trend in the mean age, mean depth, and volume of oceanic crust. margin formation are time-dependent and rely greatly on the assumed
The history of LIPs is sparse before 120 Ma; however, excess vol- sedimentation style, with either a ~40 m increase in sea level between
canism and LIP emplacement have been proposed along the edge of the 120 Ma and present day for constant sedimentation, or a more gradual
underlying large low shear velocity provinces (LLSVP) in the Pacific ~20 m increase in sea level between 200 and ~20 Ma when sedi-
basin, such as the Pigafetta basin (Madrigal et al., 2016). Geologic mentation keeps up with subsidence. When changes in the area and
evidence for further accretion events of LIPs have been compiled volume of submerged continental areas (which may be associated with
(Safonova and Santosh, 2014); these studies serve to show that the continental extension, but also other factors such as changes in con-
current record is limited both spatially and temporally, and that de- tinental geometries due to island arcs formation) are included, we find
termining robust volumetric estimate for such accreted terranes is likely a ~100 m higher sea level, compared to sea level derived using the
to remain problematic. changes in oceanic crust volume and area only. The incorporation of the
continental area and volume that is associated with passive margins and
3.4. Potential influence of factors not considered other factors, combined with maximum estimates for the influence of
back-arc basins, and LIP emplacement, for times where they are poorly
Sea-level curves assume that the continents have been “stable” constrained or not captured by current plate tectonic reconstructions

19
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

results in a ~250 m difference in sea level in the Jurassic, and helps heterogeneity, dynamic topography and the geoid. Nature 313, 541–545. https://doi.
reconcile the plate tectonic record with other sea-level curves. org/10.1038/313541a0.
Hall, R., 2002. Cenozoic geological and plate tectonic evolution of SE Asia and the SW
Pacific: computer-based reconstructions, model and animations. J. Asian Earth Sci.
Declaration of Competing Interest 20 (4), 353–431. https://doi.org/10.1016/S1367-9120(01)00069-4.
Hallam, A., 1984. Pre-Quaternary Sea-level changes. Annu. Rev. Earth Planet. Sci. 12 (1),
205–243. https://doi.org/10.1146/annurev.ea.12.050184.001225.
The authors declare that they have no known competing financial Haq, B.U., 2014. Cretaceous eustasy revisited. Glob. Planet. Chang. 113, 44–58. https://
interests or personal relationships that could have appeared to influ- doi.org/10.1016/j.gloplacha.2013.12.007.
Haq, B.U., 2017. Jurassic Sea-Level Variations: a Reappraisal. GSA Today 28 (1). https://
ence the work reported in this paper. doi.org/10.1130/GSATG359A.1.
Haq, B.U., Al-Qahtani, A.M., 2005. Phanerozoic cycles of sea-level change on the Arabian
Acknowledgements Platform. GeoArabia 10 (2), 127–160.
Haq, B.U., Hardenbol, J., Vail, P.R., 1987. Chronology of fluctuating sea levels since the
Triassic. Science 235, 1156–1167. https://doi.org/10.1126/science.235.4793.1156.
We thank Michelle Kominz and an anonymous reviewer for their Harrison, C., 1990. Long-term eustasy and epeirogeny in continents. Sea-Level Change
constructive comments which improved the manuscript. NMW was 141–158.
supported by an Australian Postgraduate Award. We would like to ac- Hasterok, D., 2013. A heat flow based cooling model for tectonic plates. Earth Planet. Sci.
Lett. 361, 34–43. https://doi.org/10.1016/j.epsl.2012.10.036.
knowledge funding from the Australian Research Council through Hays, J.D., Pitman, W.C., 1973. Lithospheric plate motion, sea level changes and climatic
grants FT130101564 (MS), DP200100966 (MS and SEW), and ecological consequences. Nature 246 (5427), 18–22.
IH130200012 (RDM), DP180102280 (JMW and SEW) and Hochmuth, K., Gohl, K., 2017. Collision of Manihiki Plateau fragments to accretional
margins of northern Andes and Antarctic Peninsula. Tectonics 36, 229–240. https://
SR140300001 (JMW). Supplementary material can be downloaded doi.org/10.1002/2016TC004333.
from: ftp://ftp.earthbyte.org/Data_Collections/Wright_etal_2020_ESR/. Ito, G., Clift, P., 1998. Subsidence and growth of Pacific Cretaceous plateaus. Earth
Planet. Sci. Lett. 161, 85–100. https://doi.org/10.1016/S0012-821X(98)00139-3.
Karlsen, K.S., Conrad, C.P., Magni, V., 2019. Deep water cycling and sea level change
References since the breakup of Pangea. Geochem. Geophys. Geosyst. 20 (6), 2919–2935.
https://doi.org/10.1029/2019GC008232.
Cesile, S.D., 2016. An investigation into oceanic large igneous province plume swell Kirschner, J.P., Kominz, M.A., Mwakanyamale, K.E., 2010. Quantifying extension of
(Honours thesis). The University of Tasmania. passive margins: Implications for sea level change. Tectonics 29. https://doi.org/10.
Abreu, V.S., Anderson, J.B., 1998. Glacial eustasy during the Cenozoic: sequence strati- 1029/2009TC002557.
graphic implications. AAPG Bull. 82 (7), 1385–1400. Kominz, M.A., 1984. Oceanic ridge volumes and sea-level change-an error analysis. Am.
Alley, R.B., Clark, P.U., Huybrechts, P., Joughin, I., 2005. Ice-sheet and sea-level changes. Assoc. Petrol. Geol. Mem. 36, 108.
Science 310 (5747), 456–460. https://doi.org/10.1126/science.1114613. Kominz, M.A., 2001. Sea Level Variations over Geologic Time. Academic Press, San
Amante, C., Eakins, B.W., 2009. ETOPO1: 1 Arc-Minute Global Relief Model: Procedures, Diego, pp. 2605–2613. https://doi.org/10.1006/rwos.2001.0255.
Data sources and Analysis. In: NOAA Technical Memorandum NESDIS NGDC-24. Kominz, M.A., 2013. Sea Level Variations over Geologic Time. Reference Module in Earth
National Geophysical Data Center, NOAA. https://doi.org/10.7289/V5C8276M. Systems and Environmental Sciences, Elsevier 2013. https://doi.org/10.1016/B978-
Bond, G.C., 1979. Evidence of some uplifts of large magnitude in continental platforms. 0-12-409548-9.04349-9.
Tectonophysics 61, 285–305. Kominz, M.A., Browning, J.V., Miller, K.G., Sugarman, P.J., Mizintseva, S., Scotese, C.R.,
Braun, J., 2010. The many surface expressions of mantle dynamics. Nat. Geosci. 3 (12), 2008. Late Cretaceous to Miocene Sea-level estimates from the New Jersey and
825. https://doi.org/10.1038/ngeo1020. Delaware coastal plain coreholes: an error analysis. Basin Res. 20 (2), 211–226.
Brune, S., Williams, S.E., Müller, R.D., 2017. Potential links between continental rifting, https://doi.org/10.1111/j.1365-2117.2008.00354.x.
CO2 degassing and climate change through time. Nat. Geosci. 10 (12), 941–946. Lambeck, K., Esat, T.M., Potter, E.K., 2002. Links between climate and sea levels for the
https://doi.org/10.1038/s41561-017-0003-6. past three million years. Nature 419 (6903), 199–206. https://doi.org/10.1038/
Cloetingh, S., Haq, B.U., 2015. Inherited landscapes and sea level change. Science 347 nature01089.
(6220), 1258375. https://doi.org/10.1126/science.1258375. Liu, L., Gurnis, M., Seton, M., Saleeby, J., Müller, R.D., Jackson, J.M., 2010. The role of
Coffin, M.F., Eldholm, O., 1992. Volcanism and continental break-up: a global compila- oceanic plateau subduction in the Laramide orogeny. Nat. Geosci. 3, 353. https://doi.
tion of large igneous provinces. Geol. Soc. Lond., Spec. Publ. 68 (1), 17–30. https:// org/10.1038/ngeo829.
doi.org/10.1144/GSL.SP.1992.068.01.02. Madrigal, P., Gazel, E., Flores, K.E., Bizimis, M., Jicha, B., 2016. Record of massive up-
Cogné, J.-P., Humler, E., Courtillot, V., 2006. Mean age of oceanic lithosphere drives wellings from the Pacific large low shear velocity province. Nat. Commun. 7. https://
eustatic sea-level change since Pangea breakup. Earth Planet. Sci. Lett. 245, 115–122. doi.org/10.1038/ncomms13309.
https://doi.org/10.1016/j.epsl.2006.03.020. Matthews, K.J., Müller, R.D., Wessel, P., Whittaker, J.M., 2011. The tectonic fabric of
Coltice, N., Seton, M., Rolf, T., Müller, R.D., Tackley, P.J., 2013. Convergence of tectonic ocean basins. J. Geophys. Res. Solid Earth 116 (B12). https://doi.org/10.1029/
reconstructions and mantle convection models for significant fluctuations in seafloor 2011JB008413.
spreading. Earth and Planetary Science Letters. Dec 1 (383), 92–100. https://doi.org/ McKenzie, D., 1978. Some remarks on the development of sedimentary basins. Earth
10.1016/j.epsl.2013.09.032. Planet. Sci. Lett. 40, 25–32.
Conrad, C.P., 2013. The solid Earth’s influence on sea level. Geol. Soc. Am. Bull. 125, van der Meer, D.G., van Saparoea, A.V.D.B., van Hinsbergen, D.J.J., van de Weg, R.M.B.,
1027–1052. https://doi.org/10.1130/B30764.1. Godderis, Y., Le Hir, G., Donnadieu, Y., 2017. Reconstructing first-order changes in
Conrad, C.P., Husson, L., 2009. Influence of dynamic topography on sea level and its rate sea level during the Phanerozoic and Neoproterozoic using strontium isotopes.
of change. Lithosphere 1, 110–120. https://doi.org/10.1130/L32.1. Gondwana Res. 44, 22–34. https://doi.org/10.1016/j.gr.2016.11.002.
Crosby, A.G., McKenzie, D., 2009. An analysis of young ocean depth, gravity and global Miller, K.G., Kominz, M.A., Browning, J.V., Wright, J.D., Mountain, G.S., Katz, M.E.,
residual topography. Geophys. J. Int. 178, 1198–1219. https://doi.org/10.1111/j. Sugarman, P.J., Cramer, B.S., Christie-Blick, N., Pekar, S.F., 2005. The Phanerozoic
1365-246X.2009.04224.x. record of global sea-level change. Science 310 (5752), 1293–1298. https://doi.org/
Crosby, A., McKenzie, D., Sclater, J., 2006. The relationship between depth, age and 10.1126/science.1116412.
gravity in the oceans. Geophys. J. Int. 166, 553–573. https://doi.org/10.1111/j. Mitrovica, J.X., Tamisiea, M.E., Davis, J.L., Milne, G.A., 2001. Recent mass balance of
1365-246X.2006.03015.x. polar ice sheets inferred from patterns of global sea-level change. Nature 409,
Demicco, R.V., 2004. Modeling seafloor-spreading rates through time. Geology 32 (6), 1026–1029. https://doi.org/10.1038/35059054.
485–488. https://doi.org/10.1130/G20409.1. Müller, R.D., Cannon, J., Williams, S., Dutkiewicz, A., 2018a. PyBacktrack 1.0: a tool for
Dutkiewicz, A., Müller, R.D., Wang, X., O’Callaghan, S., Cannon, J., Wright, N.M., 2017. reconstructing paleobathymetry on oceanic and continental crust. Geochem.
Predicting sediment thickness on vanished ocean crust since 200 Ma. Geochemistry, Geophys. Geosyst. 19 (6), 1898–1909. https://doi.org/10.1029/2017GC007313.
Geophysics, Geosystems 18. https://doi.org/10.1002/2017GC007258. Müller, R.D., Hassan, R., Gurnis, M., Flament, N., Williams, S.E., 2018b. Dynamic topo-
East, M., Müller, R.D., Williams, S., Zahirovic, S., Heine, C., 2020. Subduction history graphy of passive continental margins and their hinterlands since the Cretaceous.
reveals Cretaceous slab superflux as a possible cause for the mid-Cretaceous plume Gondwana Res. 53, 225–251. https://doi.org/10.1016/j.gr.2017.04.028.
pulse and superswell events. Gondwana Res. https://doi.org/10.1016/j.gr.2019.09. Müller, R.D., Roest, W.R., Royer, J.-Y., 1998. Asymmetric sea-floor spreading caused by
001. ridge-plume interactions. Nature 396, 455. https://doi.org/10.1038/24850.
Gordon, R.G., Jurdy, D.M., 1986. Cenozoic global plate motions. Journal of Geophysical Müller, R.D., Sdrolias, M., Gaina, C., Roest, W.R., 2008a. Age, spreading rates, and
Research: Solid Earth 91 (B12), 12389–12406. https://doi.org/10.1029/ spreading asymmetry of the world’s ocean crust. Geochemistry Geophysics
JB091iB12p12389. Geosystems 9, Q04006. https://doi.org/10.1029/2007GC001743.
Guillaume, B., Pochat, S., Monteaux, J., Husson, L., Choblet, G., 2016. Can eustatic charts Müller, R.D., Sdrolias, M., Gaina, C., Steinberger, B., Heine, C., 2008b. Long-term sea-
go beyond first order? Insights from the Permian-Triassic. Lithosphere 8 (5), level fluctuations driven by ocean basin dynamics. Science 319, 1357–1362. https://
505–518. https://doi.org/10.1130/L523.1. doi.org/10.1126/science.1151540.
Gutscher, M.-A., Olivet, J.-L., Aslanian, D., Eissen, J.-P., Maury, R., 1999. The “lost Inca Müller, R.D., Seton, M., Zahirovic, S., Williams, S.E., Matthews, K.J., Wright, N.M.,
Plateau”: cause of flat subduction beneath Peru? Earth Planet. Sci. Lett. 171, Shephard, G.E., Maloney, K.T., Barnett-Moore, N., Hosseinpour, M., 2016. Ocean
335–341. https://doi.org/10.1016/S0012-821X(99)00153-3. basin evolution and global-scale plate reorganization events since Pangea breakup.
Hager, B., Clayton, C., Richards, M., Comer, R., Dziewonski, 1985. Lower mantle Annu. Rev. Earth Planet. Sci. 44, 107–138. https://doi.org/10.1146/annurev-earth-

20
N.M. Wright, et al. Earth-Science Reviews 208 (2020) 103293

060115-012211. 10.1016/0012-821X(76)90108-4.
Müller, R.D., Zahirovic, S., Williams, S.E., Cannon, J., Seton, M., Bower, D.J., Tetley, Stein, C.A., Stein, S., 1992. A model for the global variation in oceanic depth and heat
M.G., Heine, C., Le Breton, E., Liu, S., Russell, S.H., Yang, T., Leonard, J., Gurnis, M., flow with lithospheric age. Nature 359, 123–129.
2019. A global plate model including lithospheric deformation along major rifts and Straume, E.O., Gaina, C., Medvedev, S., Hochmuth, K., Gohl, K., Whittaker, J.M., Abdul
orogens since the Triassic. Tectonics 38. https://doi.org/10.1029/2018TC005462. Fattah, R., Doornenbal, J.C., Hopper, J.R., 2019. GlobSed: Updated total sediment
Olson, P., Reynolds, E., Hinnov, L., Goswami, A., 2016. Variation of ocean sediment thickness in the world’s oceans. Geochem. Geophys. Geosyst. 20 (4), 1756–1772.
thickness with crustal age. Geochem. Geophys. Geosyst. 17, 1349–1369. https://doi. https://doi.org/10.1029/2018GC008115.
org/10.1002/2015GC006143. Sykes, T.J., 1996. A correction for sediment load upon the ocean floor: Uniform versus
Parsons, B., Sclater, J.G., 1977. An analysis of the variation of ocean floor bathymetry and varying sediment density estimations—Implications for isostatic correction. Mar.
heat flow with age. J. Geophys. Res. 82, 803–827. https://doi.org/10.1029/ Geol. 133, 35–49. https://doi.org/10.1016/0025-3227(96)00016-3.
JB082i005p00803. Taylor, B. (Ed.), 1995. Backarc Basins: Tectonics and Magmatism. Springer Science &
Pitman, W.C., 1978. Relationship between eustacy and stratigraphic sequences of passive Business Media.
margins. Geol. Soc. Am. Bull. 89, 1389–1403. Tennant, J.P., Mannion, P.D., Upchurch, P., 2016. Sea level regulated tetrapod diversity
Powell, M.J.D., 1998. Direct search algorithms for optimization calculations. Acta dynamics through the Jurassic/Cretaceous interval. Nat. Commun. 7, 12737. https://
Numerica 7, 287–336. https://doi.org/10.1017/S0962492900002841. doi.org/10.1038/ncomms12737.
Richards, M.A., Duncan, R.A., Courtillot, V.E., 1989. Flood basalts and hot-spot tracks: Torsvik, T.H., Steinberger, B., Gurnis, M., Gaina, C., 2010. Plate tectonics and net litho-
plume heads and tails. Science 246 (4926), 103–107. https://doi.org/10.1126/ sphere rotation over the past 150 my. Earth Planet. Sci. Lett. 106–112. https://doi.
science.246.4926.103. org/10.1016/j.epsl.2009.12.055.
Richards, F.D., Hoggard, M.J., Cowton, L.R., White, N.J., 2018. Reassessing the thermal Turcotte, D., Schubert, G., 2002. Geodynamics —, Second edition. Cambridge University
structure of oceanic lithosphere with revised global inventories of basement depths Press, Cambridge.
and heat flow measurements. Journal of Geophysical Research: Solid Earth 123 (10), Vail, P.R., Mitchum Jr., R.M., Thompson III, S., 1977. Seismic stratigraphy and global
9136–9161. https://doi.org/10.1029/2018JB015998. changes of sea level: part 4. Global cycles of relative changes of sea level: section 2.
Rowan, C.J., Rowley, D.B., 2014. Spreading behaviour of the Pacific-Farallon ridge Application of seismic reflection configuration to stratigraphic interpretation. pp.
system since 83 Ma. Geophys. J. Int. 197, 1273–1283. https://doi.org/10.1093/gji/ 83–97.
ggu056. Vérard, C., Hochard, C., Baumgartner, P.O., Stampfli, G.M., Liu, M., 2015. 3D palaeo-
Rowley, D.B., 2002. Rate of plate creation and destruction: 180 Ma to present. Geol. Soc. geographic reconstructions of the Phanerozoic versus sea-level and Sr-ratio varia-
Am. Bull. 114 (8), 927–933. tions. J. Palaeogeogr. 4, 64–84. https://doi.org/10.3724/SP.J.1261.2015.00068.
Rowley, D.B., 2017. Earth’s constant mean Elevation: Implication for Long-Term Sea Watts, A.B., Steckler, M.S., 1979. Subsidence and eustasy at the continental margin of
Level Controlled by Oceanic Lithosphere Dynamics in a Pitman World. The Journal of eastern North America. Deep Drilling Results in the Atlantic Ocean: Continental
Geology 125, 141–153. https://doi.org/10.1086/690197. Margins and Paleoenvironment 3, 218–234. https://doi.org/10.1029/ME003p0218.
Rowley, D.B., 2019. Oceanic axial depth and age-depth distribution of oceanic litho- Watts, A.B., Thorne, J., 1984. Tectonics, global changes in sea level and their relationship
sphere: Comparison of magnetic anomaly picks versus age-grid models. Lithosphere to stratigraphical sequences at the US Atlantic continental margin. Mar. Pet. Geol. 1
11 (1), 21–43. https://doi.org/10.1130/L1027.1. (4), 319–339.
Safonova, I.Y., Santosh, M., 2014. Accretionary complexes in the Asia-Pacific region: Wessel, P., Matthews, K.J., Müller, R.D., Mazzoni, A., Whittaker, J.M., Myhill, R.,
tracing archives of ocean plate stratigraphy and tracking mantle plumes. Gondwana Chandler, M.T., 2015. Semiautomtic fracture zone tracking. Geochem. Geophys.
Res. 25 (1), 126–158. https://doi.org/10.1016/j.gr.2012.10.008. Geosyst. https://doi.org/10.1002/2015GC005853.
Sager, W.W., Huang, Y., Tominaga, M., Greene, J.A., Nakanishi, M., Zhang, J., 2019. Whittaker, J.M., Goncharov, A., Williams, S.E., Müller, R.D., Leitchenkov, G., 2013.
Oceanic plateau formation by seafloor spreading implied by Tamu Massif magnetic Global sediment thickness data set updated for the Australian-Antarctic Southern
anomalies. Nat. Geosci. 12 (8), 661–666. Ocean. Geochem. Geophys. Geosyst. 14, 3297–3305. https://doi.org/10.1002/ggge.
Sawyer, D.S., 1985. Total tectonics subsidence: A parameter for distinguishing crust type 20181.
at the U.S. Atlantic continental margin. J. Geophys. Res. 90 (B9), 7751–7769. https:// Whittaker, J., Afonso, J., Masterton, S., Müller, R., Wessel, P., Williams, S., Seton, M.,
doi.org/10.1029/JB090iB09p07751. 2015. Long-term interaction between mid-ocean ridges and mantle plumes. Nat.
Seton, M., Whittaker, J.M., Wessel, P., Müller, R.D., DeMets, C., Merkouriev, S., Cande, S., Geosci. 8, 479–483. https://doi.org/10.1038/ngeo2437.
Gaina, C., Eagles, G., Granot, R., Stock, J., Wright, N., Williams, S.E., 2014. Williams, S.E., Whittaker, J.M., Müller, R.D., 2011. Full-fit, palinspastic reconstruction of
Community infrastructure and repository for marine magnetic identifications. the conjugate Australian-Antarctic margins. Tectonics 30 (6). https://doi.org/10.
Geochem. Geophys. Geosyst. 15 (4), 1629–1641. https://doi.org/10.1002/ 1029/2011TC002912.
2013GC005176. Williams, S.E., Wright, N.M., Cannon, J., Flament, N., Müller, R.D., 2020. Reconstructing
Snedden, J.W., Liu, C., 2010. A compilation of Phanerozoic Sea-level change, coastal seafloor age distributions in lost ocean basins. Geoscience Frontiers. https://doi.org/
onlaps and recommended sequence designations. Search and Discovery Article 10.1016/j.gsf.2020.06.004.
40594, 3. Wright, N.M., Seton, M., Williams, S.E., Müller, R.D., 2016. The Late Cretaceous to recent
Spasojevic, S., Gurnis, M., 2012. Sea level and vertical motion of continents from dynamic tectonic history of the Pacific Ocean basin. Earth Sci. Rev. 154, 138–173. https://doi.
earth models since the Late Cretaceous. AAPG Bull. 96 (11), 2037–2064. https://doi. org/10.1016/j.earscirev.2015.11.015.
org/10.1306/03261211121. Xu, X., Lithgow-Bertelloni, C., Conrad, C.P., 2006. Global reconstructions of Cenozoic
Spooner, E.T.C., 1976. The Strontium Isotopic Composition of Seawater, and Seawater seafloor ages: Implications for bathymetry and sea level. Earth Planet. Sci. Lett. 243,
Oceanic Crust Interaction. Earth Planet. Sci. Lett. 31 (1), 167–174. https://doi.org/ 552–564. https://doi.org/10.1016/j.epsl.2006.01.010.

21

You might also like