Continent-scale Hiatus Maps for the Atlantic Realm and Australia since the Upper Jurassic and links to mantle flow induced dynamic topography

Abstract

Interregional geological maps hold important information for geodynamic models. Here, we use such maps to visualize major conformable and unconformable contacts at interregional scales and at the level of geologic series from the Upper Jurassic onward across North and South America, Europe, Africa and Australia. We extract hiatus information from these paleogeological maps, which we plot in a paleogeographical reference frame to link the maps to the plate and plume modes of mantle convection. We assume that interregional patterns of hiatus surfaces are proxy records of continent-scale mantle-induced vertical motion of the lithosphere. We find significant differences in the distribution of hiatus across and between continents at the timescale of geologic series, that is ten to a few tens of millions of years (Myrs). This is smaller than the mantle transit time, which, as the timescale of convection, is about 100–200 Myrs. Our results imply that different timescales for convection and topography in convective support must be an integral component of time-dependent geodynamic Earth models, consistent with the presence of a weaker upper mantle relative to the lower mantle. Additional geological constraints together with interregional geological maps at the resolution of stages (1–2 Myrs), are needed to assist in future geodynamic interpretations of interregional geologic hiatus.

1. Introduction

An early success in geodynamics was the quantitative description of mantle convection by a boundary-layer model of high Rayleigh number and low Reynolds number flow [1]. The model came into its own when mantle convection was explored explicitly in terms of the plate and plume mode [e.g. [2–4]]. The former is associated with the cold upper thermal boundary layer, which is the lithosphere, and the latter with the hot lower thermal boundary layer, which sources plumes.

The plate mode has since then been mapped by kinematic models of lithosphere motion for the Cenozoic [5] and Mesozoic (e.g. [6]). Its temporal evolution has been linked to the generation of large-scale mantle heterogeneity through the history of subduction [7,8] and assimilated into global mantle convection simulations [9–11] to construct mantle circulation models, which from here on we will call MCM. Recently, the plume mode has been imaged by seismic tomography as localized upwellings that rise from the core-mantle boundary (CMB) to the base of the lithosphere [12–14], and the boundary-layer nature of mantle convection is now widely recognized.

Geodynamicists also understood early on that mantle convection deflects the Earth’s surface away from its isostatically compensated state [15]. Termed ‘dynamic topography’ by Hager et al. [16] the deflections are receiving renewed attention (e.g. [17]), particularly as an agent in passive margin environments [18], where the proximity to a base level allows one to gauge topographic changes better than at other places.

The boundary-layer interpretation of mantle flow makes it convenient to interpret the sedimentary expression of dynamic topography explicitly in terms of the plate and plume modes. For the plate mode, the approach was pioneered using the sedimentary record from the Cretaceous Interior Seaway and the cratonic interior of North America (e.g. [19,20]) because surface depressions induced by mantle downwellings in these regions left accommodation space to preserve a sedimentary archive. Other regions, such as the Cretaceous Eromanga Sea in Australia [21,22] and a regional unconformity of Cretaceous-Eocene age in southeast Asia [23] also record plate-mode-related vertical motion. Recently, MCMs have modelled the evolution of plate-mode-related dynamic topography since the Cretaceous [24].

It is more difficult to map the stratigraphic expression of the plume mode because the positive surface deflections create erosional/non-depositional environments, which leave time gaps in the sedimentary record. Field observations of the surface expression of the plume mode document changes in drainage patterns (e.g. [25]) and a dome-shaped uplift of 1–2 km (e.g. [26–28]) over a radius of 1000–2000 km. The resulting discontinuity surfaces in the sedimentary record are known as unconformities (e.g. [29]), although their wavelengths are so large and their amplitudes so little that at large distances an unconformity may locally be recorded as a disconformity. They preserve time missing (hiatus) from the geological record [30]. To this end, an approach of hiatus-area mapping was introduced [30,31] to highlight the long-wavelength nature of sedimentation records as explored by Sloss [32,33]. It visualizes interregional-scale unconformities because, at continental scales, what is normally perceived as a lack of data (material eroded or not deposited) becomes part of the dynamic topography signal. The method has been applied to map the temporal and spatial patterns of conformable and unconformable geological contacts across Europe [34] and Africa [35].

Continent-scale geological maps, such as the 1 : 5 Million International Geological Map of Europe and Adjacent Areas (IGME 5000) [36], are crucial databases to reveal hiatus area of geodynamic origin, that is falcogeny in the sense of Şengör [37]. They provide internally consistent compilations of geological observations, including chronostratigraphic age, lithology and geolocalization of the strata, at the scale of thousands of kilometres. This links them naturally to continent-scale elevation changes induced by mantle flow. Here, we explore interregional-scale geological maps. We identify temporal and spatial patterns of geological hiatus contacts across North and South America, Europe, Africa and Australia, under the assumption that interregional-scale conformable and unconformable contacts are proxy records of paleotopography and vertical motion. We organize our paper as follows: first, we explain our hiatus mapping method. Then we present results starting from hiatus maps for the Upper Jurassic. We find significant differences in the spatial extent of hiatus area across and between continents at the timescale of geologic series, ten to a few tens of millions of years (Myrs), which is considerably smaller than the mantle transit time [38]. We note that this negates the concepts of Stille [39,40] and Sloss [33], who argued for global synchronicity cycles. Finally, we discuss our results, place them into a geodynamic context, explore their implications for dynamic Earth models, and draw conclusions.

2. Data compilation, preparation and uncertainties

We mapped conformable and unconformable contacts at the resolution of geological series because this is the most frequently adopted temporal resolution among interregional geologic maps [31]. We also opted to map hiatus from the Upper Jurassic onward, to remain within a timescale comparable with the mantle transit time, which is about 100–200 Myrs [38]. To this end, we took the digital vector maps of Europe, Australia and North America, which describe the chronostratigraphic units within specific temporal and spatial resolutions. For South America, we compiled individual country-scale information, since only this was available at the temporal resolution of series.

Diverse open-access databases [36,41–51] provide digital information from geological maps as vector files. Some maps include information from the continental shelf and other seafloor features. We did not use this information because it also includes magnetic isochron data, which are not related to the sedimentation paleoenvironment. However, oceanic pointwise information in the form of localized stratigraphic columns from the Ocean Drilling Program (ODP) can record oceanic hiatus events. For this reason, we imported offshore data from ODP [52] Legs 100–190 and Deep Sea Drilling Project (DSDP) [53] Legs 1–95. Additionally, we used [35] for the Cenozoic series of Africa augmented by further information for the Upper and Lower Cretaceous. Table 1 summarizes our compilation of geological information. The geological maps published at continent and country scale vary both in spatial and temporal resolution. Some maps are resolved at the series level. Others provide finer or coarser geological time intervals, such as combinations of series, stages or systems, as defined in the chronostratigraphic chart [54,55]. For instance, the map may state Paleogene for the units shown. Thus time resolution falls within three categories: series, series/stages/systems mix and systems. The maps moreover use distinct naming conventions for age descriptions, including different abbreviations, languages and aggregations of time units. To handle the diversity, we adopted a standardization procedure and harmonized the time resolution among the maps. We saturated all subseries information to the series level and brought the geological unit conventions to a standard reference. This translates languages, abbreviations and combinations or ranges of geological units to the numerical value of geological time. For instance, a polygon defined as Oligocene-Miocene or Chattian-Langhian time has the same time range after the standardization and spans two series (approx. 30 Myrs). For polygons with systems resolution we assigned the hiatus information to the base of the polygon’s age range. For South America our procedure brought the country-specific maps to a unified continent-scale format. An exception had to be made for Argentina, where the temporal resolution was available only at the systems level.

Table 1.

Geological maps used in this work with their respective spatial and temporal resolution. Compilations performed at the country level for South America (see text).

region		temporal resolution	spatial resolution	reference
Australia		series and stages	1 : 1 million	Geoscience Australia [41]
Europe		series	1 : 5 million	BGR [36]
North America		series and stages	1 : 5 million	USGS [42]
South America	Argentina	systems	1 : 2.5 million	SEGEMAR [43]
	Bolivia	series	1 : 1 million	SERGEOTECMIN [44]
	Brazil	series	1 : 250 000	CPRM [45]
	Chile	stages	1 : 1 million	SERNAGEOMIN [46]
	Colombia	stages	1 : 1 million	SGC [47]
	Ecuador	series	1 : 100 000	MAGAP [48]
	Peru	stages	1 : 100 000	INGEMMNET [49]
	Uruguay	series	1 : 500 000	MIEM [50]
	Venezuela	series	1 : 500 000	USGS [51]
Africa		series and systems	1 : 5 million*	[35]
Ocean Drilling Projects		series	—	[52,53]

*Africa hiatus information taken from [35] with hiatus information added for Upper and Lower Cretaceous. Offshore data imported from ODP [52] Legs 100–190 and DSDP[53] Legs 1–95 as pointwise signal.

Following Carena et al. [35], we define conformity if a target series sits atop the one immediately preceding it in the chronostratigraphic chart, regardless of whether either series has missing stages. We define unconformity as the complementary state to conformity. This holds for any place where one or more series immediately preceding the target series are missing. The definitions apply regardless of the physical contact type between both rock units. Figures 1 and 2 provide a schematic illustration of hiatus and the extraction process for un/conformable contacts [31]. To delimit hiatus for a given series, we also include in the maps any occurrence of the immediately preceding series and categorize the signal as conformable.

Figure 1.

(a) Schematic map (top) and perspective view (bottom) of geological units in conformable contact (blue lines). (b) Same scheme showing an unconformable contact (red line), where the middle unit is missing, representing a gap (hiatus) in the geologic record. (Online version in colour.)

Figure 2.

(a) Schematic geological map for five consecutive chronostratigraphic units (T1 to T5) with T1 youngest and T5 oldest. (b) Schematic showing the extraction of un/conformable contacts for a target unit. Conformable lines for the target unit are the contours of the preceding unit. Unconformable contacts contour the contact of the unit with units older than the immediately preceding one. (Online version in colour.)

Since the temporal resolution is restricted to the series level, the un/conformity represents a time span that varies for different series. For instance, unconformity at the base of the Miocene datum is at least 11 Myrs, because this is the duration of the Oligocene series. Unconformity at the base of the Paleocene datum lasts a minimum of 34 Myrs, which is the duration of the Upper Cretaceous series. We note, however, that the hiatus duration could be longer for either case. In the former, rocks of Lower and Middle Miocene and/or Upper and Middle Eocene could be missing. In the latter, rocks of Lower and Middle Paleocene and/or Lower Cretaceous could be absent. The uncertainty of a hiatus transforms into a spatial uncertainty when plate motions are taken into account. If we take the current global root mean square plate velocity of 5 cm yr⁻¹ [56] as a representative value, temporal uncertainty for a hiatus at the series level (10–30 Myrs) translates into a minimum spatial uncertainty of 500–1500 km. Moreover, by saturating temporal resolution to the series level, we underestimate the total amount of hiatus because unconformities and hiatus at the resolution of stages may be masked at the stratigraphic resolution of series. Figure 3 illustrates these uncertainties.

Figure 3.

Schematic illustration of the temporal and spatial uncertainty of hiatus mapping [30,31,34,35]. Panels (a) and (b) show conformable and unconformable contacts, respectively. Panel (c) displays how temporal uncertainty translates into spatial uncertainty for the paleogeographical reconstruction representing hiatus. (Online version in colour.)

3. Results

(a). Geological Hiatus Maps

Figure 4 shows hiatus mapped with our method for North and South America, Europe, Africa and Australia for eight geologic series. This yields a set of eight Geological Hiatus Maps (GHMs), beginning with the Lower Cretaceous (i.e. hiatus here meaning that the Upper Jurassic is missing). We use pyGPlates [58] to reconstruct each hiatus to its past tectonic setting with a global Mesozoic-Cenozoic plate motion model [6] tied to a reference frame of Indo-Atlantic hotspots [57,59] and present the extracted signal in a plate tectonic configuration corresponding to the base of each series. Red and blue colours depict un/conformable contacts, respectively. Blank regions indicate the absence of the considered series and its immediately preceding unit.

Figure 4.

Geological Hiatus Maps (GHMs) at chronostratigraphic division of series [54] from the Base of Pleistocene datum to the Base of Lower Cretaceous datum (a–h) reconstructed paleogeographically with a global Mesozoic-Cenozoic plate motion model [6] tied to a reference frame of Indo-Atlantic hotspots [57] and shown in a plate tectonic configuration corresponding to the base of each series. Red/blue points represent un/conformable contacts, respectively. Blank regions indicate the absence of considered series and its immediately preceding unit. See text for further information. (Online version in colour.)

In the following, we describe the results for each GHM. Base of Pleistocene datum, figure 4a, presents North and South America, Greenland and Australia with predominantly unconformable contacts. Conformable contacts exist in the High Plains of North America, parts of South America, and the Australian Nullarbor Plain. Europe is dominated by conformable contacts. Africa shows a mix of un/conformable contacts, with conformable contacts located in the northwest and in the Kalahari and Congo Basins. Unconformable contacts extend through the East African Highlands and the Sahara desert. Base of Pliocene datum, figure 4b, exposes conformable contacts in North and South America, around the Gulf of Mexico, the Basin and Range, the Rocky Mountains front, the Brazilian Highlands and the western Amazon Basin. Australia shows sparse conformable contacts throughout the continent and isolated unconformable contacts in the north and in the southeast. Conformable contacts cover eastern Europe and the Iberian Peninsula, while unconformable contacts prevail in western/central Europe and in tectonically active regions in the Mediterranean. Africa exhibits unconformable contacts in the Congo Basin and the Canary-Atlas region, while conformable contacts occur in the Kalahari Basin, the Afar region and the northern edges of the continent. Base of Miocene datum, figure 4c, is dominated by unconformable contacts across the continents. Unconformable contacts abound in the western part of North America, Brazil and much of Europe, whereas isolated hiatuses exist in Australia and Africa. Conformable contacts are exposed in Greenland, western and easternmost Europe and the Kalahari Basin. Base of Oligocene datum, figure 4d, exposes conformable contacts in many regions, with a striking absence of signal across South America. Unconformable contacts are mapped in the western parts of North America, the Afar region and in Europe. Base of Eocene datum, figure 4e, features a mix of signals. Unconformable contacts exist in eastern Africa, Europe and the western parts of North America adjacent to conformable contacts in the plains of Canada. South America lacks information except for conformable contacts in the eastern Amazon. Africa reveals conformable contacts in its northern parts and the Kalahari Basin, but signal is absent in the central and southern parts of the continent. Scattered unconformable contacts are mapped across Australia. Base of Paleocene datum, figure 4f , reveals abundant conformable contacts across North and South America, Europe, Africa and Australia. Unconformable contacts are located in the northwestern part of North America and Greenland. South America exposes unconformable contacts along the Andes and the east coast of Brazil. Africa shows clusters of unconformable contacts in the Kalahari Basin, the northern Djoue Basin and the Afar region. Unconformable contacts are mapped in southern Australia near Tasmania. Base of Upper Cretaceous datum, figure 4g, is characterized by unconformable contacts, which prevail across Africa, Europe and the western parts of North America. Conformable contacts are mapped in Canada, Mexico, the northern and eastern parts of Africa, the Parana region of South America and Australia. Finally, Base of Lower Cretaceous datum, figure 4h, exhibits a mix of un/conformable contacts. Most notable are unconformable contacts in Alaska as well as a lack of signal throughout much of southern/central Africa and South America. We point out that the absence of Mesozoic/Cenozoic strata across much of Scandinavia and the cratonic part of North America precludes hiatus mapping for the Mesozoic/Cenozoic series in these regions.

(b). Base Hiatus Surfaces

The GHMs allow us to perform a spherical harmonics expansion of the hiatus signal to create Base Hiatus Surface (BHS). We adopted pyshtools [60] with fully normalized spherical harmonic coefficients [61], using a global equidistant grid of 720/1440 points in latitude/longitude for a resolution of ≈30 km between grid nodes. Numerical values of 1/−1 were assigned to un/conformable signal, respectively. Each grid node was then initialized with the nearest hiatus value that falls within a radius of 1/2 of the grid node distance. Otherwise, the grid node value was set to zero.

We performed the expansion up to spherical harmonic degree 100. However, our assumption of a geodynamic origin for interregional-scale hiatus implies the choice of a spectral window that one should consider in the BHS representations. Longstanding arguments based on dynamic models of the Geoid suggest a dominant contribution to dynamic topography of spherical harmonic degree 2 [62]. The dominance of the longest wavelength components for convectively maintained topography was challenged recently by an observational database of greater than 2000 spot measurements of residual bathymetry in the oceanic realm [63]. The latter suggests that contributions up to and including degree 30 are required to represent topography in convective support. Figure 5 illustrates the difference and reports BHS for the Base of Miocene datum for four cut off degrees (2, 10, 15 and 30) and a tapered Gaussian smoothing to the spectral coefficients. The taper width of 40 degrees allows the contribution of spectral components beyond the cut off. For the long-wavelength cut off at degree 2 there remains a 30% contribution of the original signal at degree 27, while the degree 30 cut off maintains 30% of the original signal up to degree 55. We report BHS starting with the Lower Cretaceous and assuming an intermediate cut off at degree 15 in figure 6.

Figure 5.

Base of Miocene Hiatus Surface (BHS) obtained by expanding the Miocene Geological Hiatus Map (GHM) (figure 4c) in fully normalized [61] spherical harmonics (SH) and convolving with a Gaussian taper at four different cut off values for degree 2 (a), 10 (b), 15 (c) and 30 (d), respectively. Red/blue areas represent un/conformable surfaces. Black dotted lines contour the SH signal at the ±0.1 amplitude. Hiatus data from the input GHMs are shown as blue/red dots. Centre plot: four Gaussian tapers applied in the SH expansion, see text. (Online version in colour.)

Figure 6.

Base Hiatus Surface (BHS) obtained by expanding the Geological Hiatus Maps (GHMs) (figure 4) in fully normalized [61] spherical harmonics (SH) and convolving with a Gaussian taper starting at degree 15 (compare with figure 5). BHS shown at chronostratigraphic division of series [54] from the Base of Pleistocene to the Base of Lower Cretaceous (a–h) reconstructed paleogeographically with a global Mesozoic-Cenozoic plate motion model [6] tied to a reference frame of Indo-Atlantic hotspots [57] and placed into a plate tectonic configuration corresponding to the base of each series. Blue/red colours represent no-/hiatus surfaces, indicative of low/high topography in the preceding series, respectively. Black dotted lines contour the SH signal at the ±0.1 amplitude. Hiatus data from the input GHMs are shown as blue/red dots. Black circles at Base of Miocene (c), Base of Eocene (e) and Base of Upper Cretaceous (g) maps correspond to the location of flood basalts associated with Afar, Iceland and Tristan hotspots [64]. Blank regions indicate the absence of series and its immediately preceding unit, suggesting long hiatus duration. See text for further information. (Online version in colour.)

The BHS provides information on the temporal evolution in the ratio of the area of conformal surface relative to the total area of conformal and unconformal surface. The latter can be plotted both aggregated over all continents and separate for each. The aggregated curve (figure 7) achieves a maximum in the ratio of conformable surface relative to the total area of conformable and unconformable surface at the Base of Paleocene (corresponding to topography of the Upper Cretaceous). There are also two prominent maxima in the ratio of the area of unconformable surface relative to the total area of conformable and unconformable surface at the Base of Miocene and the Base of Pleistocene, respectively. The curves for individual continents (figure 8) are more variable. They reveal considerable differences between continents and series.

Figure 7.

Ratio of the area of un/conformal (solid red/blue lines) surface relative to the total area of conformal and unconformal surface aggregated over North/South America, Europe, Africa and Australia from the Base of Lower Cretaceous to the Base of Pleistocene, indicative of mean relative elevation (blue = low, red = high) across the continents in the preceding series (see text). The spherical harmonics (SH) area of conformal and unconformable surface is taken within the amplitude range (greater than or equal to 0.1) for a tapered cut off at degree 15 (figure 6). The $//$ and $\setminus \setminus$ shaded envelopes represent the ratio variations that correspond to tapered cut offs in the SH surface at degree 2 and 30, respectively (compare with figure 5). A maximum in the ratio of conformable surface at the Base of Paleocene (corresponding to mean topography in the Upper Cretaceous) relative to the total area of conformal and unconformal surface agrees with global sea-level curves (e.g. [65,66]). Two maxima in the ratio of unconformable surface relative to the total area of conformal and unconformal surface at the Base of Miocene and the Base of Pleistocene coincide with the onset of glaciation in Antarctica [67] and the Northern Hemisphere [68], respectively (see text). The total area (within the amplitude range (greater than or equal to 0.1) for a tapered cut off at degree 15) of conformal and unconformal surface relative to the total area of the considered continents is shown by the grey curve. The grey hatched $//$ and $\setminus \setminus$ shaded envelopes represent the ratio variations that correspond to tapered cut offs in the SH area at degree 2 and 30, respectively. (Online version in colour.)

Figure 8.

Same as figure 7, but for individual continents. The curves are more variable and reveal considerable differences between continents and series. See text for interpretation. (Online version in colour.)

4. Discussion

Geodynamicists have long recognized the essential role of dynamic topography in studies of the Geoid because the mass anomalies associated with surface deflections yield gravity anomalies of comparable amplitude with the flow-inducing mantle density variations. Geoid models therefore account for dynamic topography as well as mantle density heterogeneity (e.g. [62,69,70]) However, it is difficult to separate dynamic topography from topography in isostatic support [71–73] outside the oceanic realm [63]. This has led some to doubt the existence of dynamic topography [74].

The transient nature of dynamic topography suggests to overcome this difficulty by turning to geologic archives. Ahead of his time, Bond [75,76] analysed continent-scale sediment distributions to argue for substantial uplift of continental platforms. He concluded that Africa, for instance, experienced late Tertiary uplift relative to other continents [77], in agreement with Burke and Whiteman [78]. Our interregional hiatus maps also turn to sedimentary archives, in the form of interregional unconformities. But we note that the existence of such unconformities has long been known (e.g. [32,37,40,79–86]) and that some have pointed out the need of physical models for their interpretation (e.g. [20,86]).

Our GHMs locate sedimentary rocks of any origin, including volcanic effusive and pyroclastic products that, for the purpose of mapping depositional sequences, behave like sediments. Thus, to first order, the time slices in figure 4 show, where sediments were or were not deposited (or deposited and then eroded before the deposition of the next series) in the series immediately preceding the target series. Surfaces of unconformable contact (marked in red) in the BHS (figure 6) define regions in the series immediately preceding the target series that undergo erosion and/or non-deposition, whereas areas of conformable contact (marked in blue) identify depositional regions. At the interregional scales invoked, this serves as a proxy for either exhumation and surface uplift, or burial and subsidence. Lack of signal in the BHS indicates the absence of sediments in the target series and its immediately preceding series. This describes regions that may have undergone intense and/or long-lasting erosion or non-deposition and suggests intense and/or persistent exhumation and surface uplift [30,31,34,35]. Examples for un/conformable surfaces and for lack of signal can be identified in the BHS.

South America reveals a continent-scale lack of signal at the Base of Eocene and the Base of Oligocene (figure 6d,e), indicating early Tertiary uplift in the region. This coincides temporally with the onset of rapid South Atlantic spreading rates [87] and an Eocene subaerial exposure of the Rio Grande Rise at Drill Site 516 [88]. There are also reports from thermochronological data and landscape analysis for post-rift Eocene reactivation in Brazil [89–91] and Argentina [92], and there is a Paleogene hiatus documented in Andean Foreland Basins [93].

Expansion of the total unconformable area from one time slice to the next indicates the onset of relative subsidence; it means that sediments now deposit in areas that previously underwent erosion/non-deposition. A significant expansion of unconformable area in central and northern Africa occurs at the Base of the Upper Cretaceous when compared with the Lower Cretaceous (figure 6g,h) and suggests that the Upper Cretaceous was a period of subsidence in Africa. An exception is the South African Plateau (SAP). It reveals a lack of signal suggestive of net high elevation. While this agrees with reports by some authors [94–96] calling for a Cretaceous age of the SAP topography, others suggest more recent Oligo-Miocene or younger uplift phases [35,97]. Another major expansion of unconformable area across Africa occurs at the Base of Miocene when compared with the Base of Oligocene (figure 6c,d). It implies relative subsidence in the Miocene and suggests that the Oligocene was a period of uplift in most of the continent, as noted by several authors [26,35,77] and reviewed very effectively by Burke and Gunnell [97]. A recent geologic/geodynamic analysis suggests that Africa may cover different dynamic topography domains owing to its large area. Carena et al. [35] took the presence of Upper Cretaceous to Eocene exposed marine sediments in the interior of northern Africa together with the absence of exposed Oligocene to Pleistocene marine sediments there as evidence that this region uplifted significantly after the end of the Eocene, remaining high since. Oligocene to recent sediments in northern Africa are exclusively of continental origin. Far less marine sedimentation exists in the southern half of Africa for the Cenozoic series, where it is limited to coastal regions. While none of the exposed Cenozoic sediments in the interior of southern Africa are marine, there is a complete absence of coastal marine sediments in the Oligocene and Pleistocene. From this, and from the observation that some Miocene and Pliocene marine sediments along the southern coast are now at elevations significantly above 200 m, Carena et al. [35] inferred that southernmost Africa reached a high elevation in the Oligocene, subsided in the Miocene–Pliocene, and has been high again since the Pleistocene.

Europe features a strong expansion of unconformable area at the Base of Eocene when compared with the Base of Paleocene (figure 6e,f ), indicative of relative subsidence in the Eocene. We note that the above examples of expanding unconformable area follow each in the wake of major plume events (i.e. Tristan, Lower Cretaceous; Afar, Oligocene; and Iceland, Paleocene, see figure 6).

Conformable area expansion from one time slice to the next indicates continued subsidence [31,34,35]. Prominent examples include Australia at the Base of Upper Cretaceous when compared with the Base of Lower Cretaceous (figure 6g/h), and western North America at the Base of Paleocene when compared with the Base of Upper Cretaceous (figure 6f /g). Continent-scale subsidence implied by growing conformable area in these regions has been linked to subduction at the eastern margin of Gondwana [98,99] and to the descent of the Farallon Plate beneath western North America [19,20,100].

Figures 7 and 8 show the temporal evolution in the ratio of the area of un/conformal surface relative to the total area of conformal and unconformal surface, both aggregated over all continents and separate for each. The aggregated curve (figure 7) reveals a sea-level signal. It is indicated by a maximum in the ratio of the area of conformable surface at the Base of Paleocene (corresponding to the Upper Cretaceous) relative to the total area of conformal and unconformal surface. The maximum agrees with global sea-level curves even when the amplitude of the latter is not well constrained (e.g. [65,66]). There are also two prominent maxima in the ratio of the area of unconformable surface relative to the total area of conformal and unconformal surface at the Base of Miocene and the Base of Pleistocene. These coincide with the onset of glaciation in Antarctica [67] and the Northern Hemisphere [68], respectively.

The curves for individual continents (figure 8) provide additional information: a sharp decline for North America in the ratio of conformable surface relative to the total area of conformal and unconformal surface at the Base of Eocene marks the disappearance of the Interior Seaway in the western part of the continent. South America displays a gradual growth with time in the ratio of conformable surface relative to the total area of conformal and unconformal surface. The lack of signal at the Base of Eocene and the Base of Oligocene, which we noted before in the BHS (figure 6d,e), is evinced in figure 8 by the drop in the grey curve reporting the ratio of the total conformal and unconformal surface relative to the total area of South America. Europe’s ratio of conformable surface relative to the total area of conformal and unconformal surface sinks dramatically at the Base Eocene, in agreement with the continent-scale growth of unconformable surface and the implied Eocene subsidence that followed the arrival of the Iceland Plume. Africa incurs two increases in the ratio of unconformable surface relative to the total conformal and unconformal surface at the Base of Upper Cretaceous and the Base of Miocene, presumably reflecting Upper Cretaceous and Miocene subsidence as discussed before. Notable for Australia is the increase in the ratio of conformable surface to the total un/conformable surface at the Base of Upper Cretaceous, attributed to Australia’s eastward passage over subducted oceanic lithosphere. These results are in broad agreement with the analyses of [75,76] and support the notion that there are no stable continental platforms [101].

In our discussion, we must point to the severe limitations of our method. First: GHMs strongly depend on the spatio-temporal resolution and accuracy of data compiled on geological maps. This means that the duration over which a particular hiatus area is defined depends on the temporal resolution of the input geological map, as noted before (figure 3). Our analysis is limited to the series level. But true hiatus is likely longer than indicated by the missing series, because at any one location sedimentary successions represent only a small portion of a series. This implies large temporal uncertainties in our analysis, even when only one series is absent or when the adjacent series is not fully represented in the field. While our saturation of the time intervals to the series level is dictated by the data (i.e. the geological convention), it inevitably hides shorter duration lacunae and thereby avoids artefacts related to the Sadler effect [102]. This is critical, because if shorter duration lacunae are hidden, shorter duration events from lithospheric processes may be conflated with longer duration mantle-driven signals. Essentially our method favours large time intervals and hides shorter time intervals. Krob et al. [103] deduced an uplift duration signal of 50 Myrs for the Parana-Etendeka plume. So even at the temporal resolution of series it may be difficult to detect plume-related uplift events. A similar difficulty arises when continents move laterally over different dynamic topography domains in relatively short geological time frames [104]. Future stratigraphic work should therefore respond to the geodynamic need for more precise dating of hiatus. Interregional geological maps at the resolution of stages (1–2 Myrs) are needed to reduce the uncertainties and to assist in geodynamic interpretations of hiatus.

Second: GHMs on their own do not identify the lithospheric or sublithospheric causes for continental vertical motion. Models predicting continental rise under increased horizontal stress (e.g. [105]), lithospheric folding [106] or delamination (e.g. [107,108]), which act as tectonic mechanisms within the lithosphere, must be distinguished from deeper, mantle-related effects, such as the influence of rising plumes or pressure-driven asthenosphere flow. Detailed biostratigraphy and geomorphological methods of slope investigation or planation surfaces [109] are needed in the identification of broad scale (falcogenic) structures in the sense of Şengör [37]. It is clear that viable dynamic models of lithosphere motion must provide for a coupling of tectonic and mantle-related forces (e.g. [110]) to represent the behaviour of the lithosphere as the combination of lithospheric and sublithospheric effects.

Third: GHMs are well constrained in lateral extent but not in amplitude. The latter requires independent calibration, for example, by using thermochronological data [111]. A variety of inferences provide constraints on surface uplift of the lithosphere. They include studies of river profiles (e.g. [112]), sediment compaction [113] and provenance [26,114], landform analysis [109] based on planation surfaces [115], paleoaltimetry [116], or the analysis of sediment budgets at the scale of continental margins [117–119]. Passive margins have been advocated as suitable locations for such studies [18]. MacGregor [120] summarizes episodes of margin uplift for South America and Africa, and similar inferences have been made for the Arctic [121] and the European passive margin of the North Atlantic, summarized in the Stratagem project ([122] and references therein). Inferences for an active post-rift evolution of passive margins have been collected into propositions for geodynamic models [123]. Geological hiatus maps suggest to extend the studies to broader spatial scales beyond passive margins.

(a). Geodynamic implications

Geodynamicists explore mantle convection in terms of the plate and plume modes. Hiatus maps reveal the plate mode as broad conformable surfaces at the Base of Upper Cretaceous in Australia (figure 6g) and the Base of Paleocene in western North America (figure 6f ), as noted before. Unconformable surfaces and areas of lack of signal located away from active plate margins are instead expressions of the plume mode. Seismic evidence suggests a strong plume mode [12–14], imaged for the upper [124–126] and the lower mantle as prominent regions of seismically slow velocities (e.g. [127–130]). The geodynamic analysis of these anomalies remains under debate and permits interpretations of the lower mantle anomalies primarily by elevated temperature [129,131,132] or combinations of thermal and compositional effects [133,134]. The repeated appearance of continent-scale hiatus surfaces in our maps provides additional constraints. It implies significant positive mantle buoyancies presumably related to elevated temperature.

The distribution of un/conformable surface varies at the timescale of geologic series, (i.e. ten to a few tens of Myrs). This is considerably faster than the mantle transit time which, as the timescale for convection, is about 100–200 Myrs [38,135]. The difference in the convective timescale and the timescale for topography in convective support is illustrated by geodynamic kernels. They reflect the properties of dynamic Earth models and depend strongly upon the assumed rheology (see [136] for a review). For internal loads (e.g. hot rising plumes or cold sinking slabs) passing through a uniform-viscosity mantle, the kernels predict a continuous evolution of the induced surface deflections. In other words, a comparable timescale for convection and convectively maintained topography is implied and borne out in laboratory models of isoviscous mantle flow [137]. The presence of a weaker upper mantle relative to the lower mantle, which is consistent with inferences from geodynamics [62] and mineral physics modelling [138], amplifies surface deflections for loads passing through the upper mantle. This property of dynamic Earth models makes rapid changes of convectively maintained topography geodynamically plausible.

Geological hiatus maps have implications for time-dependent geodynamic Earth models: progress has been made in understanding how to retrodict past mantle states. Early backward advection schemes (e.g. [139,140]) have given way to a formal inverse problem based on adjoint equations that provide sensitivity information in a geodynamic model relative to earlier system states. Adjoint equations have been derived for incompressible [141–143], compressible [144] and thermo-chemical [145] mantle flow, and the uniqueness property of the inverse problem has been related to the tangential component of the surface velocity field of the convection model [146]. Knowledge of the latter is essential to ensure convergence [147,148]. While plate motions are a primary surface expression of mantle convection (e.g. [3]), one needs to assimilate the tangential component of the surface velocity field (i.e. a past plate motion model) to solve the inverse problem. This makes past plate motions the input of retrodictions rather than their output, and suggests linking viable tests of retrodictions to inferences of past dynamic topography so that uncertain model parameters and state estimates can be assessed [149]. Put differently: the horizontal motion of the lithosphere cannot be predicted from mantle flow restorations, because reconstructions of past plate motion act as an input to the inverse problem, implying that it is not viable to construct self-consistent models of plate tectonics that are testable against the geologic record. However, mantle convection also induces vertical motion in the form of dynamic topography, as noted before. These can be inferred from a mantle flow retrodiction, because they are an output of the inverse problem. Geologic constraints on the history of convectively induced vertical motion of the lithosphere (that is the evolution of past dynamic topography) therefore are crucial observations to test the validity of the geodynamic modelling parameters assumed in mantle flow retrodictions. Our results imply that changes in convectively maintained topography at the timescale of geologic series and over spatial scales of a few thousand kilometres must be an integral component of time-dependent geodynamic Earth models.

5. Conclusion

The analysis of continent-scale geological maps yields powerful information for constraining large-scale geodynamic processes and models. By providing consistent compilations of geologic observations at the scale of thousands of kilometres, continent-scale geologic maps link naturally to large-scale mantle flow-induced elevation changes known as ‘dynamic topography’ [16,17]. While the latter is difficult to separate by geophysical or geodetic means from the current isostatic topography of our planet outside the oceanic realm [63], its transient nature leaves signals in sedimentary archives as conformable and unconformable (hiatus) time boundaries traceable over hundreds to thousands of kilometres. We have applied a hiatus mapping method, introduced by [30,31], as a first-order technique that uses a single manipulation of existing geological maps to construct hiatus surfaces at the temporal resolution of series across North and South America, Europe, Africa and Australia starting from the Upper Jurassic. We find significant differences in the spatial extent of hiatus surface across and between continents at the timescale of geologic series, ten to a few tens of Myrs. This is considerably smaller than the mantle transit time [38] and may reflect the effects of rapid lateral motion of continents over different dynamic topography domains in relatively short geological time-frames [104] as well as vigorous upper mantle flow in the asthenosphere facilitated by a viscosity reduction from the lower to the upper mantle as implied by response functions for dynamic Earth models (e.g. [62]). The recurrent appearance of continent-scale hiatus surfaces is consistent with the existence of significant positive mantle buoyancies, presumably induced by thermal effects and elevated temperature. This supports the notion of a strong plume mode in the mantle convection system. In the future, it is necessary to compile interregional geological maps at the temporal resolution of stages, most of which span 1–2 Myrs in duration, to reduce uncertainty and to assist in improved geodynamic interpretations of hiatus through time-dependent geodynamic Earth models capable of retrodicting past mantle flow states.

Data accessibility

As electronic supplementary material the fully normalized spherical harmonics coefficients [61] are provided for the base of each geological time series for the Cenozoic and Cretaceous.

Competing interests

We declare we have no competing interests.

Funding

This research has been supported by the European Union’s Horizon 2020 Research and Innovation Programme under the ERC-2019-STG project TEAR, grant no. 852992.