Contents lists available at ScienceDirect

Palaeogeography, Palaeoclimatology, Palaeoecology

journal homepage: www.elsevier.com/locate/palaeo

Quantifying the paleogeographic driver of Cretaceous carbonate platformdevelopment using paleoecological niche modeling

Alexandre Pohla,⁎, Marie Laugiéb,a, Jean Borgomanoa, Julien Michela, Cyprien Lanteaumea,b,Christopher R. Scotesec, Camille Fraud, Emmanuelle Polie, Yannick Donnadieua

a Aix Marseille Univ, CNRS, IRD, Coll France, INRA, CEREGE, Aix-en-Provence, FrancebMODIS Pau, 4 Rue Jules Ferry, Pau, Francec Department of Earth & Planetary Sciences, Northwestern University, Evanston, IL, USAdGroupement d'Intérêt Paléontologique, Science et Exposition, 60 bd Georges Richard, 83000 Toulon, Francee Total CSTJF, Avenue Larribeau, 64000 Pau, France

A R T I C L E I N F O

Keywords:PaleoceanographyCarbonate factoryEnvironmental constraintsGeneral circulation modelFuzzy logicAptian

A B S T R A C T

Platform carbonates are a major component of the Earth System but their spatial extent through geological timeis difficult to reconstruct, due to the incompleteness of the geological record, sampling heterogeneity and theirintrinsic complexity. Here we use coupled ecological niche modeling and deep-time general circulation modelsto predict the occurrence of platform carbonates at the global scale during the Cretaceous. Specifically, nichemodeling uses fuzzy logic to predict probable occurrence of platform carbonates as a function of sea-surfacetemperature, sea-surface salinity, primary productivity and water depth. The first three parameters derive fromCretaceous global paleoclimatic simulations using a coupled ocean-atmosphere general circulation model, whilebathymetry is based on paleogeographical reconstructions. Model predictions are validated with the well-documented and abundant geological data from the Aptian. The methodologies developed for the Aptian aresubsequently extended to other Cretaceous time intervals. The results of the niche model accurately predict thegeographic distribution of Aptian carbonate platforms if a preference for low open-marine productivity levels isassumed for the rudist-dominated carbonate factories. However, if a preference for high productivity levels isassumed, the modeling results do not match the reported global distribution of Aptian carbonate platforms. Fromthe Early Cretaceous into the Late Cretaceous sea level rose and the continents, on average, moved into lowerlatitudes. The model predicts a corresponding increase in the extent of the carbonate platforms, mainly due tothe increasing extent of shallow-water environments available to carbonate development. These results indicatethat long-term sea-level rise may have been a major factor responsible for the increase in the area of platformcarbonates during the Cretaceous.

1. Introduction

Constraining the extent and location of carbonate platforms andmore specifically reef carbonates throughout geological time has atwofold importance. From an academic perspective, reefal carbonatesplay an essential role in the global carbon cycle and thus the regulationof atmospheric CO2 concentration. Today, coral reef calcification ac-counts for> 25% of the total amount of carbonate buried in marinesediments globally (Jones et al., 2015). When extensive carbonateplatforms are subducted, decarbonation reactions provide an importantsource of volcanic carbon (Mason et al., 2017). On shorter time scales,the modulation of the production of carbonate platforms may criticallyimpact global climate state. Donnadieu et al. (2011) notably

demonstrated that short-lived cold interludes occur when the produc-tion of platform carbonates is severely reduced, such as at the Mid-dle–Late Jurassic Transition (160Ma; Dromart et al., 2003). This de-crease in global mean temperature is caused by the draw-down ofatmospheric CO2 and results in ice-sheet buildup. Beyond these ex-clusively academic interests, ancient reef carbonates also have a higheconomic value. Carbonates, reefs, and reef-associated sediments oftenform important hydrocarbon reservoirs. Burchette (2012) estimatedthat 50% to 60% of the world's conventional petroleum is hosted incarbonate reservoir rocks.

The creation of a comprehensive database of ancient reefs andcarbonate platforms provides an unparalled window into the abun-dance and geographic extent of platform carbonates during the

https://doi.org/10.1016/j.palaeo.2018.10.017Received 29 August 2018; Received in revised form 19 October 2018; Accepted 20 October 2018

⁎ Corresponding author.E-mail address: pohl@cerege.fr (A. Pohl).

Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

Available online 23 October 20180031-0182/ © 2018 Elsevier B.V. All rights reserved.

T

Phanerozoic (e.g., Kiessling et al., 1999[updated], 2003). Nevertheless,such a compilation necessarily suffers limitations associated with in-complete preservation and heterogeneous geographic sampling(Kiessling, 2005). These problems limit utilization of the database whenanalyzing the evolution of platform carbonates through geological time(Kiessling, 2001, 2005). A promising, complementary approach wasrecently suggested by studies that investigated the environmentalcontrols on the distribution (Couce et al., 2012; Michel et al., 2018) andproduction (Jones et al., 2015) of shallow-water coral reefs. Based onthe environmental limitations to reefal development (see Kleypas et al.,1999; Lauchstedt et al., 2017), Couce et al. (2012) built transfer func-tions deriving the probability of occurrence of coral reefs from en-vironmental parameters, such as sea-surface temperature.

Here we build on these pioneering studies and propose spatially-explicit models that predict the occurrence of platform carbonatesbased on the estimates of environmental variables provided by globalsimulations of Cretaceous climate and marine biogeochemistry. We firsttest the validity of our predictive model on the Aptian (EarlyCretaceous), which has long been recognized as a period of prolificcarbonate platform development and is one of the best known timeintervals of the Cretaceous (e.g., Steuber, 2002; Skelton and Gili, 2011).We then extend our simulations to the earliest and latest Cretaceous inorder to quantify the impact of the paleogeographic changes on thedevelopment of platform carbonates throughout the Cretaceous.

2. Methods

The rationale is to predict the geographic occurrence of platformcarbonates by mapping the environmental parameters that have beenpreviously documented to control the production of shallow-water,marine carbonates. In this section, we first design a formula that useslocal environmental constraints to predict the likelihood of carbonateoccurrence at each location (i.e. model grid point; Section 2.1). We thenintroduce results from four Cretaceous climatic simulations that pro-vide the above-mentioned environmental constraints (Section 2.2).

2.1. Building the paleoecological niche model

Couce et al. (2012) used species distribution modeling to predict thepresent-day occurrence of shallow-water coral reefs based on numerouspredictive variables including sea-surface temperatures, the intensity ofcyclone activity, and dust. However, in order to predict occurrences ofplatform carbonates in deep time, we were restricted by the environ-mental variables that can be simulated in general circulation models.Temperature and salinity were the first oceanographic parametersshown to control carbonate sedimentation (Lees, 1975). In addition,they provide a reasonable estimate of carbonate saturation at a globalscale (e.g., Kleypas et al., 1999; Jiang et al., 2015). An additional cri-tical parameter was net primary productivity, which supplies food toheterotrophic organisms and provides a fair picture of water transpar-ency, an essential control of photo-autotrophic production. An estimateof water depth obtained from paleogeographic reconstructions (Scotese,2014a, 2014b) further constrained regions of neritic carbonate pro-duction. In summary, four parameters were used in the following cal-culations: sea-surface temperature (SST), salinity (SSS), net primaryproductivity (NPP) and water depth (bathymetry). The first threevariables (SST, SSS and NPP) come from global simulations of Cretac-eous climate and marine biogeochemistry and the fourth, the estimateof water depth, was based on paleogeographical reconstructions.

It is noteworthy that the spatial resolution of the global, deep-timegeneral circulation model employed to simulate oceanic fields of SST,SSS and NPP in the Cretaceous (~200 km×200 km; see Section 2.2)does not allow resolving intra-shelf variations in these oceanographicparameters. The adopted approach consists in predicting the suscept-ibility of occurrence of the platform carbonates at the platform scalebased on the conditions that characterize the neighboring open ocean

(see discussion in Section 4.1).Fuzzy logic (Zadeh, 1965) was used to compute the susceptibility of

Cretaceous platform carbonate occurrences in each point of the Earth asa function of SST, SSS and NPP (Eq. (1)). For each environmentalvariable under investigation (i.e., SST, SSS or NPP), a membershipfunction f(x) (Fig. 1, see Eq. (1)) was used to normalize global fields ofthe variable representative of June–July–August and

Fig. 1. Fuzzy functions. Each curve represents the susceptibility of occurrenceas a function of the values of (A) SST, (B) SSS and (C) NPP. In panels (A) and(B), histograms represent the number of coral reef occurrences (left y-axis)reported in the present-day ocean as a function of the values of the parameterduring respectively winter (blue shading) and summer (yellow shading). Greyshadings represent the superposition of the summer and winter histograms. Inpanel (A), the dashed black line represents the function established based on thedistribution of present-day shallow-water coral reefs while the solid black line isthe function adapted to Cretaceous platform carbonates (right y-axis). In panel(C), red and blue lines respectively represent the functions established to re-present an affinity for either low or high open-ocean productivity levels. Valuesof the parameters used to define each fuzzy function are given in Section S3.(For interpretation of the references to colour in this figure legend, the reader isreferred to the web version of this article.)

A. Pohl et al. Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

223

December–January–February. The resulting 2 normalized, seasonalfields (values between 0 and 1) were then averaged to obtain the spa-tially varying contribution of the variable to the likelihood of the oc-currence of the platform carbonates (see Eq. (1)). The contributions ofSST, SSS and NPP were subsequently multiplied by their respectiveweighting factors (respectively a, b and c in Eq. (1)). The resultingsusceptibility of occurrence was then sorted by bathymetry. Only lo-calities that occur in shallow-water environments were preserved; alldeeper-water occurrences were eliminated, similar to Jones et al.(2015). The final output (S in Eq. (1)) is a latitudinal/longitudinal gridof values (0 to 1) that represent the predicted susceptibility of the oc-currence of the platform carbonates. Grid points with values above0.85, a value determined empirically by comparison with an extensiveindustrial (TOTAL in-house) and public (Kiessling et al., 2003) data-base, were used to define the potential ecological niche of Cretaceousplatform carbonates.

= × + × + × ×S [a f(SST) b f(SSS) c f(NPP)] bathy_mask (1)

with:- S: susceptibility of occurrence of the platform carbonates;- a+ b+ c=1: inter-parameter weighting factors;- bathy_mask: value 1 over shallow-water platforms and 0 in open-

ocean areas;- f(x): susceptibility of occurrence for each parameter x, or equiva-

lently, contribution of the parameter x to the final prediction S.The membership functions f(x) were first defined based on statistics

established on present-day reef carbonate occurrences (similar toKleypas et al., 1999) and then adapted to the Cretaceous based onpublished literature. The sedimentary profile of Cretaceous carbonateplatforms generally features rudist bivalves in the inner platform andcorals in a more distal position (Gili et al., 1995; Johnson et al., 2001;Borgomano et al., 2002; Fenerci-Masse et al., 2005; Pomar and Hallock,2008; Masse and Fenerci-Masse, 2011). Such carbonate platforms be-long to the tropical shallow-water carbonate factory sensu Schlager(2000). We attempted to simulate the extent of this factory at the globalscale based on the environmental requirements of the main bioticcontributors. Rudists usually constitute the bulk of these carbonateplatforms (Pomar and Hallock, 2008), but they are now extinct andtheir ecology is poorly known (Scott, 1995; Fenerci-Masse et al., 2005;Gili and Goetz, 2018). As a first attempt, f(SST) and f(SSS) (Eq. (1))were thus defined based on the current distribution of coral-dominatedtropical reef platforms (i.e., photozoan carbonates; James, 1997). To doso, we extracted the values of SST and SSS at each of their present-daycoordinates and designed functions that satisfactorily reproduce thedistribution of these variables (Fig. 1A, B). The geochemical analysis ofrudist shelves provided temperature estimates as high as 30–37 °C(Steuber, 1999; Immenhauser et al., 2005; Steuber et al., 2005). Thiscontrasts with present-day shallow-water coral reefs that do not thrivein waters warmer than 30 °C (Hoegh-Guldberg, 1999). In order to ac-count for the adaptation of the rudist bivalves to high ocean tempera-tures, the temperature function primarily established based on the oc-currences of present-day coral reefs – f(SST) – was then extended tohigher sea-surface temperatures to adequately represent the rudist-dominated Aptian carbonate platforms (Fig. 1A). Given the presumablybroader thermal tolerance of the rudists, the temperature function wasalso extended towards lower values. Based on a statistical analysis ofpresent-day coral reef tolerances, 18 °C was identified as the lowesttemperature supporting coral reef persistence during winter. Undersuch cool conditions, coral reefs can continue to develop provided thatsummer temperatures are above 24 °C. In our model of Cretaceouscarbonate platforms, the latter constraint was discarded and tempera-tures above 18 °C were considered favorable to the development ofrudist-dominated platform assemblages (Fig. 1A).

A straightforward comparison of present-day NPP and simulatedNPP values for the Cretaceous was not possible. The general circulationmodel used in this study (MITgcm) does not simulate the same high

NPP values that are reported for modern coastal environments. This ismainly due to the fact that riverine input and dust deposition are notrepresented in the model and the model resolution (~200 km) does notcapture fine-scale coastal upwelling systems that are only a few tens ofkilometers wide (see the model description in Section 2.2) An addi-tional explanation may be the simplified representation of NPP (seediscussion in Section 4.2).

Two theoretical, opposite NPP membership functions (i.e., f(NPP) inEq. (1)) were built to discriminate between the high NPP values typi-cally simulated along strongly upwelling western continental marginsand lower NPP values that characterize the water masses downwellingalong eastern continental margins (Fig. 1C). These two domains re-present preferential conditions for respectively heterotrophic andphoto-autotrophic carbonate productions (Schlager, 2005; Michel et al.,2018). The ecological affinities of Cretaceous rudists are today thesubject of controversy (e.g. Steuber, 2000; Skelton and Gili, 2011). Mostauthors suggest that rudists were heterotrophic; a few others suggest aphoto-autotrophic affinity. We considered both options and tested forboth a photo-autotrophic and heterotrophic affinity for the platformcarbonates by considering each of the two opposite f(NPP) membershipfunctions that favor either low or high NPP levels.

Weighting factors (a, b and c in Eq. (1)) were defined using theanalytical hierarchy process (AHP) method (Saaty, 1987), which as-signs higher weights to environmental parameters that are expected tohave a higher impact on the distribution of the platform carbonates.The adopted set of factors (a= 0.660 [SST], b= 0.062 [SSS] andc= 0.278 [NPP]) gives a greater weight to SST and NPP. This is sup-ported by the hierarchy of constraints suggested by Couce et al. (2012)in their attempt to predict present-day shallow-water coral reef occur-rences (see also Kleypas et al., 1999).

2.2. Simulating Cretaceous environmental conditions

To obtain maps of Cretaceous environmental parameters (i.e., SST,SSS, NPP), we ran Cretaceous global climatic simulations using theMassachusetts Institute of Technology general circulation model(MITgcm; Marshall et al., 1997; Adcroft et al., 2004). This is a coupledocean-atmosphere-biogeochemistry model that has been recently ap-plied to simulate the climate of other deep-time periods (Brunetti et al.,2015; Pohl et al., 2017a, 2017b).

We ran the model on a curvilinear cubed-sphere grid (CS48; Forgetet al., 2015) featuring a mean equatorial resolution of 1.85°× 1.85°(~200 km×200 km). In the ocean, we defined 28 layers verticallywith thicknesses gradually increasing from 10m at the ocean surface to748m at the bottom, with 18 levels defining the upper 1000m of thewater column. Effects of mesoscale eddies were parametrized as anadvective process (Gent and McWilliams, 1990) and an isopycnal dif-fusion (Redi, 1982). Vertical mixing processes in the ocean's surfaceboundary layer and the interior were represented using the nonlocal K-Profile Parameterization (KPP) scheme of Large et al. (1994). Thephysical ocean-atmosphere model was coupled with a module ofmarine biogeochemistry. The latter was configured to simulate netprimary productivity in the ocean as a function of available photo-synthetically active radiation and phosphate concentration, withphosphate constituting the single limiting nutrient (see discussion inSection S1.2 in Supporting Information). Because the oceanic residencetime of phosphate is much longer than the oceanic turnover time scale(10 to 40 kyrs; Ruttenberg, 1993; Wallmann, 2003), the globally-aver-aged oceanic phosphate concentration was fixed in the model. Riverineand atmospheric sources were not represented and sedimentation wasnot allowed (Dutkiewicz et al., 2005).

Paleogeographical reconstructions for 150Ma (late Jurassic),120Ma (early Aptian), 110Ma (early Albian) and 80Ma (earlyCampanian) of Scotese (2014a, 2014b) were used as boundary condi-tions. The bathymetry of oceanic crust in the deep ocean, most of whichhas been lost during subduction, was replaced by the age-depth

A. Pohl et al. Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

224

estimate of Müller et al. (2008). Model value for solar luminosity wasdecreased by 1% compared to the present-day value, in agreement withmodels of solar evolution (Gough, 1981). Cretaceous atmospheric CO2

concentration (pCO2) is assumed to range between 2 and 8 times thepreindustrial atmospheric level (PAL= 280 ppm; Bice et al., 2006;Donnadieu et al., 2016). We adopted a pCO2 of 2 PAL (560 ppm) for theearly Aptian, a time when ocean temperatures were at the lower end ofCretaceous temperature range (Steuber et al., 2005; O'Brien et al.,2017). A constant value of 2 PAL was used in every Cretaceous climatemodel run in order to highlight the impact of the changes in the con-tinental configuration, all other things being equal. Baseline runs oneach continental configuration were conducted using a median orbitalconfiguration that is relatively similar to the present-day configuration.Additional simulations with contrasting orbital parameters were run forthe Aptian (see Section S1.3).

Identical initial conditions were used in every simulation. We de-fined the initial ocean temperature as a theoretical latitudinal tem-perature gradient, characterized by equatorial and polar ocean surfacetemperatures of respectively 40 °C and 10 °C, and an ocean bottom in-itial potential temperature of 10 °C. A uniform initial ocean salinity of35 psu (practical salinity units) was imposed. Phosphate in the oceanwas initialized with its present-day depth profile, and dissolved organicphosphorus was null at the beginning of the model runs. The ocean-atmosphere model was integrated for 3000 years in order to ensuredeep-ocean equilibrium. The marine biogeochemistry module wasturned on over the last 1000 years of the model runs. The last 50 yearswere used to build the monthly climatology files used for this analysis.A thorough description of the model and boundary conditions alongwith a table of experiments is provided in Section S1.

3. Results

3.1. Climate results

Fig. 2 shows that the SST simulated at 120Ma is in overall agree-ment with published estimates from O'Brien et al. (2017), althoughsome data points of O'Brien et al. (2017) lie beyond the SST range si-mulated in the model. Observed model-data mismatch has various po-tential explanations. It is plausible that our model does not accuratelysimulate Cretaceous climate. In particular, the “low pole-to-equatortemperature gradient” issue is a long-standing question. It is frequentlyargued that general circulation models are not able to capture moreequable latitudinal temperature gradients presumably characteristic ofwarmer, greenhouse climates (see Huber and Caballero, 2011; Rose andFerreira, 2013). However, recent geochemical data (Pucéat et al., 2007;Bernard et al., 2017) indicate that Cretaceous meridional temperaturegradients may have been steeper than supposed and thus much moresimilar to the present-day gradient. This suggests that numerical modelscould well provide a reasonable picture of warm greenhouse climates(see Donnadieu et al., 2016). Huber and Caballero (2011) further de-monstrated that current ocean-atmosphere general circulation modelsare able to capture most of the climatic signal reconstructed for theearly Eocene greenhouse based on proxy data, suggesting that the firstorder physics of climate are well represented in these models.

Besides modeling limitations, it is well known that the Aptian was aperiod characterized by rapid and abrupt climatic changes. Notably,high-resolution geochemical (TEX86) records demonstrate a strongwarming during the Oceanic Anoxic Event (OAE) 1a (up to 4 °C in theproto North Atlantic Ocean; Naafs and Pancost, 2016), followed bycooling of similar magnitude (McAnena et al., 2013; Bottini et al.,2015). The remainder of the period is characterized by a protractedcooling trend, interrupted by an episode of warming during the latestAptian (Bottini et al., 2015). The warming associated with OAE 1a canbe seen in Fig. 2. Data points shown in red, which are associated withparticularly high SST values, belong to OAE 1a warm interval (O'Brienet al., 2017). Our model runs do not capture this short-term event, the

duration of which is estimated to represent 1.11 ± 0.11Myr(Malinverno et al., 2010). Our goal was to simulate average backgroundAptian climate. We did not attempt to simulate unusual warmingphases induced by abrupt volcanic events such as OAE 1a (Mehay et al.,2009), which would at least require a doubling of the pCO2 in the model(Naafs et al., 2016). In any case, these anomalous temperature excur-sions would not be representative of the environmental conditions thatfavored platform carbonate development throughout the Aptian. In thiscontext, modeled temperatures do not cover the entire temperaturerange documented in geochemical data. Nevertheless, the models doaccount for seasonality (light grey envelope in Fig. 2), orbital variations(black dashed envelope in Fig. 2) and changes in the SST at the specificlocations where temperature proxy data have been recovered (verticaldark grey lines shown for each yellow point in Fig. 2), which helps tobring models and data into much closer agreement.

Key climatic results for additional simulations conducted at 150Ma,110Ma and 80Ma are given in Table 1. In these model runs, only thecontinental configuration was modified compared to our baseline earlyAptian run at 120Ma. The model predicts a warming trend throughoutthe Cretaceous with an increase in global surface temperatures by 4.1 °Cfrom 150Ma to 80Ma solely due to the changing continental config-uration. This warming mainly results from the gradual increase inocean surface area (by 13% from 150 to 80Ma, Table 1) and associated

Fig. 2. Comparison between our Aptian climatic simulations and the latitudinalthermal gradient based on the proxy data of O'Brien et al. (2017). The thickblack line represents the zonally-averaged annual mean SST simulated in ourbaseline run with 2 PAL CO2 and median orbital parameters. The grey enveloperepresents the range of zonally-averaged monthly SST encountered throughoutthe year in the same model run, while dashed black lines represent the envelopeextended to 4 additional simulations featuring contrasting orbital configura-tions (Section S1.3). Each data point (grey, yellow and red dots) represents themedian of the SST values reported at a specific location using either TEX86

H or∂18Opl (O'Brien et al., 2017), red dots being associated with Ocean Anoxic Event1a. For each data point the vertical, thick black line gives the spread betweenthe minimum and maximum values reported at this location. The vertical, thinblack line further represents the calibration uncertainty, which amounts to2.5 °C for TEX86

H and 0.7 °C for ∂18Opl after O'Brien et al. (2017). In order tomaximize the number of data points, a 20-Myr window centered on 120Ma wasdefined and every temperature estimate provided by O'Brien et al. (2017) be-tween 110Ma and 130Ma was included. Vertical dark grey lines represent theSST range encountered at the specific locations of each of the 3 yellow datapoints from O'Brien et al. (2017) throughout the simulations with differentorbital configurations. It differs from the dashed black envelope, which re-presents the range of zonally-averaged values. (For interpretation of the refer-ences to colour in this figure legend, the reader is referred to the web version ofthis article.)

A. Pohl et al. Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

225

decrease in surface albedo. This warming trend is in agreement withboth previous modeling studies (Donnadieu et al., 2016; Ladant andDonnadieu, 2016) and SST proxy data (O'Brien et al., 2017).

3.2. Testing the validity of the model using the Aptian database

Fig. 3 compares the results of simulations featuring an affinity ofAptian platform carbonates for either low or high NPP levels (designedhereafter as low and high productivity models, respectively). In order todetermine which model best fits geological observations, modelagreement was scored with information in the PaleoReef database(Kiessling et al., 1999[updated], public interface: https://paleo-reefs.pal.uni-erlangen.de/reefs/searchreef_public.php) (Fig. 3). The score iscalculated to represent the proportion of the points of reference that arecaptured in the simulation. A platform carbonate occurrence that iscaptured or not in the model is assigned a score of 1 or 0, respectively.Scores for all reference points are then averaged to obtain the overallscore of the simulation. We emphasize that the score does not fullytestify of the validity of the simulation since an overestimation of theextent of the platforms in the model does not lower the score. The scorevalue has to be combined with visual inspection of the results. Regionswhere the model predicts the occurrence of platform carbonates thatare not documented in the geological record have to be discussed (seebelow).

The low marine productivity model matches 81% of the observed

data points. It predicts that carbonate platforms are likely to occuralong East Africa, in the Mediterranean Tethys and in the Caribbeanregion. Carbonate platforms are less likely to occur along the westerncoast of South America. Both of these predictions satisfactorily matchthe spatial patterns reported in the reef database (Fig. 3A, B). In con-trast, the high productivity model matches only 44% of the observedoccurrences. The high productivity model notably misses the widelydocumented occurrences in the Mediterranean Tethys and predicts thedevelopment of platform carbonates along the Pacific margin of SouthAmerica, where no occurrences are reported in the reef database(Fig. 3C, D).

Fig. 4 illustrates the relative importance of sea-surface temperature(SST), sea-surface salinity (SSS), and net primary productivity (NPP) forthe predictions of carbonate platform occurrence shown in Fig. 3 (seeSection 2.1 for a description of the ecological niche model, Eq. (1) inparticular). SST and to a lesser extent SSS suggest a preferred low-la-titude zone for the development of the platform carbonates (Fig. 4A, B)while NPP modulates the values of susceptibility simulated within thislow-latitude zone (Fig. 4C, D). The low productivity model (Fig. 4C)favors the subtropics (~30° NS) and excludes the equatorial latitudesand low-latitude western continental margins where upwelling pro-motes high NPP levels (Fig. 4C). In the high productivity model, thepatterns are reversed (Fig. 4D), with low susceptibility values simulatedin the Tethys notably the Mediterranean Tethys where simulated NPP islow due to the development of an oligotrophic gyre (Fig. 5). Conse-quently, the high productivity model fails to predict the majority of thecarbonate platforms (Fig. 3D).

Because the low productivity model provides the best fit to the reefdatabase, it is adopted, hereafter, as the preferred trophic model for theAptian shallow-water factory. Comparison between the predictions ofthe simulation and the extent of known Aptian carbonate platforms(Kiessling et al., 2003) confirms that the resulting model effectivelycaptures the main patterns of carbonate platform development reportedbased on geological data (Fig. 6).

The Aptian predictions however, are not perfect. The simulated

Table 1Climatic results.

Paleogeography Global surface temperature(°C)

Ocean surface area(×106 km2)

80Ma 23.4 391.3110Ma 22.7 366.9120Ma 21.9 366.3150Ma 19.3 345.3

Fig. 3. Predicted susceptibility fields (top panels) and their equivalents thresholded at 0.85 (bottom panels) for platform carbonate models considering an affinity foreither low productivity levels (left panels) or high productivity levels (right panels). In each panel, the coastline is represented with the thick black line and the edgeof the shallow-water platforms is shown with the thin black line. In panels (B) and (D), the prediction (blue shading) is compared to the n=73 occurrences of coralsand rudists (orange dots and stars) of Aptian age referenced in the public PaleoReef Database (Kiessling et al., 1999[updated]). Model agreement is scored with theseobservations. Orange dots represent reef occurrences of the database that are predicted by the simulation while orange stars represent reef occurrences that arepreserved in the geological record but not predicted in the model. A global score is computed based on each data point, the value of which is given at the top of panels(B) and (D). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

A. Pohl et al. Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

226

extent of the platform carbonates is overestimated, when compared tothe reconstruction of Kiessling et al. (2003), in the southern subtropicallatitudes along the northern margin of India and in the nascent,southern South Atlantic. No rudists are documented in the SouthAtlantic despite the presence of carbonate platforms (Granier and Dias-Brito, 2015). In Brazil, taxa previously identified as rudists (Granieret al., 1991; white point in Fig. 7), turned out to be mostly large bi-valves with rudist-like morphologies (Granier and Dias-Brito, 2015).Based on the analysis of the faunal associations, Granier and Dias-Brito(2015) estimated that unfavorable temperature and salinity conditionswere not suitable explanations for the absence of rudists, thus indirectly

supporting our niche modeling results. Granier and Dias-Brito (2015)further identified the lack of oceanographic connection with theTethyan domain as the most probable explanation for the absence ofrudists. Interestingly, our general circulation model for the earlyCampanian (80Ma) confirms that water masses originating in Tethysdo not enter the South Atlantic Ocean despite the opened seaway be-tween the Central and South Atlantic Oceans. During the early Cam-panian ocean circulation flows northward from the South Atlantic to-wards the Central Atlantic. The Brazilian locality studied by Granierand Dias-Brito (2015) is separated from the northern part of the oceanbasin due to the establishment of a well-developed anticlockwise gyre(Fig. 7). Regarding the overestimated extent of the carbonate platformsalong the northern margin of India, it is difficult to provide any defi-nitive conclusions since most of the geological record was lost duringHimalayan collision. Sensitivity tests demonstrate that the first-orderpatterns of the simulated extent of the platform carbonates are robustwhen the orbital configuration is varied (Section S4).

3.3. Extending the model to other Cretaceous time intervals

The model predicts a monotonic increase in platform carbonatearea, by a factor of 2.7 between 150Ma and 80Ma (from 6.7 to18.1×106 km2, solid blue line in Fig. 8A). These values satisfactorilycapture the increase in the extent of the platform carbonates reportedby Kiessling et al. (2003) and Walker et al. (2002) (solid and dashed

Fig. 4. Contributions of each oceanographic parameter to the predictions shown in Fig. 3: (A) SST, (B) SSS, (C) NPP in the low marine productivity model, (D) NPP inthe high marine productivity model. Colors represent susceptibilities, with values ranging between 0 and 1. The thick stepped black line is the coastline, the thinstepped black line is the outer limit of the shallow-water environments, thus delimiting the bathymetry mask provided to the niche model. Emerged landmasses areshaded white.

Fig. 5. Mean annual ocean circulation and nutrient concentration simulated inthe Aptian baseline run in the Mediterranean Tethys, averaged over the first150m of the water column. Background shading represents the phosphateconcentration (mmol m−3). Vectors represent ocean currents. Landmasses areshaded black.

Fig. 6. Extent of the Aptian carbonate platforms (A) simulated in this study and (B) after Kiessling et al. (2003, see also Sohl, 1987). For panel (B), the original map ofKiessling et al. (2003) was georeferenced and the extent of the platforms was digitized and plotted on the reconstruction used in the current study.

A. Pohl et al. Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

227

black lines in Fig. 8A, respectively). Two main underlying mechanismsmay explain this increase in platform carbonate area: (1) changes in theenvironmental conditions (SST, SSS and NPP) that were favorable tocarbonate platform development and/or (2) changes in the bathymetryand configuration of the continents, that increased the area of theshallow-water environments available for carbonate deposition. Herewe try to disentangle these two contributions.

Fig. 8B shows that the area of the regions characterized by climaticconditions favorable to the development of platform carbonates firstincreases from 150Ma to 120Ma and then remains essentially constantfrom 120Ma to 80Ma (black line in Fig. 8B). These trends poorly re-produce the monotonic increase in the surface area occupied by theplatform carbonates (Fig. 8A), suggesting that changes in sea-surfacetemperature, sea-surface salinity, and net primary productivity are notresponsible for the strongly positive long-term trend. On the otherhand, the well-documented long-term Cretaceous sea-level rise (seecompilation by Müller et al., 2008) induces an increase in both the totalocean surface area and the surface area of the shallow-water environ-ments within the latitudinal zone favorable for reefal development (i.e.,45°N–45°S) (Fig. 9). The increase in the area of low-latitude, shallow-water environments, assuming that 46% of these shallow-water en-vironments will be occupied by carbonate platforms (value calculatedfor 150Ma), predicts an increase in carbonate platform cover (dashedblue line in Fig. 8A) similar to both the increase reported by Kiesslinget al. (2003; solid dark line in Fig. 8A), and the increase in carbonateplatform cover simulated by the model (solid blue line in Fig. 8A).

Fig. 7. Mean annual ocean circulation simulated at 80Ma in the South andCentral Atlantic Oceans, averaged over the first 150m of the water column.Background shading and vectors represent ocean currents speed, expressed incm s−1. The white point represents the location where bivalves with rudist-likemorphologies, which were previously interpreted as rudists (Granier et al.,1991), are reported by Granier and Dias-Brito (2015; Fazenda Cafuz, Sergipe,Brazil).

Fig. 8. Simulated Cretaceous trend compared to the geological database. (A) Surface area of the carbonate platforms simulated in the model throughout theCretaceous (solid blue line) compared to the trends published by Kiessling et al. (2003; solid black line) and Walker et al. (2002; dashed black line). The dashed blueline represents the purely paleogeographic contribution to the simulated trend. This latter surface contribution represents the surface area occupied by the platformcarbonates computed based on the extent of the shallow-marine environments between 45°N and 45°S (see Fig. 9) considering that the occupancy of the shallow-water seas by the platform carbonates did not change compared to 150Ma. The difference between the solid and dashed blue lines therefore represents thecontribution of other factors such as climate change and the location of the shallow-water environments (e.g., eastern vs. western ocean margin). (B) Surface area ofthe portion of the global ocean climatically favorable to platform carbonate development (black line) and the area of its spatial intersection with the shallow-waterenvironments thus representing the surface area effectively occupied by platform carbonates in the model (blue line, similar to panel (A)). Note the log scale. (Forinterpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 9. Evolution throughout the Cretaceous of the surface area occupied by theocean (blue line, left y-axis), the shallow-water environments (solid red line,right y-axis) and the shallow-water environments between 45°N and 45°S (da-shed red line, right y-axis) based on Scotese (2014a, 2014b). (For interpretationof the references to colour in this figure legend, the reader is referred to the webversion of this article.)

A. Pohl et al. Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

228

These results suggest that the paleogeographic evolution, notably theincreasing extent of shallow-water environments induced by a rising sealevel, may have been the major reason for the increase in carbonateplatform extent during the Cretaceous.

Our results are in line with the conclusions of Walker et al. (2002)who demonstrated a strong correlation between the area of low-latitudeshallow-water environments available to carbonate deposition and thearea of shallow-water carbonates during the Phanerozoic. Kiesslinget al. (2003) later acknowledged the need for additional extrinsic fac-tors to explain the Phanerozoic variations in carbonate platform sizeand distribution.

4. Discussion

4.1. Implications for future research

The first-order agreement between the spatial extent of the Aptianplatform carbonates documented in geological data and simulated ex-tent of Aptian platform carbonates based on temperature, salinity andproductivity constitutes a robust argument supporting that climaticconstraints can be used to predict the occurrence of platform carbo-nates. Relationships between the development of platform carbonatesand changes in the surrounding physical environment have a longhistory in the literature (e.g., Sohl, 1987; Schlager, 2000), but suchrelationships have rarely been tested in a quantitative manner (seeprevious attempts by Lees (1975), Couce et al. (2012) and Jones et al.(2015) on modern platform carbonates). Our study demonstrates that,in spite of the large uncertainties, this approach can be applied in deeptime based on simulations conducted with general circulation models.Beyond its predictive potential, such a quantified approach makes itpossible to disentangle the respective influences of different mechan-isms such as climate change or paleogeographic evolution.

Our simulations have additional implications for the environmentalconditions that may have been most favorable to the development ofrudist-dominated carbonate platforms. By testing two opposite open-ocean marine productivity models for the Aptian tropical shallow-marine factory, we demonstrated that the best model-data agreement isobtained when considering an affinity for low productivity levels. Ourmodel is not able to satisfactorily reproduce the extent of the Aptianplatform carbonates when considering an affinity for high marineproductivity levels. The simulation of carbonate platforms in regions oflow open-marine productivity is in agreement with the presence ofcorals, i.e., oligotrophic organisms, along the outer margin of theseplatforms (Gili et al., 1995; Johnson et al., 2001; Borgomano et al.,2002; Fenerci-Masse et al., 2005; Pomar and Hallock, 2008; Masse andFenerci-Masse, 2011). It is also supported by several studies that in-vestigated the relationship between nutrient supply and the style ofcarbonate production during the Cretaceous, including during the Ap-tian. Föllmi et al. (2006) described the temporal evolution of thebenthic marine communities along the northern Tethyan margin duringthe Early Cretaceous. They reported an alternation between phases ofcarbonate production dominated either by photozoan or heterozoancommunities, keeping in mind that they included the coral-rudist as-sociations in the photozoan community. Based notably on the correla-tion between the record of oceanic phosphorus accumulation (Föllmi,1995) and the observed modes of carbonate production, they identifiedchanges in trophic levels as a major factor driving the changes in thenature of benthic marine production in the Early Cretaceous. Specifi-cally, Föllmi et al. (2006) suggested that carbonate production by coral-rudist-dominated assemblages (i.e., the photozoan factory) was in-hibited by high nutrient levels, although Skelton and Gili (2011)highlighted that other mechanisms, such as carbonic acidification ofocean waters, may also have contributed to the interruption of platformdevelopment during specific time intervals. Another argument in sup-port of our low-productivity model is provided by Huck et al. (2010,2012). The authors documented major disruptions of carbonate

production by the rudist-(coral)-nerinid platform community during theLower Aptian, characterized by the pervasive growth of micro-encrusters (i.e., Lithocodium/Bacinella). The analysis of the faunal as-sociations combined with a review of the ecological affinities of thisgreen algae led Huck et al. (2010, 2012) to postulate that Lithocodiumprobably constituted an opportunistic taxon in competition with coralsfor the same habitat, which episodically flourished during periods ofhigh nutrient influx at the expense of the coral-rudist assemblages. Suchinterpretation of the geological record is, thus, in line with our mod-eling results suggesting the preferential development of the rudist-dominated platforms in low open-ocean productivity settings.

Although our results provide interesting insights on the trophic af-finities of Cretaceous carbonate platforms, we emphasize that our re-sults cannot be used to infer the trophic affinity of the taxa found ininner platform settings, such as rudists. Indeed, our global generalcirculation model gives access only to open-ocean productivity levels;trophic conditions may have been significantly different in moreproximal settings, due to combined effects of the potential disconnec-tion from the open ocean and the influence of the nearby continent.

4.2. Limitations

As in any deep-time study, our approach has some obvious limita-tions. Some of these limitations may serve as a basis for future work.One of the major remaining uncertainties is the input of terrigenousmaterial from adjacent land areas. The latter is known to regionallyinhibit carbonate sedimentation (e.g., Wilson, 1975) and its impact mayhave been enhanced under (super)greenhouse conditions (see Philip(2003) for a discussion of the impact of the input of terrigenous clasticson carbonate sedimentation in the Peri-Tethyan areas during the Cre-taceous). Constructing a map of the sources of terrigenous material ischallenging. Detailed knowledge of both the spatially-varying rate oferosion and the location of the main river deltas would be required,which necessitates major assumptions regarding the fine-scale topo-graphy and the nature of the land surface (i.e., vegetation cover andlithology) and geodynamic settings. Therefore no terrigenous mask hasbeen used in our simulations and our approach only attempts to identifyregions where oceanographic conditions may have been favorable tothe development of platform carbonates.

Regarding the climate model, there are uncertainties in theboundary conditions, and the biogeochemical component of our cou-pled general circulation model is relatively simple. However, the pre-sent work, as well as previous studies (Pohl et al., 2017a, 2017b),provide interesting insights into global patterns of primary productivityin past oceans, which cannot be otherwise accessed through the geo-logical record (e.g., Servais et al., 2016). In addition, these patterns ofproductivity are relatively robust since they are tightly linked to thesimulated wind belts, which are a robust model output (Pohl et al.,2017a). Still, primary productivity is computed implicitly in the modelas a simple function of phosphate concentration and photosyntheticallyactive radiation (Section S1.2). The ecology of the producers is notrepresented, so that the setup does not discriminate between phyto-plankton types, the requirements of which may vary in terms of natureand availability in nutrients. This simplification may lead to an un-derestimation of the spatial heterogeneity in simulated NPP. The modeldoes not take into account nutrient input through continental runoff ordust deposition, which results in an underestimation of simulated NPPin coastal areas, especially at river mouths. We highlight as a futureresearch target, the use of an up-to-date ecological model validated bythe present-day and accounting for riverine input, such as the versatileDarwin model (e.g. Barton et al., 2010).

It should also be noted that the use of a global ocean model raises aspatial scale problem. The accurate simulation of the complex ocea-nographic conditions and geometries that typify shallow-marineCretaceous platforms requires a high spatial resolution, both horizon-tally and vertically. The great computational cost of such high-

A. Pohl et al. Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

229

resolution models, combined with the necessity to run the simulationsfor at least 2000 years in order to ensure deep-ocean equilibrium,precludes the application of such experimental setups to deep-timestudies. Provided that the boundary conditions and notably bathymetryare sufficiently well constrained, using global simulations to provideboundary conditions for a more detailed regional model may enabledetailed studies of specific ocean basins in the future. Investigating fine-scale patterns of platform carbonate development may, however, re-quire including environmental parameters that are not accounted for inthe present attempt, such as local hydrodynamics or light limitation atdepth. The high spatial and temporal resolution needed for such studiesmay require significant rebuilding of the ecological niche model.

Another caveat is the partially circular reasoning necessarily asso-ciated with use of paleobathymetry obtained from paleogeographicmodels to simulate the spatial occurrences of platform carbonates.Indeed, the location and surface area of the shallow-water environ-ments are key boundary conditions for our niche model. They definethe area available to carbonate deposition, i.e., the bathymetry mask.The problem is that the reconstruction of these environments is partlybased on bibliographic compilation of geological lithofacies data. Inother words, the paleogeographic reconstruction already contains someof the information that we aim to extract from our models. Of course,the paleogeographic reconstruction also includes shallow-water en-vironments that do not correspond to carbonate platforms and shallow-water areas reconstructed based solely on the interpretation of tectonichistory. We therefore emphasize that the paleogeography alone cannotreflect the extent of the platform carbonates. Climatic constraints de-rived from our general circulation models are necessary to discriminatebetween shallow-marine environments favorable or unfavorable toplatform carbonate development. This is best exemplified by the fre-quency that platform carbonates are found in the model in shallow-marine environments, worldwide (30 to 39%). Paleogeographic re-constructions are also necessarily incomplete and miss some of theshallow-water shelves. In these cases, our approach is of particular in-terest in identifying the regions where oceanographic conditions mayhave been favorable to the development of platform carbonates, even ifno sedimentary record is documented. Provided that it is used withcaution, this predictive approach has the potential to complement ourknowledge of the geological record.

5. Conclusions

We have presented a model that simulates the occurrence ofCretaceous platform carbonates based on environmental conditionsderived from general circulation model simulations. An ecologicalniche model was developed that reproduces the Cretaceous tropicalshallow-marine carbonate factory. This model reproduces the spatialpatterns of Aptian (~120Ma) carbonate platform development whenan affinity for low productivity levels is considered. The alternativemodel featuring an affinity for high marine productivity levels does notagree with the occurrences reported in the geological record. Additionalsimulations conducted on continental configurations representative ofother Cretaceous time intervals (i.e., 150, 110 and 80Ma), with allother parameters being equal, reproduce at first order the long-termincrease in the extent of the carbonate platforms documentedthroughout the Cretaceous. Analysis of the underlying mechanismsreveals that the main drivers of this trend are the increasing extent ofthe shallow-water environments in a climatic zone favorable to reefaldevelopment (45°N–45°S), and the long-term rise in sea level during theCretaceous. Our simulations therefore suggest that the increase in theextent of carbonate platforms during the Cretaceous may have beenprimarily driven by the concomitant paleogeographic changes.

Acknowledgments

Thanks are expressed to Total E&P for funding a significant part of

the project and for granting permission to publish. Contribution to datainterpretation is acknowledged to S. Raillard, A. Virgone, J. Kenter andJ.-N. Ferry, TOTAL CSTJF, Pau. A.P., M.L., J.M., Y.D. and J.B. thank the‘SIGEO’ GIS service at CEREGE for providing access to the ArcGISsoftware, and the CEA/CCRT for providing access to the HPC resourcesof TGCC under the allocation 2014-012212 made by GENCI. DavidFerreira (U. Reading) is thanked for providing essential support in in-cluding the pCO2 into the MITgcm. The authors thank Alexis Godet (UTSan Antonio) and an anonymous reviewer for their constructive re-views, and Thomas Algeo (U. Cincinnati) for editorial handling.

Data availability

Data that support the results of this study are available on request tothe authors.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.palaeo.2018.10.017.

References

Adcroft, A., Campin, J.M., Hill, C., 2004. Implementation of an atmosphere-ocean generalcirculation model on the expanded spherical cube. Mon. Weather Rev. 132,2845–2863. https://doi.org/10.1175/MWR2823.1.

Barton, A.D., Dutkiewicz, S., Flierl, G., Bragg, J., Follows, M.J., 2010. Patterns of diversityin marine phytoplankton. Science 327, 1509–1511. https://doi.org/10.1126/science.1184961.

Bernard, S., Daval, D., Ackerer, P., Pont, S., Meibom, A., 2017. Burial-induced oxygen-isotope re-equilibration of fossil foraminifera explains ocean paleotemperatureparadoxes. Nat. Commun. 8, 1134. https://doi.org/10.1038/s41467-017-01225-9.

Bice, K.L., Birgel, D., Meyers, P.A., Dahl, K.A., Hinrichs, K.-U., Norris, R.D., 2006. Amultiple proxy and model study of Cretaceous upper ocean temperatures and at-mospheric CO2 concentrations. Paleoceanography 21, 1–17. https://doi.org/10.1016/S0016-7037(99)00091-5.

Borgomano, J., Masse, J.-P., Maskiry, Al.S., 2002. The lower Aptian Shuaiba carbonateoutcrops in Jebel Akhdar, northern Oman: impact on static modeling for Shuaibapetrol eum reservoirs. AAPG Bull. 86, 1513–1529. https://doi.org/10.1306/61EEDCE2-173E-11D7-8645000102C1865D.

Bottini, C., Erba, E., Tiraboschi, D., Jenkyns, H.C., Schouten, S., Sinninghe Damsté, J.S.,2015. Climate variability and ocean fertility during the Aptian Stage. Clim. Past 11,383–402. https://doi.org/10.5194/cp-11-383-2015.

Brunetti, M., Vérard, C., Baumgartner, P.O., 2015. Modeling the Middle Jurassic oceancirculation. J. Palaeogeogr. 4, 371–383. https://doi.org/10.1016/j.jop.2015.09.001.

Burchette, T.P., 2012. Carbonate rocks and petroleum reservoirs: a geological perspectivefrom the industry. Geol. Soc. Lond., Spec. Publ. 370, 17–37. https://doi.org/10.1144/SP370.14.

Couce, E., Ridgwell, A., Hendy, E.J., 2012. Environmental controls on the global dis-tribution of shallow-water coral reefs. J. Biogeogr. 39, 1508–1523. https://doi.org/10.1080/07055900.2001.9649686.

Donnadieu, Y., Dromart, G., Goddéris, Y., Pucéat, E., Brigaud, B., Dera, G., Dumas, C.,Olivier, N., 2011. A mechanism for brief glacial episodes in the Mesozoic greenhouse.Paleoceanography 26, PA3212. https://doi.org/10.1029/2010PA002100.

Donnadieu, Y., Pucéat, E., Moiroud, M., Guillocheau, F., Deconinck, J.-F., 2016. A better-ventilated ocean triggered by Late Cretaceous changes in continental configuration.Nat. Commun. 7, 10316. https://doi.org/10.1038/ncomms10316.

Dromart, G., Garcia, J.-P., Gaumet, F., Picard, S., Rousseau, M., Atrops, F., Lécuyer, C.,Sheppard, S.M.F., 2003. Perturbation of the carbon cycle at the Middle/Late Jurassictransition: Geological and geochemical evidence. Am. J. Sci. 303, 667–707. https://doi.org/10.2475/ajs.303.8.667.

Dutkiewicz, S., Follows, M.J., Parekh, P., 2005. Interactions of the iron and phosphoruscycles: a three-dimensional model study. Glob. Biogeochem. Cycles 19, GB1021.https://doi.org/10.1029/2004GB002342.

Fenerci-Masse, M., Masse, J.P., Pernarcic, E., 2005. Quantitative stratigraphy of rudistlimestones and its bearing on spatial organisation of rudist communities: the LateBarremian, Urgonian, sequences of Provence (S.E. France). Palaeogeogr.Palaeoclimatol. Palaeoecol. 215, 265–284. https://doi.org/10.1016/j.palaeo.2004.09.008.

Föllmi, K.B., 1995. 160m.y. record of marine sedimentary phosphorus burial: coupling ofclimate and continental weathering under greenhouse and icehouse conditions.Geology 23, 859–862. https://doi.org/10.1130/0091-7613(1995)023<0859:MYROMS>2.3.CO;2.

Föllmi, K.B., Godet, A., Bodin, S., Linder, P., 2006. Interactions between environmentalchange and shallow water carbonate buildup along the northern Tethyan margin andtheir impact on the Early Cretaceous carbon isotope record. Paleoceanography 21,PA4211. https://doi.org/10.1029/2006PA001313.

Forget, G., Campin, J.M., Heimbach, P., Hill, C.N., Ponte, R.M., Wunsch, C., 2015. ECCOversion 4: an integrated framework for non-linear inverse modeling and global ocean

A. Pohl et al. Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

230

state estimation. Geosci. Model Dev. 8, 3071–3104. https://doi.org/10.5194/gmd-8-3071-2015.

Gent, P.R., McWilliams, J.C., 1990. Isopycnal mixing in ocean circulation models. J. Phys.Oceanogr. 20, 150–155. https://doi.org/10.1175/1520-0485(1990)020<0150:IMIOCM>2.0.CO;2.

Gili, E., Goetz, S., 2018. Treatise Online no. 103: Part N, Volume 2, Chapter 26B:Paleoecology of Rudists.

Gili, E., Skelton, P.W., Vicens, E., Obrador, A., 1995. Corals to rudists – an en-vironmentally induced assemblage succession. Palaeogeogr. Palaeoclimatol.Palaeoecol. 119, 127–136. https://doi.org/10.1016/0031-0182(95)00064-X.

Gough, D.O., 1981. Solar interior structure and luminosity variations. Sol. Phys. 74,21–34. https://doi.org/10.1007/BF00151270.

Granier, B., Dias-Brito, D., 2015. End of a modern geological myth: there are no rudists inBrazil! Paleobiogeographic implications. Carnets Géol. 15, 123–136.

Granier, B., Berthou, P.-Y., Poignant, A.-F., 1991. Constructions bio-sédimentaireslaminées, Lithothamnium et Parachaetetes de la Formation Riachuelo (Albien) dubassin de Sergipe (Nord-Est du Brésil). 10. Geociências, São Paulo, pp. 169–181.

Hoegh-Guldberg, O., 1999. Climate change, coral bleaching and the future of the world'scoral reefs. Mar. Freshw. Res. 50, 839–866. https://doi.org/10.1071/MF99078.

Huber, M., Caballero, R., 2011. The early Eocene equable climate problem revisited. Clim.Past 7, 603–633. https://doi.org/10.5194/cp-7-603-2011.

Huck, S., Rameil, N., Korbar, T., Heimhofer, U., Wieczorek, T.D., Immenhauser, A., 2010.Latitudinally different responses of Tethyan shoal-water carbonate systems to theEarly Aptian oceanic anoxic event (OAE 1a). Sedimentology 57, 1585–1614. https://doi.org/10.1111/j.1365-3091.2010.01157.x.

Huck, S., Heimhofer, U., Immenhauser, A., 2012. Early Aptian algal bloom in a neriticproto-North Atlantic setting: Harbinger of global change related to OAE 1a? Geol.Soc. Am. Bull. 124, 1810–1825. https://doi.org/10.1130/B30587.1.

Immenhauser, A., Nägler, T.F., Steuber, T., Hippler, D., 2005. A critical assessment ofmollusk 18O/16O, Mg/Ca, and 44Ca/40Ca ratios as proxies for Cretaceous seawatertemperature seasonality. Palaeogeogr. Palaeoclimatol. Palaeoecol. 215, 221–237.https://doi.org/10.1016/j.palaeo.2004.09.005.

James, N.P., 1997. The cool-water carbonate depositional realm. In: James, N.P., Clarke,J.A.D. (Eds.), Cool-Water Carbonates. Society of Economic Paleontologists andMineralogists, Special Publication 56. pp. 1–20. https://doi.org/10.2110/pec.97.56.0001.

Jiang, L.-Q., Feely, R.A., Carter, B.R., Greeley, D.J., Gledhill, D.K., Arzayus, K.M., 2015.Climatological distribution of aragonite saturation state in the global oceans. Glob.Biogeochem. Cycles 29, 1656–1673. https://doi.org/10.1002/2015GB005198.

Johnson, C.C., Sanders, D., Kauffman, E.G., Hay, W.W., 2001. Patterns and processesinfluencing Upper Cretaceous reefs. In: Kiessling, W., Flügel, E., Golonka, J. (Eds.),Phanerozoic Reef Patterns. SEPM Special Publication, pp. 549–585.

Jones, N.S., Ridgwell, A., Hendy, E.J., 2015. Evaluation of coral reef carbonate produc-tion models at a global scale. Biogeosciences 12, 1339–1356. https://doi.org/10.5194/bg-12-1339-2015.

Kiessling, W., 2001. Paleoclimatic significance of Phanerozoic reefs. Geology 29,751–754. https://doi.org/10.1130/0091-7613(2001)029<0751:PSOPR>2.0.CO;2.

Kiessling, W., 2005. Habitat effects and sampling bias on Phanerozoic reef distribution.Facies 51, 24–32. https://doi.org/10.1017/S009483730000542X.

Kiessling, W., Flügel, E., Golonka, J., 1999. Paleoreef maps: evaluation of a comprehen-sive database on Phanerozoic reefs. AAPG Bull. 83, 1552–1587 (updated).

Kiessling, W., Flügel, E., Golonka, J., 2003. Patterns of Phanerozoic carbonate platformsedimentation. Lethaia 36, 195–225. https://doi.org/10.1111/j.1365-3091.1986.tb00533.x.

Kleypas, J.A., McManus, J.W., Meñez, A.B., 1999. Environmental limits to coral reefdevelopment: where do we draw the line? Am. Zool. 39, 146–159. https://doi.org/10.1093/icb/39.1.146.

Ladant, J.-B., Donnadieu, Y., 2016. Palaeogeographic regulation of glacial events duringthe Cretaceous supergreenhouse. Nat. Commun. 7, 12771. https://doi.org/10.1038/ncomms12771.

Large, W.G., McWilliams, J.C., Doney, S.C., 1994. Oceanic vertical mixing: a review and amodel with a nonlocal boundary layer parameterization. Rev. Geophys. 32, 363–403.https://doi.org/10.1029/94RG01872.

Lauchstedt, A., Pandolfi, J.M., Kiessling, W., 2017. Towards a new paleotemperature-proxy from reef coral occurrences. Sci. Rep. 7, 10461. https://doi.org/10.1038/s41598-017-10961-3.

Lees, A., 1975. Possible influence of salinity and temperature on modern shelf carbonatesedimentation. Mar. Geol. 19, 159–198. https://doi.org/10.1016/0025-3227(75)90067-5.

Malinverno, A., Erba, E., Herbert, T.D., 2010. Orbital tuning as an inverse problem:chronology of the early Aptian oceanic anoxic event 1a (Selli Level) in the CismonAPTICORE. Paleoceanography 25, PA2203. https://doi.org/10.1029/2009PA001769.

Marshall, J., Adcroft, A., Hill, C., Perelman, L., Heisey, C., 1997. A finite-volume, in-compressible Navier Stokes model for studies of the ocean on parallel computers. J.Geophys. Res. 102, 5753–5766. https://doi.org/10.1029/96JC02775.

Mason, E., Edmonds, M., Turchyn, A.V., 2017. Remobilization of crustal carbon maydominate volcanic arc emissions. Science 357, 290–294. https://doi.org/10.1126/science.aan5049.

Masse, J.-P., Fenerci-Masse, M., 2011. Drowning discontinuities and stratigraphic corre-lation in platform carbonates. The late Barremianeearly Aptian record of southeastFrance. Cretac. Res. 32, 659–684. https://doi.org/10.1016/j.cretres.2011.04.003.

McAnena, A., Flögel, S., Hofmann, P., Herrle, J.O., Griesand, A., Pross, J., Talbot, H.M.,Rethemeyer, J., Wallmann, K., Wagner, T., 2013. Atlantic cooling associated with amarine bioticcrisis during the mid-Cretaceous period. Nat. Geosci. 6, 558–561.https://doi.org/10.1038/ngeo1850.

Mehay, S., Keller, C.E., Bernasconi, S.M., Weissert, H., Erba, E., Bottini, C., Hochuli, P.A.,2009. A volcanic CO2 pulse triggered the Cretaceous Oceanic Anoxic Event 1a and abiocalcification crisis. Geology 37, 819–822. https://doi.org/10.1130/G30100A.1.

Michel, J., Borgomano, J., Reijmer, J.J.G., 2018. Heterozoan carbonates: when, whereand why? A synthesis on parameters controlling carbonate production and occur-rences. Earth Sci. Rev. 182, 50–67. https://doi.org/10.1016/j.earscirev.2018.05.003.

Müller, R.D., Sdrolias, M., Gaina, C., Steinberger, B., Heine, C., 2008. Long-term sea-levelfluctuations driven by ocean basin dynamics. Science 319, 1357–1362. https://doi.org/10.1126/science.1151540.

Naafs, B.D.A., Pancost, R.D., 2016. Sea-surface temperature evolution across AptianOceanic Anoxic Event 1a. Geology 44, 959–962. https://doi.org/10.1130/G38575.1.

Naafs, B.D.A., Castro, J.M., De Gea, G.A., Quijano, M.L., Schmidt, D.N., Pancost, R.D.,2016. Gradual and sustained carbon dioxide release during Aptian Oceanic AnoxicEvent 1a. Nat. Geosci. 9, 135–139. https://doi.org/10.1016/j.epsl.2011.02.031.

O'Brien, C.L., Robinson, S.A., Pancost, R.D., Damsté, J.S.S., Schouten, S., Lunt, D.J.,Alsenz, H., Bornemann, A., Bottini, C., Brassell, S.C., Farnsworth, A., Forster, A.,Huber, B.T., Inglis, G.N., Jenkyns, H.C., Linnert, C., Littler, K., Markwick, P.,McAnena, A., Mutterlose, J., Naafs, B.D.A., Püttmann, W., Sluijs, A., van Helmond,N.A.G.M., Vellekoop, J., Wagner, T., Wrobel, N.E., 2017. Cretaceous sea-surfacetemperature evolution: constraints from TEX86 and planktonic foraminiferal oxygenisotopes. Earth Sci. Rev. 172, 224–247. https://doi.org/10.1016/j.earscirev.2017.07.012.

Philip, J., 2003. Peri-Tethyan neritic carbonate areas: distribution through time anddriving factors. Palaeogeogr. Palaeoclimatol. Palaeoecol. 196, 19–37. https://doi.org/10.1016/S0031-0182(03)00311-0.

Pohl, A., Donnadieu, Y., Le Hir, G., Ferreira, D., 2017a. The climatic significance of LateOrdovician-early Silurian black shales. Paleoceanography 32, 397–423. https://doi.org/10.1002/2016PA003064.

Pohl, A., Harper, D.A.T., Donnadieu, Y., Le Hir, G., Nardin, E., Servais, T., 2017b. Possiblepatterns of marine primary productivity during the Great OrdovicianBiodiversification Event. Lethaia 132, 525–531. https://doi.org/10.1175/1520-0426(2000)017<0525:ARTLSI>2.0.CO;2.

Pomar, L., Hallock, P., 2008. Carbonate factories: a conundrum in sedimentary geology.Earth Sci. Rev. 87, 134–169. https://doi.org/10.1016/j.earscirev.2007.12.002.

Pucéat, E., Lécuyer, C., Donnadieu, Y., Naveau, P., Cappetta, H., Ramstein, G., Huber,B.T., Kriwet, J., 2007. Fish tooth δ18O revising Late Cretaceous meridional upperocean water temperature gradients. Geology 35, 107–110. https://doi.org/10.1130/G23103A.1.

Redi, M.H., 1982. Oceanic isopycnal mixing by coordinate rotation. J. Phys. Oceanogr.12, 1154–1158. https://doi.org/10.1175/1520-0485(1982)012<1154:OIMBCR>2.0.CO;2.

Rose, B.E.J., Ferreira, D., 2013. Ocean heat transport and water vapor greenhouse in awarm equable climate: a new look at the low gradient paradox. J. Clim. 26,2117–2136. https://doi.org/10.1175/JCLI-D-11-00547.1.

Ruttenberg, K.C., 1993. Reassessment of the oceanic residence time of phosphorus. Chem.Geol. 107, 405–409. https://doi.org/10.1016/0009-2541(93)90220-D.

Saaty, R.W., 1987. The analytical hierarchy process – what it is and how it is used. Math.Model. 9, 161–176. https://doi.org/10.1016/0270-0255(87)90473-8.

Schlager, W., 2000. Sedimentation rates and growth potential of tropical, cool-water andmud-mound carbonate systems. In: Insalaco, E., Skelton, P.W., Palmer, T.J. (Eds.),Carbonate Platform Systems: Components and Interactions. TheMicropalaeontological Society, Special Publications 178. The Geological Society ofLondon, pp. 217–227. https://doi.org/10.1144/GSL.SP.2000.178.01.14.

Schlager, W., 2005. Carbonate sedimentology and sequence stratigraphy. In: Concepts inSedimentology and Paleontology. SEPM, Special Publications 8, Tulsa, Oklahoma.https://doi.org/10.2110/csp.05.08. (200 pp.).

Scotese, C.R., 2014a. Late Cretaceous Atlas of Paleogeograhic Maps, PALEOMAP Atlas forArcGIS, volume 2, The Cretaceous, Maps 16–22, Mollweide project, PALEOMAPProject, Evanston, IL. https://doi.org/10.13140/2.1.4691.3284.

Scotese, C.R., 2014b. Early Cretaceous Atlas of Paleogeograhic Maps, PALEOMAP Atlasfor ArcGIS, volume 2, The Cretaceous, Maps 23–31, Mollweide project, PALEOMAPProject, Evanston, IL. https://doi.org/10.13140/2.1.4099.4560.

Scott, R.W., 1995. Global environmental controls on Cretaceous reefal ecosystems.Palaeogeogr. Palaeoclimatol. Palaeoecol. 119, 187–199. https://doi.org/10.1016/0031-0182(94)00068-9.

Servais, T., Perrier, V., Danelian, T., Klug, C., Martin, R., Munnecke, A., Nowak, H.,Nützel, A., Vandenbroucke, T.R.A., Williams, M., Rasmussen, C.M.O., 2016. Theonset of the “Ordovician Plankton Revolution” in the late Cambrian. Palaeogeogr.Palaeoclimatol. Palaeoecol. 458, 12–28. https://doi.org/10.1016/j.palaeo.2015.11.003.

Skelton, P.W., Gili, E., 2011. Rudists and carbonate platforms in the Aptian: a case studyon biotic interactions with ocean chemistry and climate. Sedimentology 59, 81–117.https://doi.org/10.1016/S0031-0182(03)00450-4.

Sohl, N.F., 1987. Cretaceous Gastropods: contrasts between Tethys and the TemperateProvinces. J. Paleontol. 61, 1085–1111.

Steuber, T., 1999. Isotopic and chemical intra-shell variations in low-Mg calcite of rudistbivalves (Mollusca-Hippuritacea): disequilibrium fractionations and late Cretaceousseasonality. Int. J. Earth Sci. 88, 551–570. https://doi.org/10.1007/s005310050284.

Steuber, T., 2000. Skeletal growth rates of Upper Cretaceous rudist bivalves: implicationsfor carbonate production and organism-environment feedbacks. In: Insalaco, E.,Skelton, P.W., Palmer, T.J. (Eds.), Carbonate Platform Systems: Components andInteractions. Geological Society, London, Special Publications 178. pp. 21–32.https://doi.org/10.1144/GSL.SP.2000.178.01.03.

Steuber, T., 2002. Plate tectonic control on the evolution of Cretaceous platform-carbo-nate production. Geology 30, 259–262. https://doi.org/10.1130/0091-7613(2002)030<0259:PTCOTE>2.0.CO;2.

A. Pohl et al. Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

231

Steuber, T., Rauch, M., Masse, J.-P., Graaf, J., Malkoč, M., 2005. Low-latitude seasonalityof Cretaceous temperatures in warm and cold episodes. Nature 437, 1341–1344.https://doi.org/10.1306/052803740007.

Walker, L.J., Wilkinson, B.H., Ivany, L.C., 2002. Continental drift and Phanerozoic car-bonate accumulation in shallow-shelf and deep-marine settings. J. Geol. 110, 75–87.https://doi.org/10.1086/324318.

Wallmann, K., 2003. Feedbacks between oceanic redox states and marine productivity: a

model perspective focused on benthic phosphorus cycling. Glob. Biogeochem. Cycles17, 1084. https://doi.org/10.1029/2002GB001968.

Wilson, J.L., 1975. Principles of carbonate sedimentation. In: Wilson, J.L. (Ed.),Carbonate Facies in Geologic History. Springer-Verlag, New York Heidelberg Berlin,pp. 1–19. https://doi.org/10.1007/978-1-4612-6383-8.

Zadeh, L.A., 1965. Fuzzy sets. Inf. Control. 8, 338–353. https://doi.org/10.1016/S0019-9958(65)90241-X.

A. Pohl et al. Palaeogeography, Palaeoclimatology, Palaeoecology 514 (2019) 222–232

232