New palaeoclimatic constraints from paleosols on the Middle-Late Jurassic landscape, Western Colorado, U.S.A

ABSTRACT The Wanakah Formation and Tidwell Member of the Morrison Formation record intervals of paedogenesis in the Paradox Basin and Central Colorado trough, western Colorado during the Middle-Late Jurassic. Detailed field description of paleosols and paedogenic carbonates at different stratigraphic horizons from three localities document four main pedotypes: vertisols, gleysols, oxisols, and protosols. Generally, gleysols reflect reducing conditions, protosols and oxisols reflect oxidizing conditions, and vertisols reflect fluctuations in oxidizing/reducing conditions (seasonality). Major elemental geochemical ratios in samples from these paleosols suggest variable redox conditions and a sub-humid to humid paleoclimate with seasonal precipitations during paleosol development. Estimated mean annual temperature based on elemental analyses of paleosol B in paleosols from Ribbon Trail and Escalante Canyon in western Colorado suggest mesic – thermic paleoclimate. Mean annual precipitation indicates sub-humid to humid regional palaeoclimatic conditions marked by seasonal precipitation. Clay mineralogy of these paleosols, determined by X-ray diffraction, shows the abundance of illite most likely formed from smectite alteration due to episodic wetting and drying. δ13C and δ18O isotope analyses of carbonate veins in these paleosols suggest that they formed recently from meteoric water and therefore are not considered for palaeoclimatic reconstructions. Mean annual temperature and mean annual precipitation estimates in the Wanakah Formation and Tidwell Member of the Morrison Formation provide new information on the local and regional palaeoclimatic and paleoenvironmental conditions during the Middle Jurassic in western Colorado as well as data for comparison with the Upper Morrison paleoclimate from northern and southern localities in the western United States.


Introduction
The Jurassic sedimentary units in the western United States have received considerable attention due to their extensive fossil records which are of both palaeontological and geologic interest (e.g.Dodson et al. 1980;Jensen and Padian 1989).Palaeontological, palaeoclimatic, and paleoenvironmental studies on the Morrison Formation have placed important constraints on the Jurassic palaeogeography of western United States (e.g.Hasiotis 2004;Parrish et al. 2004;Bernier and Chan 2006).The existence and subsequent extinction of dinosaurs, as well as the rich ecosystem, diverse fauna and flora populations, during the deposition of the Morrison Formation constrain the Late Jurassic palaeoclimatic and paleoenvironmental conditions (Demko et al. 2004;Rees et al. 2004;Turner et al. 2004).
Recently, Myers et al. (2014) used multi-geochemical proxies in paleosols from the upper Morrison Formation across the western United States to interpret the Late Jurassic paleoclimate as changing from semiarid, during the Kimmeridgian (~157.3 to 152.1 Ma), to subhumid, during the Tithonian (~152.1 to 145.0 Ma), with seasonal precipitation.Spatial variations in climatic conditions are also reported within the Morrison depositional basin and span from arid in the south to sub-humid in the north (Demko et al. 2004;Myers et al. 2014).However, the palaeoclimatic record in paleosols that formed before and at the onset of the Morrison Formation deposition, i.e. during the latest Middle to earliest Upper Jurassic, a period of rapid transition in the regional climate and depositional environment in southwest Laurentia as recorded in the Callovian (~166.1 to 163.5 Ma; Wanakah Formation) -Oxfordian (~163.5 to 157.3 Ma; Tidwell Member of the Morrison Formation) interval has not been fully documented.Thick packages of these paleosols are preserved in the Paradox Basin and the Central Colorado trough (CCT) localities in western Colorado, which covers ~19,000 km 2 .This study explores

Geologic setting and regional palaeogeography
Previous studies (e.g.Peterson 1994) showed that paleosols of the Jurassic Wanakah and Morrison Formations formed in the Paradox Basin in largely stable terrestrial (lacustrine) and marginal marine (tidal-flat) environments, respectively.The Middle Jurassic palaeogeography in the Cordilleran foreland, within the Western Interior of Laurentia, was a vastly arid and low topographic plain except for Central Colorado, where the NW -SE trending Ancestral Front Range (AFR) intracratonic uplift was prominent and extended southeastwards (Blakey 1994;Blakey and Ranney 2008).Marine incursions expanded towards the continental interior, while exotic terranes were accreted on Laurentia's active western margin, setting the stage for Cordilleran magmatism further west.
The early Late Jurassic (~160 Ma) marked the second episode of marine transgression into the continent following the continuous subduction and volcanism on the western continental margin (Lawton 1994;DeCelles 2004).The Cordillera magmatic arcs migrated eastwards (Smith et al. 1993) due to crustal shortening while significant uplifts in the southern end of the North American craton formed the Mogollon Highland (Bilodeau 1986) causing the regression of the seaway.Sediments delivered into western intracratonic basins during the Morrison time were mainly sourced from the west (Cordilleran arc) and the south (Mogollon highland in southern Arizona and New Mexico) (Peterson 1994).

Basis of paleosols as palaeoclimatic and paleoenvironmental archive
Paleosols record the palaeoclimatic and paleoenvironmental conditions at the time of their development because they form by weathering of parent materials on stable, non-eroding ancient landscapes that were exposed at the earth's surface and preserved over relatively long geologic timescales (Kraus 1999;Sheldon and Tabor 2009).As a result, paleosols have been used to reconstruct paleoenvironmental and palaeoclimatic conditions (e.g.Retallack 1994b; Prochnow et al. 2006;Sheldon and Tabor 2009;Torres and Gaines 2013).Paleosols generally lack bedding structures/depositional features associated with typical sedimentary rocks because they form from the in-situ alteration (physical, biological, and chemical weathering) of parent materials.However, paleosols host paedogenic features that are distinctive and associated with their development mode, even after undergoing burial and compaction.
Paleosols are commonly classified based on (a) horizonation -which relates to continuity, thickness, and colour of a paleosol profile, (b) pedofacies -which relates to the texture (loam, sandy loam, clay loam, clayey), sediment size ranging from very fine, fine, medium, coarse, to very coarse, and peds ranging from massive, platy, wedgy, to angular blocky, (c) reaction to acid which could be no reaction, weak, moderate, or strong, (d) presences of paedogenic features, e.g.slickensides, root trace, organic matter and, (d) redoximorphic features, e.g.presence of iron oxide pigments, calcite nodules/concretions (Retallack 1988;Cerling and Quade 1993;Mack et al. 1993;Schad 2016).Geochemical proxies and transfer functions based on major elements in paleosols have been developed and applied in previous studies to determine paedogenic processes and interpret palaeoclimatic and paleoenvironmental conditions (Nesbitt and Young 1982;Maynard 1992;Fedo et al. 1995;Sheldon et al. 2002;Retallack 2005;Sheldon 2006a; Sheldon and Tabor 2009;Rosenau et al. 2013;Myers et al. 2014;Tabor and Myers 2015;Driese et al. 2016).The principles behind some of these proxies are briefly described below.

Geochemical proxies in paleosols
A range of paleoenvironmental and palaeoclimatic conditions, e.g. chemical weathering, paleotemperature, paleoprecipitation, atmospheric CO 2 estimates, and provenance are quantified in ancient soils using thermodynamic and geochemical models that have been successfully applied to study paedogenesis in modern soils (Nesbitt and Young 1982;Retallack 1994aRetallack , 1997;;Sheldon et al. 2002;Prochnow et al. 2006;Sheldon and Tabor 2009;Driese et al. 2016).These models assume that the overall bulk chemical composition of paleosols is preserved during diagenesis (Mora et al. 1998) and metamorphism (Barrientos and Selverstone 1987), although the cation exchange capacity and base saturation may be altered after burial (Retallack 1991).The extent of chemical alteration in a soil profile can be estimated from the ratio of soluble cations (e.g.Ca, Fe, Mg, Na, and K) versus an immobile element (e.g.Al, Ti, Zr) when the concentrations are normalized to the lowermost horizon along a paleosol profile.
For example, in a stable sedimentary basin undergoing net aggradation of paleosols, the mobilization and removal of soluble cations (e.g.Ca, Na, Mg, and K) from the top of the soil profile, is an indication of the weathering intensity.Soil weathering intensity is influenced by the duration of subaerial exposure, mean annual temperature (MAT), and mean annual precipitation (MAP) at the time of formation (Torres and Gaines 2013).When soluble cations such as Ca, Na, Mg, and K are removed from the top of the soil profile due to high weathering intensity, silicate minerals (e.g.kaolinite) that are rich in Al and Si form while some heavy and insoluble minerals, such as zircon, become relatively enriched (Torres and Gaines 2013).With reduced weathering intensity, carbonates, smectites, and other secondary alkaline earth minerals form in response to the hydrolysis of soluble cations in soils (Sheldon and Tabor 2009).
A palaeoclimatic trend such as MAP is estimated from the geochemical indices of alteration (Sheldon et al. 2002).MAP estimates are used to categorize paleoclimate based on the amount of yearly precipitation: arid (50-205 mm), semiarid (250-500 mm), subhumid (500-1000 mm), and humid (1000-2000 mm) (Bull 1991).The most used proxy to assess the intensity of chemical weathering is the Chemical Index of Alteration (CIA) (Nesbitt and Young 1982;Fedo et al. 1995).CIA is calculated using the molecular proportions of certain major oxides according to the equation by Nesbitt and Young (1982): CIA can also be used to quantify the conversion of feldspar to clay minerals (Selvaraj et al. 2006).Typical CIA values for potassium feldspars are 50 while the average CIA value of clay minerals in shale ranges from 70 to 75 (Nesbitt and Young 1982).Maynard (1992) proposed the CIA without potassium (CIA-K) equation to eliminate the effect of metasomatism.
CaO* is the total quantity of CaO measured excluding CaO derived from calcium carbonate.Sheldon et al. (2002) derived a geochemical climofunction based on the relationship of MAP and CIA-K for North American paleosols.These authors defined R 2 = 0.72 and standard error = ±184 mm.The climofunction estimates annual precipitation in the range of 200-1600 mm/yr.However, it does not apply to calcic subsoil horizons because paedogenic carbonates can synthetically reduce the precipitation estimates by altering the CIA-K coefficient significantly.Paedogenic carbonates are typically precipitated in MAP<1000 mm environments since carbonates dissolution increases with high precipitation rates (Royer 1999).Sheldon (2006b) derived an exact quantitative relationship between MAT and Al/Si ratio (i.e.clayeyness of Bt horizon) in protosol defined as standard error is ±0.6°C and R 2 = 0.96.Paleotemperature estimates obtained using this relationship on paleosols of different geologic ages (Palaeozoic -Cenozoic) are consistent with estimates derived from plant and other biogenic materials (Sheldon 2006b and references therein).While the application of some of these geochemical proxies in paleosols constitutes a significant part of this study, our interpretations were not based solely on them.Our interpretations incorporate complementary information such as paedogenic features documented from field observations, X-ray diffraction (XRD) of both bulk sample and clay fraction, petrographic analysis of thin sections, and carbon and oxygen stable isotopes.

Field methods and sampling
Paleosol samples were collected from the Wanakah Formation and Tidwell Member of the lower Morrison Formation from three study sites: Ribbon Trail, Escalante Canyon, and Sawpit, along an NW -SE transect in western Colorado (Figure 1).Paleosols at these sites are wellpreserved and range from well-developed, to moderately-developed, to poorly-developed at Ribbon Trail, Escalante Canyon, and Sawpit, respectively.Figure 2 shows representative outcrops of these paleosols at Ribbon Trail and Escalante Canyon.The outcrops were observed to identify paedogenic features and develop stratigraphic correlations (Figure 3).Some examples of paedogenic features documented include colour mottles (using the Munsell Colour Chart; Supp.Table S1), the presence or lack of root traces, ped structure, slickenside development, and calcite concretions/nodules.Samples were collected along the stratigraphic section at horizons where an observable change in pedofacies is observed.Twenty-eight paleosol samples were collected; sixteen were from Ribbon Trail, seven from Escalante Canyon, and four from Sawpit (Figure 3).

Analytical techniques
X-ray diffraction analysis (XRD) was performed on the bulk paleosol samples (n = 4) and clay fractions (<2 μm size particle, n = 16) for mineral identification using the Rigaku Ultima IV X-ray diffractometer at Southern Illinois University Carbondale, Illinois.Thin sections used for the XRD of bulk samples were prepared by Spectrum Petrographics Inc., Vancouver, Washington, U.S.A. Claysize fractions were separated following standard methods described by Moore and Reynolds (1997).Samples were gently disaggregated to very fine grains and ovendried at 60°C for ~20 hours to reduce clumping.The powdered samples were soaked in deionized water and stirred thoroughly, with drops of 70% isopropyl alcohol solution added to aid deflocculation for ~4 hours.Suspended clay separates were extracted using a pipette, smeared on a glass slide to orient the clays, and allowed to air-dry for about 18 hours.XRD spectra were acquired for the air-dried sample and subsequently for samples heated (up to ~60°C) in an oven for ethylene glycol solvation for 5 hours.The samples were all scanned using Cu-Kα radiation with accelerating potential and tube current of 40 kV and 44 mA, over a 2theta range of 2 to 40° for clay separates and 10 to 60° for the bulk samples.Micromorphological features in the paleosols were examined on thin sections using a petrographic microscope equipped with a Leica digital camera.Images of distinctive paedogenic features found in the different paleosol horizons were taken for further analysis and interpretation.
An average of 30 g per paleosol sample was sent to ALS Geochemistry Analytical Lab in Reno, Nevada, U.S.A for whole-rock geochemical analyses.Samples were finely crushed (<2 mm) and then pulverized to <75 μm.A pulverized split of ~0.04 g per sample was used for the analyses.Major elements were analysed using both X-Ray Fluorescence (XRF) and Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES) after heating (up to 1000°C to achieve loss on ignition) and complete dissolution of the respective sample in four separate acids (i.e.HCl, HF, HNO3, and HClO4).For ICP-MS and XRF analyses, the following standards (AGV-1, BHVO-2, BCR-2, DTS-1, JB-2, JB-3, JP-1, and MRG; data from Jochum et al. 2005) were used for calibration with a precision of 0.01% and an accuracy, estimated from repeated analysis of the same sample, of<3% for all major elements.Elemental concentrations from the analyses were reported in wt.%.
Mobile cations/Ti ratios were used to assess the depletion and accumulation of elements in the Bt horizon of the respective profiles by normalizing the ratios to the lowest sampled stratigraphic horizon.The redox condition of the respective paleosol horizon was assessed using the multi-element ratio of redox- sensitive elements; Fe/Ti vs. Fe + Mn/Al.The multielement ratios were all calculated using their molar concentrations.The ratios of major element geochemical data for the twenty-eight paleosol samples are presented in Supp.Table S1 and displayed in Figure 6.The multi-element ratios are in molar concentrations, normalized to the lowest stratigraphic horizon of each unit (i.e. the Wanakah Formation and Tidwell Member) to track accumulation/depletion of the elements across multiple horizons.Some of the major element ratios that are routinely used to quantify paedogenic processes include (1) ∑base elements/Ti ratio (∑base elements = Ca+Mg+Na+K) which is used to assess leaching from the top of the soil profile and accumulation of Ti during weathering at normal pH conditions (Supp.Table S1; Figure 6 left); (2) K+Na/Al ratio is used to assess salinization from the accumulation of alkali elements as soluble salts that are not leached under normal weathering conditions (Supp.Table S1; Figure 6 centre); and (3) the Al/Si ratio is used to assess the clayeyness (i.e.hydrolysis) of paleosols since Al preferentially accumulates during the formation of clay minerals (Supp.Table S1; Figure 6

right).
Nine paedogenic calcite concretion/vein samples present at certain stratigraphic intervals within some paleosol horizons were collected at the three localities for stable isotope analysis.The carbonate samples were mostly homogenous.The δ 13 C and δ 18 O compositions in the carbonate samples were analysed at Southern Illinois University Mass Spectrometry Facility using a Thermo Scientific Delta-V Plus isotope ratio mass spectrometer (IRMS) equipped with an on-line gas preparation and introduction Gasbench II system.δ 13 C and δ 18 O values were corrected for isotopic fractionation by normalizing to base values (in ‰) relative to the respective ratios of δ 13 C and δ 18 O in the Vienna Pee Dee Belemnite (V-PDB) standard.National Standards NBS 18 and NBS19 were used for calibration with a precision of 0.01‰.Standard deviations for δ 13 C and δ 18 O were respectively 0.05 and 0.07‰.

Field identification and classification of paleosols
Field recognition, description, and classification of paleosols in this study are based on the morphology and pedofacies methods of Retallack (1988) and Mack et al. (1993).Based on the classification scheme of Mack et al. (1993), four classes of paleosols were identified namely vertisol, gleysol, oxisol, and protosol in the Wanakah Formation and Tidwell member of the Morrison Formation (Figure 3; Supp.Table S1).Vertisols are dark green paleosols characterized as having very high and homogeneous clay content (>35%); they exhibit shrink and swell features (desiccation cracks), slickensides, and are moderately leached.Gleysols are a modified redox variant of vertisols and are often referred to as gleyed vertisols.They are well-leached, mostly greenish-grey (Munsell colour: 5 G 4/1 to 5 GY 4/1; Supp.Table S1), and have patches of yellow and reddish-brown.Gleysols host paedogenic features like slickensides due to their modest clay content.Oxisols are identified based on their bright reddish colour, extreme weathering, presence of 1:1 phyllosilicates, accumulation of Fe and Al oxides, or mixtures of both, within the paleosol horizon.Protosols are weakly developed paleosol horizons that host some paedogenic carbonates, root traces, and are typically light brown to tan in colour.All four classes of paleosols are present at Ribbon Trail and Escalante Canyon, while three (excluding protosols) are present at Sawpit (Figure 3).Common paedogenic features found within these paleosol horizons at the three localities are shown in Figure 3.

Clay mineralogy
Figure 4 shows representative XRD spectra of oriented clay separates (air-dried and glycolated) from different paleosol horizons in the Wanakah Formation and Tidwell Member samples.The XRD spectra show the presence of three clay minerals (smectite, illite, and kaolinite) and minor goethite that have varying peak strength and sharpness (Figure 4).Changes in peak intensities and shift in peak positions in the low angle direction were observed in some samples (e.g. Figure 4(b-d)) after heating in ethylene glycol, suggesting the presence of randomly interstratified illite/smectite minerals.
Strong reflections at 2θ angle of 26.60-26.72° in the representative samples are interpreted as an admixture of illite and quartz minerals (Figure 4).Minor peaks occurring at two evenly spaced intervals at 2θ angle = ~8.81°and 17.74° in all the samples confirm the abundance of the illite phase.The RT-5Jw sample displays a sharp calcite peak at 2θ = 29.38°and a moderate quartz peak at 2θ = 20.81°(Figure 4(c)).Weak peaks interpreted as kaolinite phase occurred at 2θ angle = ~12.35°,24.35°, and 34.91° in the RT-6Jw sample (volcanic ash layer; Figure 4(b)), with only the primary peak (2θ angle = ~12.35°)occurring in the other samples (Figure 4(a,cd)).A weak shoulder peak at 2θ angle = ~30.97°interpreted as dolomite occurs in the EC-3Jmt (Gleysol) sample (Figure 4(d)).Minor peaks occurring at 2θ angle = ~19.70°and 19.78° in the RT-5JW (oxisol) and EC-3Jmt (gleysol) samples, respectively, are interpreted as a trace amount of chlorite (Figure 4(c,d)).Goethite (110 peak) is present at 2θ angle = ~25° in the oxisol.While it may be difficult to quantitatively assess the abundance of each clay mineral in the samples using the XRD spectra, the relative peak intensities show which of the clay minerals is most abundant and might reflect paedogenic processes during paleosol development.Ethylene glycol solvation of the samples revealed the presence of mixed-layering of illite/smectite, particularly at lowerorder reflections (Figure 4).The kaolinite phase is present in all the samples, although in minor amounts.

Micromorphology of paleosols
Thin-section images of distinctive paedogenic features in representative samples from the different paleosol horizons are presented in Figure 5.The samples show a homogenous clay matrix, consistent with results from the XRD spectra (Figure 4).Calcite veins are common in the RT-2Jmt paleosol sample from the vertisol horizon (Figure 5(a)).The RT-6Jw sample (volcanic ash) is predominantly clay with dispersed grains of quartz (Figure 5(b)).The RT-5Jw sample is representative of the oxisol horizon (Figure 5(c)).Oxisol horizons are distinctive because they contain a high concentration of iron oxy-hydroxide.Figure 5(d) shows a root trace, a common paedogenic characteristic of gleysols.Root traces filled with iron oxide are pervasive in the macroscopic sample.

Multi-elemental ratios
Figure 6 graphically represents the geochemical data of Supplementary Table S1.The Fe/Ti vs. Fe + Mn/Al ratio (Figure 7) is used to assess redox conditions since Fe and Mn are highly redox-sensitive and Ti and Al are immobile under diagenetic conditions (Sheldon and Tabor 2009;Torres and Gaines 2013).Smaller values of both ratios reflect high mobilization of soluble elements under reducing conditions (Figure 8).Iron is soluble and mobile under reducing conditions (e.g. in vertisols or gleysols) and precipitated under oxidizing conditions (e.g. in oxisols or protosols).

Stable isotopes of carbon and oxygen
The stable isotope values of oxygen and carbon from paedogenic carbonates from different paleosol horizons are shown in Supp.Table S2 and plotted in Figure 9.The carbonate samples are primarily small, homogeneous calcite concretions and veins found in the vertisol, gleysol, and oxisol horizons, except for a biogenic carbonate (microbialite) sample -from a greyish, 10-15 cm thick micritic bed with wavy laminafound in the protosol horizon at Escalante Canyon.The paedogenic carbonates from vertisols (δ 13 C = −1.86 ± 0.94‰) show slightly higher δ 13 C values than carbonates from the protosol (−3.25 ± 0.97‰) and gleysol (−2.38 ± 0.49‰) horizons (Supp.Table S2; Figure 9).No trend is observed in δ 18 O values.The range of δ 13 C and δ 18 O isotope values are −3.97 to −1.29 ‰ and −15.2 to −10.3 ‰, respectively, in all the paedogenic carbonates.

Inferences on processes from multi-elemental ratios
As previously discussed, the Al/Si ratio and the CIA index are proxies for clayeyness and show a positive correlation with paleosol horizons along all profiles.Using these two indicators to distinguish two types of episodic sequences (low-and moderate-high soil-water saturation) in certain paleosol horizons along the stratigraphic profiles (Figure 6).In the Ribbon Trail section, three periods of moderate-high soil-water saturation are identified along the profile: RT-1Jw → RT-8Jw; RT-10Jw → RT-13Jw; and RT-14Jw → RT-3Jmt.Two periods of low soilwater saturation are also identified in the section: RT-8Jw → RT-10Jw and RT-13Jw → RT-14Jw.Overall, the Al/Si ratio increases across 18 m up section.We interpret this trend as reflecting a lesser contribution of carbonates and greater contribution of feldspars in the protolith.In the Escalante Canyon section, two periods of moderatehigh soil-water saturation are identified along the profile: EC-3Jw → EC-5Jw and EC-1Jmt → EC-3Jmt.Two periods of low soil-water saturation are also identified: EC-2Jw → EC-3Jw and EC-5Jw → EC-1Jmt.Finally, in the Sawpit section, two periods of moderate-high soil-water saturation are identified: SP-4Jw → SP-5Jw and SP-6Jw → SP-7Jw.Only one period of low soil-water saturation was recognized: SP-5Jw → SP-6Jw.
Elemental Fe, Al, and organic C are reported in wt.% (Suppl.Table S1).The paleosol samples are in general more enriched in elemental Al than Fe.Carbon is relatively low (<1 wt.%) across the paleosol samples compared to Fe and Al.Paleosol samples from the vertisol and gleysol horizons contain slightly higher values of Fe and Al than those from the oxisol and protosol horizons, although with a few exceptions (e.g.EC-2Jw sample).    2 provides the stratigraphic position of the samples.Isotopic data from pedogenic carbonates in late quaternary vertisols formed under humid climate, from the San Bernard River, east of the texas coastal plain in Brazoria County, Texas, U.S.A. (Mintz et al., 2011) are provided for comparison: M1 (calcite precipitated under saturated conditions); M2 (calcite precipitated under saturated oxidizing conditions); and M4 (calcite precipitated under saturated hybrid conditions, i.e. mixture of conditions in M1 and M2).The δ 18 O values of these modern carbonates are similar but their δ 13 C values are variable.The values in this study are close to M2 (saturated oxidizing conditions).Further comparison is provided with four pedogenic carbonate nodules (PC1 to PC4) in paleosols from the Upper Jurassic Morrison Formation in east-central Colorado (Dunagan and Turner, 2004).These other isotopic ratios in pedogenic nodules are homogenous.The δ 13 C value are lower than in the samples of this study but the δ 18 O values are higher.
The Fe/Ti vs. (Fe + Mn)/Al plots for paleosols from the three localities (Figure 7) show a trend in redox conditions most likely reflective of the paleoenvironment during paedogenesis.Paleosols classified as vertisols and gleysols (i.e. the greenish-grey to grey horizons) trend towards a more reducing environment while those classified as protosols and oxisols (i.e. the brown and reddish horizons) reflect oxidizing conditions (Supp.Table S1; Figure 7).
Fe/Ti ratios range from 3.07 to 19.41 at Ribbon Trail, 2.05 to 7.39 at Escalante Canyon, and 2.85 to 4.20 at Sawpit.(Fe + Mn)/Al range from 0.12 to 0.32 at Ribbon Trail, 0.07 to 0.27 at Escalante Canyon, and 0.12 to 0.24 at Sawpit.The Ti/Al ratio (Figure 8) is another important elemental geochemical ratio and has been used as a proxy to discriminate provenance (mafic vs. felsic sources) of the parent material of paleosols (e.g.Sheldon and Tabor 2009;Myers et al. 2014).The higher the Ti/Al ratio, the more mafic the parent material was.Overall, Ti/Al varies in a relatively narrow range between 0.04 and 0.06 which is typical of paleosols developed over mudstones and sandstones (Sheldon and Tabor 2009).

Alteration minerals
Clay mineralogy in the studied shows the abundance of paedogenic illite, an alteration phase that is also reported to be abundant in the Upper Jurassic Morrison Formation in other localities (Owen et al. 1989;Turner and Peterson 1999;Myers et al. 2014).Alteration of smectites from repeated drying and wetting could explain the abundance of illite (Torres and Gaines 2013).However, smectite is present in minor amounts mostly in the vertisol and gleysol paleosol horizons (Figure 4(a,c)).The occurrence of paedogenic smectite suggests a moderately-drained paleoenvironment with elevated pH conditions and high activity of base cations (Borchardt 1989), which is reflective of a climate with moderate precipitation rates (Torres and Gaines 2013).The low concentration of paedogenic kaolinite in all the samples (i.e. from the weak peak intensity relative to illite and smectite) suggests that the paleoclimate during paedogenesis was not particularly arid, especially because kaolinite may also form in high MAP, well-drained conditions (Gibson et al. 2000).Kaolinites are reported to form in low pH environments that are well-drained, with extremely low base cation activity and silica input (Dixon 1991).The consistency in clay mineralogy in the Wanakah Formation and Tidwell Member from the three study sites suggests that no abrupt changes in the paleoenvironment occurred during their development.

Palaeoclimatic conditions
Geochemical elemental ratio plots show the episodic sequence of alternating soil-water saturation ranging from low to moderate (Figure 6), particularly in samples from the Ribbon Trail, CO study site where the paleosols are well-developed.The Fe/Ti vs. Fe+Mn/Al plot supports the interpretation of seasonal palaeoclimatic conditions.The paleoenvironments alternate between an oxidizing to a reducing environment during soil development (Figure 7).Major element ratios generally provide complementary information on some geochemical processes during paedogenesis.While we acknowledge that post-depositional fluids, in general, may affect the chemical compositions of sedimentary units, in this specific area fluids tend to flow primarily along high permeability sandstones (Ejembi et al. 2020), not along paleosol horizons.The geochemical data for the paleosol samples seems consistent in tracking the mobilization and enrichment of redox-sensitive elements under varying paleoenvironments (Figure 7).
A spatial summary of the distribution of humid to semi-arid palaeoclimatic conditions is provided in Figure 10.The provenance of the Morrison Formation sediments is also shown on this map (Ejembi et al. 2021).MAP estimates from this study are comparable to values reported for the Upper Morrison deposition in southern Montana (Myers et al. 2014).The periodic wetting and drying inferred from the geochemical proxies (Figure 6) might be due to the seasonality in the regional paleoclimate which potentially imparted on the paleoprecipitation and paedogenesis.

Paleoprecipitation
Using the climofunction of Sheldon et al. (2002), estimates of MAP based on the non-vertic horizon of paleosols in the three localities are ~847-1260 for at Ribbon Trail, ~ 1200-1360 for Escalante Canyon, and ~438-943 for Sawpit.Estimated MAT based on clayeyness of the protosol horizon yielded 9.4°C for Ribbon Trail and 17.0° C for Escalante Canyon, which corresponds to mesic and thermic climates, respectively (Bull 1991).
Vertisols and gleysols record periods marked by increased soil moisture content.An increase in precipitations (as indicated by the high MAP estimates) may result in subaqueous conditions and poor soil drainage.However, the low content of organic matter (<1% C) preserved in vertisols and gleysols suggests that although the paleoenvironment recorded by these paleosols shows reducing conditions, they were not anoxic.Under anoxic conditions, the rate of oxygen consumption by plant respiration is faster than it is replenished by atmospheric diffusion.Low oxygen, in the case of a strongly reducing environment, favours the preservation of organic matter by slowing down its decomposition (Armstrong et al. 1991).Oxisols and protosols reflect periods of extended subaerial exposure, moderate precipitation (although they can also form in well-drained, high MAP environments), and oxidation from atmospheric oxygen favours the accumulation of Fe and Al oxides (Silver et al. 1999) (Figure 5(c)).
The isotopic values of δ 13 C (−3.97 to −1.29 ‰) and δ 18 O (−15.20 to −10.30 ‰) in paedogenic carbonates from different paleosols horizons limited internal variation.The values of δ 13 C reflect carbonates derived from dissolved inorganic sources (DIC) in soils.Typically, carbonates derived from organic matter (OM) have values << −15.20 ‰ while carbonates derived directly from atmospheric CO 2 sources have values close to 0 ‰.Thus, the δ 13 C (−3.97 to −1.29 ‰) values in these carbonates reflect precipitation in soils that are nearly at equilibrium with the atmospheric CO 2 .δ 18 O values show a limited range of compositions that can be explained by several effects.As suggested by the presence of sparry calcite veins in concretions (Supplemental Figure ), these isotope values may reflect meteoric diagenesis (Mintz et al. 2011).Indeed, these authors reported carbonate nodules in modern vertisols that precipitated in active soils during different seasons yet displayed similar δ 18 O values regardless of variations in their δ 13 C values, the temperature of formation, and evapotranspiration.However, fluxes in evapotranspiration and seasonal precipitation exert a strong influence on the δ 18 O values of soil pore water, which invariably affects the oxygen isotopic signatures of carbonates precipitated in soils.Alternatively, the observed δ 18 O values may also be controlled by significant orographic effect such as that reported elsewhere by Peters et al. (2013).

Protoliths
The Ti/Al ratios for all the Wanakah Formation and Tidwell Member samples from the three localities range from 0.01-0.06(Figure 8).The values are low and comparable between samples and study sites.The lack of variation and low values of the Ti/Al ratios for paleosols from the three localities suggest that the sedimentary protoliths for the Wanakah Formation and Tidwell Member of the Morrison Formation were derived from continental felsic sources rather than mafic sources, consistent with the derivation and recycling of detrital zircons in sandstones from these in the Paradox Basin and CCT (Ejembi et al. 2021).Provenance analysis of detrital zircons from the Wanakah Formation and Tidwell Member shows an abundance of zircon grains that were potentially derived from the McClure Mountain syenite of the Wet Mountains, Colorado, and the granitic and rhyolitic provinces of the Amarillo-Wichita uplift in Oklahoma (Potter-McIntyre et al. 2016;Ejembi et al. 2021).
Whole-rock geochemistry, clay mineralogy, and paedogenic features in paleosols from the Wanakah Formation and Tidwell Member of the Morrison Formation suggest that the palaeoclimatic conditions that existed during the Callovian to Oxfordian in the Paradox Basin were characterized by alternating periods of wet and dry cycles due to seasonality in paleoprecipitations.The lower Wanakah Formation appears to have recorded a distinct high precipitation event across two profiles.Because the nature of existing exposures limits the number of samples per profile, it is challenging to draw conclusions regarding any seasonal behaviour in observed climatic variations.Regardless of this challenge, the data from this study and our robust interpretation of the results offer new palaeoclimatic constraints based on paleosols from these southwestern paleoenvironments and thus integrate into the regional Jurassic palaeoclimatic trend in western Colorado as reported by previous studies.

Conclusion
Paleosol development and absence of marine fossils in the Wanakah Formation and Tidwell Member of the Morrison Formation in the Paradox Basin of Colorado confirm previous palaeogeographic continental interpretations of this region of the United States.Major element geochemical proxies further constrain the palaeoclimatic environment as dominated by episodic subaqueous and subaerial conditions during paedogenesis.More specifically, the episodic wetting and drying conditions caused by seasonal precipitation and/or flooding had a strong influence on paedogenesis, characterized by the abundance of illite from alteration of smectite.In addition, MAT estimates from Ribbon Trail (9.4°C) and Escalante Canyon (17.0°C) correspond to mesic and thermic paleoclimates, respectively.MAP estimates (~1,050 mm/yr at Ribbon Trail, ~1280 mm/ yr for Escalante Canyon, and ~700 mm/yr for Sawpit) suggest sub-humid to humid regional palaeoclimatic conditions.Overall, the MAP estimates based on geochemical climofunction on well-constrained, non-vertic paleosol horizons from three distinct and yet nearby localities are relatively high.Results from this study show that local paleoenvironmental conditions, such as location across a floodplain may substantially impact temperature and precipitation estimates and bias palaeoclimatic interpretations.Finally, the subhumid and semiarid palaeoclimatic conditions determined for the Upper Morrison Formation at Sawpit, Escalante Canyon and Ribbon Trail are in agreement with those determined elsewhere for the same geological periods in the western United States (Demko et al. 2004;Myers et al. (2014).

Figure 1 .
Figure 1.Map showing the location of study areas and extent of the Jurassic outcrops in western Colorado and southeastern Utah.Map compiled from USGS digital maps library for Utah and Colorado.General stratigraphy after Dunagan and Turner (2004).

Figure 2 .
Figure 2. Field photographs showing (a) macromorphologies of paleosols in the upper section of the Wanakah formation (Jw) and Tidwell member (Jmt) of the Morrison Formation at ribbon trail, CO.The channel sandstone (above yellow dashed line) is the sandstone marker bed a of Jmt -the contact between the middle Jurassic (Jw) and upper Jurassic (Jmt) deposition in western Colorado.(b) middle-late Jurassic outcrop section in Escalante Canyon to the southeast.

Figure 3 .
Figure 3. Measured stratigraphic sections of middle-late Jurassic outcrops in the three study areas show intervals of paleosol formation (coloured).Key paedogenic features are documented along the stratigraphic section.Paleosol classification after Mack et al. (1993).

Figure 6 .
Figure 6.Multi-elements ratio plots of major geochemical proxies of paleosol profiles from the three localities.From left to right: normalized cations/Ti plots, (K+na)/Al plots, Al/Si plots and MAP plots with uncertainty equal to width of triangle symbol.The dash line across the profiles shows the lower boundary of the Tidwell member.Samples are from paleosol horizons in Figure 3 and sampling points along measured stratigraphic section for the three sites are in supp.table 1.

Figure 7 .
Figure 7. Fe/Ti vs. Fe+Mn/Al plot for paleosols in the three localities.

Figure 8 .
Figure 8. Ti/Al ratios vs. stratigraphic position (in metres) for the middle-late Jurassic paleosols from the three localities.

Figure 9 .
Figure 9. Carbon (δ 13 C) and oxygen ((δ 18 O) stable isotopes data of pedogenic carbonates from the three study sites, in color symbols.Supp Table2provides the stratigraphic position of the samples.Isotopic data from pedogenic carbonates in late quaternary vertisols formed under humid climate, from the San Bernard River, east of the texas coastal plain in Brazoria County, Texas, U.S.A.(Mintz et al., 2011) are provided for comparison: M1 (calcite precipitated under saturated conditions); M2 (calcite precipitated under saturated oxidizing conditions); and M4 (calcite precipitated under saturated hybrid conditions, i.e. mixture of conditions in M1 and M2).The δ 18 O values of these modern carbonates are similar but their δ 13 C values are variable.The values in this study are close to M2 (saturated oxidizing conditions).Further comparison is provided with four pedogenic carbonate nodules (PC1 to PC4) in paleosols from the Upper Jurassic Morrison Formation in east-central Colorado(Dunagan and Turner, 2004).These other isotopic ratios in pedogenic nodules are homogenous.The δ 13 C value are lower than in the samples of this study but the δ 18 O values are higher.

Figure 10 .
Figure 10.Map showing the regional paleoclimate trend in western United States during the callovian-oxfordian interval based on palaeoclimatic proxies in paleosols from the wanakah formation and Tidwell member of the Morrison Formation in the Paradox Basin and central Colorado trough (CCT) in western Colorado.Regional outcrop distribution of the Jurassic wanakah formation and Morrison Formation and extent of the Morrison depositional basin in western United States are modified after Peterson (1972); Tweto and Schoenfeld (1979); Craig et al. (1955); and Galli (2014).Solid lines mark state boundaries: MT-Montana; ND-North Dakota; SD-South Dakota; ID-Idaho; WY -Wyoming; NE-Nebraska; NV-Nevada; UT-Utah; CO-Colorado; KS-Kansas; AZ-Arizona; NM-New Mexico; OK-Oklahoma; TX-Texas.