Weathering Records from an Early Cretaceous Syn-Rift Lake

Abstract

The Aptian–Albian interval represents a significant cooling phase within the Cretaceous “hothouse” climate, marked by dynamic climatic fluctuations. High-resolution continental records are essential for reconstructing terrestrial climate and ecosystem evolution during this period. This study examines a lacustrine-dominated succession of the Shahezi Formation (Lishu Rift Depression, Songliao Basin, NE Asia) to access paleo-weathering intensity and paleoclimate variability between the Middle Aptian and Early Albian (c. 118.2–112.3 Ma). Multiple geochemical proxies, including the Chemical Index of Alteration (CIA), were applied within a sequence stratigraphic framework covering four stages of lake evolution. Our results indicate that a hot and humid subtropical climate predominated in the Lishu paleo-lake, punctuated by transient cooling and drying events. Periods of lake expansion corresponded to episodes of intense chemical weathering, while two distinct intervals of aridity and cooling coincided with phases of a reduced lake level and fan delta progradation. To address the impact of potassium enrichment on CIA values, we introduced a rectangular coordinate system on A(Al₂O₃)-CN(CaO* + Na₂O)-K(K₂O) ternary diagrams, enabling more accurate weathering trends and CIA corrections (CIA_corr). Uncertainties in CIA correction were evaluated by integrating geochemical and petrographic evidence from deposits affected by hydrothermal fluids and external potassium addition. Importantly, our results show that metasomatic potassium addition cannot be reliably inferred solely from deviations in A-CN-K diagrams or the presence of authigenic illite and altered plagioclase. Calculations of “excess K₂O” and CIA_corr values should only be made when supported by robust geochemical and petrographic evidence for external potassium enrichment. This work advances lacustrine paleoclimate reconstruction methodology and highlights the need for careful interpretation of weathering proxies in complex sedimentary systems.

1. Introduction

The Cretaceous period from the Aptian to the Santonian (ca. 125–80 Ma), known as the “hothouse Earth” period, represents the most extreme greenhouse period over the last 150 Ma, characterized by elevated land surface temperatures and a significantly lower latitudinal temperature gradient compared to the present day [1,2,3,4]. Nevertheless, many studies have suggested the occurrence of transient cooling events during the Early Cretaceous [5,6,7,8,9]. Short-term fluctuations in atmospheric CO₂ concentrations imply that paleoclimate evolution has exhibited periodic variability [10,11]. Large igneous province (LIP) eruptions may have reinforced greenhouse conditions through the release of volcanic CO₂ [8], thereby initiating and regulating global warming events [12]. Additional geological evidence, such as the intensification of global weathering [13], the Oceanic Anoxic Events (OAEs) [14,15], the occurrence of Cretaceous Oceanic Red Beds (CORBs) [16,17], and the presence of giant glendonites [9], also reflects pronounced paleoclimatic changes during this time. These contrasting records suggest that the Early Cretaceous climate was highly dynamic and subject to a variety of environmental drivers [11,18].

Prior investigations into the Cretaceous paleoclimate have predominantly focused on marine deposits [19,20], given their globally correlatable sea level changes, minimal terrestrial sediment influence, and compositional homogeneity. In contrast, continental sequences are generally less continuous due to extensive erosion and depositional hiatuses, complicating precise stratigraphic correlations and restricting their application in paleoclimate reconstruction. Consequently, the reconstruction of terrestrial paleoclimate change from continental records remains limited. Nevertheless, continuous continental successions can provide higher-resolution reconstructions of terrestrial climate change and ecosystem evolution [21,22]. Compared to oceans, lakes have relatively smaller volumes of sediment and water, leading to their higher sensitivity to subtle environmental shifts, such as changes in land surface temperature, humidity, and hydrological cycles [23,24]. The geological history preserved in lacustrine sediments is, therefore, crucial for building comprehensive global paleoclimate models and understanding the mechanisms driving environmental change [25,26,27]. Moreover, continental weathering regimes control the flux of nutrients and terrestrial organic matter delivered to marine systems via runoff [28]. Thus, systematic studies of continental environments, particularly the intensity of chemical weathering driven by paleoclimate variability, are also essential for interpretations of marine bioproductivity and organic carbon cycling [29,30].

As one of the world’s largest, long-lived inland lake basins, the Songliao Basin provides an exceptionally complete sedimentary archive of Cretaceous terrestrial paleoclimate evolution [22,31]. A sequence stratigraphic framework and depositional pattern of syn-rift lake basins have been established based on the investigation of the Lower Cretaceous Shahezi Formation in the southeastern Songliao Basin [32]. The origins of terrigenous organic matter in the lacustrine deposits have been interpreted through direct correlation between depositional facies, chemical weathering conditions, and organic geochemical characteristics [33]. These studies demonstrate that the lacustrine deposition recorded in the Shahezi Formation can serve as a natural laboratory for reconstructing the Early Cretaceous paleoclimate conditions. In the present study, additional mudstone cores from the Shahezi Formation were analyzed to address the following specific questions:

• What were the tectonic setting and sedimentary provenance of the Lishu paleo-lake during the deposition of the Shahezi Formation in the Songliao Basin, NE Asia?
• How can chemical weathering conditions be reconstructed while accounting for external controls and potential uncertainties?
• What were the prevailing paleoclimate conditions in the Lishu paleo-lake watershed during the Early Cretaceous?

To address these questions, major and trace element concentrations were analyzed to reconstruct paleo-weathering intensity and interpret paleoclimatic conditions. Multiple external factors that may influence interpretations of chemical weathering intensity, such as protolith (parent rock) composition, sediment recycling, and post-depositional alteration, were assessed using mineralogical and petrographic results. To illustrate the petrographic expressions of potassium metasomatism and its impact on bulk-rock geochemistry, Shahezi Formation mudstones were compared with Lower Silurian Longmaxi (or Lungmachi) Formation mudstones (Sichuan Basin, southwestern China), which have been demonstrated to be affected by low-temperature hydrothermal fluids and associated metasomatism in prior studies [34,35]. Finally, a chemical weathering profile of the Shahezi Formation was constructed within a sequence stratigraphic framework. Potential uncertainties associated with diagenetic alteration and calculation methodologies were also discussed in this study. These findings may contribute to a better practice of paleo-weathering reconstruction in future research and have important implications for understanding the Early Cretaceous paleoclimate in continental environments.

2. Geological Settings

The Songliao Basin is a Mesozoic rift basin located in northeastern Asia (Figure 1a). Its stratigraphy is predominantly composed of Jurassic–Cretaceous strata, overlain by minor Cenozoic deposits [31]. During the Cretaceous, the Songliao Basin was positioned at mid- to high-paleolatitudes within the boreotropical climate zone (approximately 40° N to 50° N) [36]. The Lishu Rift Depression is an asymmetric half-graben, located in the southeastern part of the Songliao Basin (Figure 1b) [37]. Its stratigraphic architecture and basin configuration have been significantly influenced by several faults (Figure 1c,d). The Sangshutai Fault, which extends from the Mesozoic basement to the pre-Quaternary deposits, acts as the basin-bounding master fault along the western margin of the Lishu Rift Depression. The tectono-stratigraphic profile of the Lishu Rift Depression is subdivided into two phases: syn-rift and post-rift. The syn-rift succession, comprising the Upper Jurassic–Lower Cretaceous Huoshiling, Shahezi, Yingcheng, and Denglouku Formations, can be further divided into three stages: rift initiation, rift climax, and rift deceleration [32,37,38,39]. The Shahezi Formation, which is the focus of this study, represents the phase of the most intense rifting, with a subsidence rate up to 385 m/Ma [37,39].

The Shahezi Formation represents fan-deltaic and lacustrine deposits, consisting of interbedded conglomerates, coarse- to fine-grained sandstones, siltstones, and mudstones. Its chronology is constrained by the presence of volcanic tuff layers, including sedimentary tuff (113.9 ± 0.9 Ma, 117.9 ± 1.6 Ma) and rhyolitic crystal tuff (118.2 ± 1.5 Ma) [40,41], obtained from the SK-2e borehole drilled by the International Continental Scientific Drilling Program (ICDP) (Figure 1b). Based on zircon U-Pb ages from these tuff layers, a floating astronomical timescale has been established for the SK-2e borehole, indicating that the Shahezi Formation was developed during the Middle Aptian to Early Albian (118.2–112.3 Ma) period, spanning approximately 5.92 Ma [42].

This study focuses on a 90-m-thick, lacustrine-dominated interval of the Shahezi Formation, recovered by the JLYY1 borehole in the Lishu Rift Depression (Figure 2) [43]. A sedimentological log, presented by Wang et al. (2025), provides a detail description of the high-resolution changes in lithology, organic richness, mineralogical composition, and depositional systems tracts within this interval [32,33,44]. Their research identified several facies associations that reflect lake base-level fluctuations: subaqueous channel fills (FA2-1), pro-fan delta (FA3), nearshore and offshore lacustrine (FA5 and FA6), and sublacustrine fan (FA7) (Figure 2). The pro-fan delta deposits, characterized by co-sets of flaser, wavy, and lenticular beddings, represent the distal and subaqueous portions of fan-delta systems [32]. Nearshore and offshore lacustrine are further subdivided into four sub-facies associations: lake beach and shoreface (FA5-1), littoral (FA5-2), sublittoral (FA6-1), and profundal (FA6-2), corresponding to four distinct lake zones that redefined by the annual minimum water level, fair-weather wave base, and storm wave base [23,45,46,47,48,49]. The lake beach and shoreface comprises mud-to-silt-sized deposits characterized by ripple cross-lamination, hummocky cross-stratification, and upward-thinning bedsets, which are indicative of shallow, energetic environments that were influenced by oscillatory wave action, longshore currents, and storm currents [32,33]. The littoral facies is composed of thick successions of massive argillaceous mudstones with minor intercalation of silty laminae, reflecting substantial terrestrial siliciclastic inputs in long-lasting shallow lake environments, occasionally punctuated by low-density gravity flows [32,33]. Compared to the littoral deposits, sublittoral mudstones were deposited in deeper, relatively poorly oxygenated water settings, as indicated by higher organic matter and pyrite framboid contents, as well as abundant woody clasts and biogenic fragments [32,33]. Profundal facies, representing periods of the highest lake levels and pronounced shoreline transgression, are distinguished by high TOC (total organic carbon) contents, fine laminations, abundant pyrite aggregates, and a paucity of woody detritus and fossils [32,33]. Sublacustrine fan deposits, predominantly consisting of massive mud-rich conglomerates and sand- to pebble-bearing mudstones, are interpreted to be the results of cohesive mud-rich debris flow within deep lake settings, likely triggered by fault activity [32,33].

Figure 1. (a) The geographical position of the Songliao Basin in northeastern Asia; (b) the tectonic units, fault development, and distribution of rift depressions within the Songliao Basin [31]. The faults are the basin central faults (F1 and F6), Nenjiang–Balihan Fault (F2), Jiamusi–Yitong Fault (F3), Dunhua–Mishan Fault (F4), and Xilamulun River–Yanji Suture Zone (F5). The locations of six boreholes analyzed in this study are highlighted, which were drilled in the Xujiaweizi (borehole 1: SK-2e), Changling (borehole 2: YS3, borehole 3: SL2, borehole 4: S103, and borehole 5: B2), and Lishu (borehole 6: JLYY1) rift depressions; (c,d) interpreted seismic reflection profiles of the Lishu Rift Depression, illustrating basin geometry, major faults, sequence stratigraphic framework, and depositional facies distribution during the sedimentation period spanning the Huoshiling to Denglouku formations [39]. The T2 to T5 surfaces represent seismic reflectors serving as the boundaries between different formations. (e) A lithostratigraphic and sedimentological section from the JLYY1 borehole in the Lishu Rift Depression, showing the rifting stages, formations, lithologies, sedimentary cycles, and depositional environments. The dating of the Shahezi Formation is referenced from Ma et al. (2023) [42]. The stratigraphic column lacks the Jurassic Huoshiling Formation and the Paleozoic basement, as these strata were not drilled by the JLYY1 borehole.

Figure 1. (a) The geographical position of the Songliao Basin in northeastern Asia; (b) the tectonic units, fault development, and distribution of rift depressions within the Songliao Basin [31]. The faults are the basin central faults (F1 and F6), Nenjiang–Balihan Fault (F2), Jiamusi–Yitong Fault (F3), Dunhua–Mishan Fault (F4), and Xilamulun River–Yanji Suture Zone (F5). The locations of six boreholes analyzed in this study are highlighted, which were drilled in the Xujiaweizi (borehole 1: SK-2e), Changling (borehole 2: YS3, borehole 3: SL2, borehole 4: S103, and borehole 5: B2), and Lishu (borehole 6: JLYY1) rift depressions; (c,d) interpreted seismic reflection profiles of the Lishu Rift Depression, illustrating basin geometry, major faults, sequence stratigraphic framework, and depositional facies distribution during the sedimentation period spanning the Huoshiling to Denglouku formations [39]. The T2 to T5 surfaces represent seismic reflectors serving as the boundaries between different formations. (e) A lithostratigraphic and sedimentological section from the JLYY1 borehole in the Lishu Rift Depression, showing the rifting stages, formations, lithologies, sedimentary cycles, and depositional environments. The dating of the Shahezi Formation is referenced from Ma et al. (2023) [42]. The stratigraphic column lacks the Jurassic Huoshiling Formation and the Paleozoic basement, as these strata were not drilled by the JLYY1 borehole.

Based on the interpretation of depositional environments, sequence boundaries and flooding surfaces were identified to construct a sequence stratigraphic framework for the lacustrine-dominated interval. This interval overlies subaqueous channel sandstones, marked by a sharp and distinct contact interpreted as a flooding surface at 3167.6 m (Figure 2). The upper boundary, characterized by rip-up clasts and lag deposits, forms a sharp erosive surface (sequence boundary 2, 3077.85 m, Figure 2), where subaqueous channel sandstones abruptly overlie littoral lacustrine mudstones. Sequence boundary 1 (3122.05 m, Figure 2) is represented by an upward transition from littoral mudstones to mud-rich conglomerates within a sublacustrine fan setting. Flooding surfaces within this interval are defined by lithological transitions from underlying pro-fan delta, lake shore, and littoral siltstones and mudstones to overlying sublittoral and profundal mudstones. The maximum flooding surface (MFS, 3140.9 m, Figure 2), indicating the peak of shoreline transgression, is marked by horizontally stratified, finely laminated, and argillaceous siliceous mudstones of the profundal lacustrine facies. Based on these surfaces, this interval has been subdivided into four genetically related parasequence sets, representing four stages of lacustrine evolution (TST-1, HST-1, HST-2, and HST-3) (Figure 2) [32,33].

The Shahezi Formation mudstones analyzed in this study were compared with the Longmaxi Formation mudstones to illustrate the petrographic signatures and geochemical alterations associated with potassium metasomatism. The Lower Silurian Longmaxi Formation (444.43–438.13 Ma) was deposited in a semi-silled epicontinental sea or shelf environment, known as the Yangtze Shelf Sea [50]. This shelf sea formed as a result of the accretion of the Cathaysian Block to the Yangtze Block along the northwestern margin of Gondwana, during the transition from the Late Ordovician to the Early Silurian period [51,52]. During this period, thick marine successions, primarily composed of fine-grained, siliciclastic-dominant rocks (black shales), were widely distributed across the Sichuan Basin, including the Wufeng Formation, Kuanyinchiao Bed (or Guanyinqiao Bed, distinguished by its Hirnantian fauna shell layers), and the Longmaxi Formation [51,53,54,55].

3. Data and Methods

3.1. Sample Preparation

The Shahezi Formation mudstone samples were collected from the JLYY1 and SK-2e boreholes and subsequently processed for experiments. Small core chips were crushed into silt-sized powder using a tungsten disk mill (Retsch RS 200, Retsch Ltd., Haan, Germany) at 1000 rpm for 5 s. The resulting powders were further ground using a Retsch Planetary Ball Mill PM 100 in an 80 mL agate pot, with 20 mL of powder and 5 agate balls for 20 min at 400 rpm, yielding a grain size of less than 10 µm. The homogenized powders were then split into three aliquots for the determination of TOC contents, as well as major and trace element concentrations. Large pieces of the samples were cut perpendicular to the sedimentary bedding and further polished for optical and electron microscopy. One mudstone sample from the Longmaxi Formation was collected from the He201 borehole that was located in the southwestern Sichuan Basin for electron microscopy.

3.2. TOC Measurement, Optical and Electron Microscopy

Bulk-rock TOC contents were measured for 62 samples using a LECO CS230 C/S Determinator (LECO Co., Ltd., St. Joseph, MI, USA) at the Petroleum Geology Laboratory, China University of Petroleum (Beijing, China). Sixty-eight standard petrographic thin sections, prepared from fifty samples, were photographed and analyzed using an Olympus BX51 transmitted light microscope equipped with cellSens Imaging Software (Version 1.12, Olympus Corporation, Tokyo, Japan) at the University of Liverpool. SEM-EDS (scanning electron microscope and energy-dispersive X-ray spectroscopy) analyses were conducted on 23 samples to examine microscale rock fabrics and mineral types using a Zeiss GeminiSEM 450 equipped with an X-Max 50 mm² EDS (Carl Zeiss Ltd., Jena, Germany) at the SEM Shared Research Facility, University of Liverpool. Backscattered electron (BSE) images were acquired using a 20 kV and 1 nA beam with a work distance of 10 mm. Detailed information on the samples’ preparation and analytical procedures can be found in Wang et al. (2025) [32].

3.3. X-Ray Fluorescence Spectrometry (XRF)

Major element concentrations were determined for 93 mudstone samples collected from the JLYY1 and SK-2e boreholes using X-ray fluorescence spectrometry at the Environmental Laboratory, Central Teaching Labs, University of Liverpool. Pressed pellets were prepared by mixing 6.0 g of sample powder with 1.5 g of cellulose-based binder (CEREOX BM-0002-1, FLUXANA Ltd., Bedburg-Hau, Germany) using an FLUXANA XRF Mixer at 3000 rpm for 60 s. The mixture was then pressed into pellets with a diameter of 40 mm and a height of 3.4 mm using an Atlas Manual Hydraulic Press 15T (Specac Ltd., Orpington, UK), applying approximately 10 tons of pressure for 2 min of dwell time, followed by a 1 min release. The pellets were subsequently loaded into holders and scanned for 20 min using a Rigaku Supermini 200 WD XRF instrument (Rigaku Corporation, Tokyo, Japan). A glass pellet of NIST standards (National Institute of Standards and Technology) was analyzed along with each batch of 11 samples for data calibration. Concentrations of 11 major elements (Na, Mg, Al, Si, P, S, K, Ca, Ti, Mn, and Fe) were reported as oxides, with the total Fe contents expressed as Fe₂O₃. Analytical precision, evaluated by replicate analyses, was better than ±5% (2σ) of the reported values.

3.4. Inductively Coupled Plasma-Mass Spectrometry (ICP-MS)

Trace element concentrations were determined for 34 mudstone samples obtained from the JLYY1 borehole using an Agilent 7500 ICP-MS (Agilent Technologies, Inc., Santa Clara, CA, USA) at the Wuxi Research Institute of Petroleum Geology of SINOPEC (Wuxi, China). For each sample, 0.05 g of powder was dissolved in ultrapure water within a polytetrafluoroethylene crucible. To remove silicon, the solutions were treated with 1 mL of HF and heated to dryness at 150 °C. The residues were then sealed in a steel jar and subjected to digestion with a mixture of 1 mL of HF and 0.6 mL of HNO₃ in an oven at 190 °C for over 96 h. The resulting solution was evaporated to remove excess HF. Next, the sample was treated with 1 mL of concentrated HNO₃ and heated until the solution appeared as emulsion droplets. This step was repeated twice. After that, 1.6 mL of HNO₃ was added to the solution, which was heated at 140 °C for 3–5 h. After cooling, the solution was diluted to 50 mL with ultrapure water and 1 mL of reference solution containing 500 ng/g Rb. The final solution was analyzed, with the analytical uncertainties being better than ±5% (2σ) for most elements and 5–10% for other elements at low concentrations. Data quality was monitored using Chinese national standard materials.

3.5. Geochemical Proxies

3.5.1. Chemical Index of Alteration (CIA)

The Chemical Index of Alteration (CIA) is a commonly utilized proxy for assessing paleo-weathering conditions, calculated based on the molar proportions of the relatively immobile element (Al³⁺) in comparison to three more labile cations (i.e., Ca²⁺, Na⁺, and K⁺) within the silicate fraction of deposits [56].

CIA = Al₂O₃/(Al₂O₃ + CaO* + Na₂O + K₂O) × 100

(1)

where all oxides are measured in molar concentrations (mol/100 g), and CaO* specifically denotes the CaO content derived from the silicate fraction, thereby excluding contributions from carbonate minerals (calcite and dolomite) and phosphates (apatite) [56,57].

CaO* = CaO − CO₂ (calcite) − 0.5 CO₂ (dolomite) − 10/3 × P₂O₅ (apatite)

(2)

where the coefficients 0.5 and 10/3 indicate the molar concentrations of Ca in dolomite and apatite, respectively [58]. In the present study, CaO* was determined using the approach proposed by McLennan (1993), where CaO* is the lesser of (CaO − 10/3 × P₂O₅) and Na₂O [59].

3.5.2. Plagioclase Index of Alteration (PIA)

The Plagioclase Index of Alteration (PIA) was introduced by Fedo et al. (1995) to account for the effects of potassium metasomatism on CIA values [58]. Compared to the CIA, the PIA formula excludes K₂O from both the numerator and the denominator (Equation (3)), thereby removing the influence of the Al₂O₃ associated with K-feldspar. The PIA is suitable for assessing the weathering intensity of provenance areas where the feldspar component is dominated by plagioclase.

PIA = (Al₂O₃ − K₂O)/(Al₂O₃ + CaO* + Na₂O − K₂O) × 100

(3)

3.5.3. Index of Compositional Variability (ICV)

The Index of Compositional Variability (ICV) quantifies the concentration of Al₂O₃ relative to other major oxides in rocks (Equation (4)) [60]. SiO₂ was excluded from the ICV formula to avoid the effects of quartz dilution. Since the ICV value is related to the composition of minerals with varying resistance to weathering, it serves as an indicator of rock compositional maturity and can provide insights into tectonic setting and sediment recycling [60,61].

ICV = (Fe₂O₃ + K₂O + Na₂O + CaO* + MgO + MnO + TiO₂)/Al₂O₃

(4)

where all variables represent the molar concentrations of oxides (mol/100 g), and CaO* is the same as that used in the CIA calculation.

3.5.4. Ln(Al₂O₃/Na₂O) and Total Carbonate Content

The presence of carbonate minerals in varying concentrations may complicate the application of CIA for reconstructing paleo-weathering history [62,63]. Hence, the Ln(Al₂O₃/Na₂O) index and total carbonate content (calcite and dolomite) were employed as supplementary indicators [62,63,64]. The mineral composition data of the Shahezi Formation mudstones were obtained from Wang et al. (2025) [32].

3.5.5. Weathering Index of Parker (WIP)

The Weathering Index of Parker (WIP) was proposed to quantify the molecular proportions of mobile alkali and alkaline earth metals, i.e., Na⁺, Mg²⁺, K⁺, and Ca²⁺ (Equation (5)) [65].

WIP = (CaO*/0.7 + 2 × Na₂O/0.35 + 2 × K₂O/0.25 + MgO/0.9) × 100

(5)

Lower WIP values correspond to more significant chemical weathering, which contrasts with the trend observed in CIA values. Since the WIP only reflects the concentrations of labile elements, it may be influenced by the dilution effect of quartz and thus tends to overestimate the degree of weathering in SiO₂-rich rocks [66].

3.5.6. Rubidium-to-Strontium (Rb/Sr) Ratio

The ratio of Rb to Sr has been widely used as an indicator of paleo-weathering intensity due to the distinct geochemical behaviors of these elements [67,68]. During the chemical weathering of feldspars, Sr is preferentially leached into solution, while Rb tends to be retained by illite, resulting in weathered rocks that are enriched in Rb and depleted in Sr [69]. Low Rb/Sr ratios in marine or lacustrine sediments suggest substantial Sr input from source areas that experienced intense chemical weathering [69,70,71]. Similarly, the strontium-to-copper (Sr/Cu) ratio has also been employed to infer paleo-weathering conditions [68,72]. However, in this study, copper in the Shahezi Formation mudstones may have been derived from volcanic ashfall into the lake system [73], as supported by the frequent occurrence of tuffaceous mudstones and tuff layers (Figure 2). Consequently, Sr/Cu is not considered a reliable proxy for reconstructing the paleo-weathering intensity in the Shahezi Formation and is, therefore, excluded from this analysis.

3.6. Major Element Concentrations of the Longmaxi Formation Mudstones

The major element geochemical data for the Longmaxi Formation mudstones were adapted from Wang (2021) [74]. In that study, 139 mudstone samples collected from HD-1 and XD-1 boreholes were analyzed using ICP-OES (Inductively Coupled Plasma-Mass Spectrometry). Concentrations of nine major elements (Fe, Al, K, Na, Mg, Ti, Ca, Si, and Mn) were reported as oxides, while the total phosphorus content was expressed as elemental P (ppm). Based on these data, CIA values for the Longmaxi Formation were calculated in the present study.

3.7. Data Treatment

The z-score, quantifying the deviation of a data point from the mean, was calculated to identify outliers in the element concentrations. A threshold z-score of ±2.5 was employed to determine the outliers. Additionally, elemental concentrations were normalized to the reference values to identify distribution patterns and potential anomalies. In this study, the elemental concentrations of Post-Archaean Australian Shale (PAAS), as reported by Taylor and McLennan (1985), were selected as the reference standard to calculate the enrichment factors for major-element oxides (Equation (6)) [75,76]. Major-element oxide X is considered enriched with respect to PAAS when EF_x exceeds 1, and depleted when EF_x is less than 1.

EF_x = (X/Al) _sample/(X/Al)_PAAS

(6)

Rare earth element (REE) concentrations were divided by their values in chondrite, reported by Boynton (1984), for normalization [77]. Since Eu exhibits a significant fractionation relative to other REEs, its anomaly was calculated to infer protolith characteristics in the provenance area (Equation (7)) [78].

Eu/Eu* = Eu_N/(Sm_N × Gd_N)^0.5

(7)

where Eu_N, Sm_N, and Gd_N represent the chondrite-normalized concentrations of Eu, Sm, and Gd, respectively. Values greater than 1 indicate a positive Eu anomaly, while values less than 1 indicate a negative anomaly.

4. Results

4.1. Bulk TOC Concentrations and Mineral Composition (Shahezi Formation)

The TOC concentrations vary across the four lacustrine depositional sequences of the Shahezi Formation (Figure 2). The organic-rich TST-1 interval exhibits TOC contents ranging from 0.63% to 3.86%, with an average of 2.09%. These values are slightly lower than those found in the HST-1 deposits, where the TOC values ranged from 0.3% to 4.05%, averaging 2.21%. In contrast, the HST-2 and HST-3 intervals are characterized by relatively low organic carbon contents, with average TOC concentrations of 0.78%, and individual values ranging from 0.49% to 1.91%.

The Shahezi Formation mudstones have a mixed mineralogical composition, dominated by clay minerals (averaging 51.7%) and non-clay mineral silicates (averaging 35.1%), with lower proportions of carbonates (averaging 10.4%), pyrite (averaging 1.3%), and other minerals (averaging 1.5%, including barite, siderite, gypsum, anhydrite, and augite). The clay mineral fraction consists mainly of interstratified illite/smectite, illite, and chlorite (Figure 3). Non-clay mineral silicates primarily occur as detrital and authigenic quartz, ranging in size from silt to clay, with minimal feldspar particles (Figure 3). The carbonate minerals are predominantly calcite, comprising between 0.2% and 48.5% of the total, and averaging 9.9%, while dolomite is below 8.5%. Carbonate minerals occurred as cryptocrystalline laminae, fibrous sparry bands, and shell fragments (Figure 3). Carbonate content exhibits a marked upward increase in the lower part of the lacustrine interval, rising from generally 0.5%–16.4% in TST-1 to 9.5%–48.9% in HST-1 (Figure 2). At the base of HST-1, near the maximum flooding surface, the carbonate content increases sharply from 10% to 48.9%, followed by a general decrease toward values of 0.6–2.8% at the top of HST-3. The HST-1 deposits are distinguished by considerably higher carbonate contents compared to the other three depositional intervals (Figure 2).

4.2. Major Element Geochemistry (Shahezi Formation)

The concentrations of all major elements are presented as enrichment factors (EFs) relative to the PAAS to minimize uncertainties associated with the measurements. As illustrated in Figure 4, the enrichment factors vary in terms of their oxide types and depositional periods. The enrichment factors of Na₂O, TiO₂, K₂O, SiO₂, and Fe₂O₃ exhibit minor changes across the four periods. The samples are depleted in Na₂O and TiO₂ (EFs < 1), whereas the enrichment factors for K₂O, SiO₂, and Fe₂O₃ fluctuate around 1 from TST-1 to HST-3. In contrast, the MnO, MgO, P₂O₅, and CaO show pronounced stratigraphic variation. The lower section of the lacustrine interval (TST-1 and HST-1) is characterized by a relatively higher MnO content compared to the upper section (HST-2 and HST-3). The relative abundance of P₂O₅ peaks in TST-1 and HST-2, followed by lower values in HST-1 and HST-3. The TST-1 to HST-2 deposits are enriched in MgO and CaO (EFs > 1), whereas HST-3 is the only interval depleted in MgO and CaO. Notably, the enrichment factors values of CaO in HST-2 vary significantly between the two samples analyzed in this study, and the results may be influenced by the limited sample size. Overall, the average enrichment factors of the nine major oxides in the Shahezi Formation mudstones can be ranked as follows: Na₂O < TiO₂ < K₂O < SiO₂ < Fe₂O₃ < MnO < MgO < P₂O₅ < CaO.

4.3. Rare Earth Element (REE) Geochemistry (Shahezi Formation)

The total content of the light rare earth elements (LREE: La to Eu) averages 185.50 ppm and varies from 136.02 ppm to 235.21 ppm. In comparison, the total contents of heavy rare earth elements (HREE: Gd to Lu) average 20.58 ppm, with a range of 14.80 to 30.79 ppm. The LREEs are enriched relative to the HREEs, as indicated by the LREE/HREE ratios ranging from 7.64 to 10.83, with an average of 9.05. The REE distribution patterns display significantly steeper LREE curves compared to the HREE curves, with pronounced troughs at Eu, which is indicative of negative Eu anomalies (Figure 5). The calculated values of Eu/Eu* range from 0.43 to 0.79, with an average of 0.54.

4.4. Chemical Weathering Proxies (Shahezi Formation)

The chemical weathering indices (CIA, WIP, Ln(Al₂O₃/Na₂O), PIA, and Rb/Sr) were combined with the compositional variation proxy (ICV) to determine the chemical weathering conditions of provenance and access paleoclimate evolution during the Shahezi Formation period. The z-scores were calculated to compare different chemical weathering indices with different ranges of values and to summarize a general vertical trend of weathering intensity within the sequence stratigraphic framework. As shown in Figure 6, the chemical weathering intensity of the Shahezi Formation exhibits a gradual increase from TST-1 to HST-1, followed by a general decrease from HST-1 to HST-3. All proxies suggest that the chemical weathering intensity reached its maximum during HST-1 (3124.52–3131.22 m), with CIA ranging from 75.01 to 79.23. The transition from HST-1 to HST-2 marks a dramatic shift in chemical weathering intensity, with CIA decreasing from 76.73 to 72.44. The HST-3 is characterized by the weakest weathering conditions (with CIA values approaching 70). Additionally, the upper portion of HST-3 (3089.19–3091.69 m) records a short-time enhancement of chemical weathering, as evidenced by Ln(Al₂O₃/Na₂O) and PIA values.

4.5. Microscopic Occurrence of (Na, K, Ba) Feldspar (Longmaxi Formation)

The microscopic relationships and elemental compositions of albite, K-feldspar, and barium feldspar in the Longmaxi Formation mudstones were characterized using SEM-EDS. As shown in Figure 7, albite grains are partially replaced by K-feldspar, while the original morphologies of the albite particles were preserved. Anhedral K-feldspar is commonly rimmed or in contact with barium feldspar, which may also extend into the interior of the K-feldspar grains along cleavage planes. The K-rich barium feldspar has been identified as hyalophane, as evidenced by EDS spectra.

5. Discussion

5.1. Tectonic Settings of the Lishu Rift Depression

Siliciclastic rocks, generated through the weathering, transport, and deposition of protoliths in various tectonic settings, exhibit distinct sedimentary geochemical signatures [59,79]. As shown by the La-Th-Sc and Th-Sc-Zr/10 diagrams (Figure 5a,b), most samples plot within the domain of the continental island arc, with a few indicating active or passive continental margin settings. During the Early Cretaceous, the Songliao Basin was developed between two active continental margins: the eastern Sikhote-Alin and the northeastern Mongol-Okhotsk orogenic belts (Figure 1a). The Sikhote-Alin Belt resulted from the subduction of the paleo-Pacific plate beneath the Paleo-Asian continent [82], while the Mongol-Okhotsk Belt was generated following the closure of the Mongol-Okhotsk Ocean, which was driven by the northwestward subduction of oceanic crust beneath the Siberia Block and the subsequent collision between the Siberia Block and the Mongolia–North China continent [83,84,85,86]. These tectonic processes, and specifically, the subduction of the paleo-Pacific plate, triggered a backarc extension and gravitational collapse [87,88], resulting in the initiation of rifting within the Songliao Basin [31,89]. Thus, the Lishu Rift Depression was developed within an extensional tectonic regime of a continental island arc during the Shahezi Formation period.

5.2. Sedimentary Provenance

REE concentrations and Al₂O₃/TiO₂ ratios are widely applied as indicators of sedimentary provenance and protolith characteristics. Due to their relative geochemical stability and low mobility during weathering, deposition, and diagenesis, REE abundances and chondrite-normalized patterns are considered to be reliable proxies for reconstructing sediment source and protolith types [59,75,79,90,91]. According to Hayashi et al. (1997), Al₂O₃/TiO₂ ratios typically range from 3 to 8 for mafic igneous rocks and from 21 to 70 for felsic igneous rocks [92]. Furthermore, mafic igneous rocks are typically characterized by an absence of Eu anomalies and low LREE/HREE ratios, whereas felsic igneous rocks show pronounced negative Eu anomalies and high LREE/HREE ratios [61,75].

In this study, the Shahezi Formation mudstones exhibit similar chondrite-normalized REE patterns (Figure 5d), suggesting a uniform sediment provenance [61,75]. These REE distribution curves closely resemble those of the upper continental crust (Figure 5d), and most samples plot within the acidic arc source domain on the La/Th versus Hf diagram (Figure 5c) [80]. Therefore, the protolith of the Shahezi Formation mudstones is suggested to be predominantly derived from upper continental crustal igneous rocks. Additionally, the high LREE/HREE ratios (ranging from 7.64 to 10.83, with an average of 9.05), pronounced negative Eu anomalies (Eu/Eu* values < 0.79 and distinct troughs at Eu in the REE patterns), and elevated Al₂O₃/TiO₂ ratios, collectively indicate that the Shahezi Formation mudstones mainly originated from felsic igneous sources (Figure 5d and Figure 8a).

5.3. Chemical Weathering Intensity Reconstructed by CIA Values

5.3.1. Controls on CIA Indicator: Protolith Composition

As many factors, including the protolith composition, CaO contents in non-silicate fractions, sediment recycling, and diagenetic alteration, may have influences on the CIA values, the application of CIA as an indicator to link chemical weathering intensity with climate conditions must be approached with caution [56,58,59,93,94,95]. The A-CN-K ternary diagram, representing the molecular proportions of Al₂O₃, CaO* + Na₂O, and K₂O, is particularly useful for differentiating the compositional variations caused by weathering from those resulting from sediment provenance [58,96]. An estimate of provenance composition can be obtained by drawing a best-fit line through the linear array of sample points on the ternary diagram [94]. The best-fit line (the solid arrow in Figure 8b), representing the actual weathering trend, has an intersection point P with the Pl–Kfs line. A point P is expected to represent the original composition of unweathered rocks because the low CIA value of 50 corresponds to incipient chemical weathering [58,94].

Figure 8. Geochemical diagrams of the Shahezi Formation mudstones from the JLYY1 borehole, Lishu Rift Depression, showing the protolith characteristics and chemical weathering conditions during four depositional periods (TST-1 to HST-3). (a) Al₂O₃ versus TiO₂ plot for protolith discrimination, with the threshold values for igneous rock types referenced from Hayashi et al. (1997) [92]. (b) Ternary A-CN-K diagram, highlighting the CIA and CIA_corr values as well as weathering trends. The solid gray points represent the Shahezi Formation data from the SK-2e borehole in this study and four additional boreholes from Yang (2019) [97]. The black circles denote compositions of Post-Archean Australia Shale (PAAS), global average shale (AS), and various unweathered rocks and minerals (data from Wedepohl, 1971; Taylor and McLennan, 1985; Fedo et al., 1995) [58,75,98]. The solid arrow indicates the actual chemical weathering trend, while the dashed arrow represents the predicted progressive chemical weathering trend. Mineral abbreviations: clinopyroxene (Cpx), hornblende (Hbl), plagioclase (Pl), K-feldspar (Kfs), biotite (Bt), muscovite (Mu), illite (Ill), smectite (Sm), gibbsite (Gi), chlorite (Chl), and kaolinite (Ka). (c) The weathering trend is defined by an approximately linear relationship between CIA and WIP. UCC = upper continental crust. (d) Th/Sc versus Zr/Sc diagram, with a base map referenced from McLennan et al. (2003) [99]. The TTG (tonalite–trondhjemite–granodiorite) suite represents an association of rocks comprising the majority of the Archean continental crust.

Figure 8. Geochemical diagrams of the Shahezi Formation mudstones from the JLYY1 borehole, Lishu Rift Depression, showing the protolith characteristics and chemical weathering conditions during four depositional periods (TST-1 to HST-3). (a) Al₂O₃ versus TiO₂ plot for protolith discrimination, with the threshold values for igneous rock types referenced from Hayashi et al. (1997) [92]. (b) Ternary A-CN-K diagram, highlighting the CIA and CIA_corr values as well as weathering trends. The solid gray points represent the Shahezi Formation data from the SK-2e borehole in this study and four additional boreholes from Yang (2019) [97]. The black circles denote compositions of Post-Archean Australia Shale (PAAS), global average shale (AS), and various unweathered rocks and minerals (data from Wedepohl, 1971; Taylor and McLennan, 1985; Fedo et al., 1995) [58,75,98]. The solid arrow indicates the actual chemical weathering trend, while the dashed arrow represents the predicted progressive chemical weathering trend. Mineral abbreviations: clinopyroxene (Cpx), hornblende (Hbl), plagioclase (Pl), K-feldspar (Kfs), biotite (Bt), muscovite (Mu), illite (Ill), smectite (Sm), gibbsite (Gi), chlorite (Chl), and kaolinite (Ka). (c) The weathering trend is defined by an approximately linear relationship between CIA and WIP. UCC = upper continental crust. (d) Th/Sc versus Zr/Sc diagram, with a base map referenced from McLennan et al. (2003) [99]. The TTG (tonalite–trondhjemite–granodiorite) suite represents an association of rocks comprising the majority of the Archean continental crust.

In this study, the distribution of sample points derived from the JLYY1 borehole in the ternary diagram was too limited to generate a reasonable best-fit line. Therefore, additional data from other boreholes (i.e., SK-2e, SL2, B2, S103, and YS3) were incorporated as supplementary evidence (gray points in Figure 8b). Furthermore, a rectangular coordinate system was established to conduct a linear regression analysis of the data points (see Appendix A for more details). In this way, the oxide compositions (K_i, A_i, CN_i) of the samples were transformed to their respective orthogonal coordinates (X_i, Y_i), which allows for a robust determination of the best-fit line and the intersection point P (K_p, A_p, CN_p) based on the least squares method. As shown in Figure 8b, the best-fit line was derived for 118 samples from the Shahezi Formation mudstones. A strong correlation between X_i and Y_i (R² = 0.82, p-value < 0.001) indicates a well-defined actual weathering trend and suggests a similar protolith composition among samples from four boreholes. The intersection point P is located near tonalite (plagioclase/K-feldspar = 5.3), indicating the felsic origin of the Shahezi Formation mudstones examined in this study.

5.3.2. Controls on CIA Indicator: Potassium Metasomatism

Potassium metasomatism results in an increase in K₂O concentration of rocks when K-bearing fluids are available, which commonly occurs in siliciclastic rocks [58]. Its petrographic expression is the transformation or creation of K-bearing minerals, e.g., the reconstitution of detrital plagioclase or K-feldspar by low-temperature phases (authigenic adularia) or the transformation of aluminous clay minerals (smectite and kaolinite as mineral matrix) to illite [58,100]. Such mass-transport-related diagenetic alteration can decrease the CIA values and lead to a deviation of the linear array of sample points in a A-CN-K diagram. As indicated by the dashed arrow in Figure 8b, the predicted progressive chemical weathering trend is a straight line that originates from the intersection point P and is parallel to the A-CN sideline. Under the effects of K metasomatism, the actual chemical weathering trend (the solid arrow) deviates from the “idealized” trends, which leads to an underestimation of CIA values. To correct the excess potassium, the CIA values were corrected to CIA_corr by calculating the “pre-alteration” K₂O contents (K₂O_corr) [95].

K₂O_corr = [mAl₂O₃ + m(CaO* + Na₂O)]/(1 − m)

(8)

where m represents the molecular fraction of the K₂O for the protolith sample; in the present study, m can be determined from the intersection point P, which represents the composition of protolith rocks before the chemical weathering process (m = K_p, see Appendix A for the detailed calculation).

For the Shahezi Formation mudstones, the CIA_corr values range from 77.6 to 85.3, with an average of 82.7, which are higher than the CIA values (69.6 to 79.2, averaging 75.0). In addition, the CIA_corr values exhibit a similar stratigraphic variation trend with the CIA values (Figure 6 and Figure 9a), which is supported by their positive correlation (R² = 0.45, p-value < 0.001). Since K-feldspar constitutes only a minor proportion in all samples, never exceeding 25% of the total feldspar content [32], it is appropriate to use PIA proxies to correct for the effects of potassium metasomatism by excluding K-feldspar from the relevant equations. In this study, CIA_corr and PIA exhibit a strong positive correlation (R² = 0.97, p-value < 0.001), which is notably higher than the correlation between CIA and PIA (R² = 0.62, p-value < 0.001). The strong correlation between CIA_corr and PIA demonstrates that the influence of potassium metasomatism can be effectively minimized when using the two proxies.

5.3.3. Controls on CIA Indicator: Sediment Recycling

Multiple events of sedimentation and weathering tend to generate compositionally mature rocks that are rich in clay minerals such as kaolinite and chlorite, and have high CIA values [60]. CIA values calculated from the recycled sediments may overestimate the paleo-weathering intensity [33,101]. The clay minerals contain a higher fraction of Al₂O₃ than non-clay mineral silicates, relative to other oxides; therefore, they are characterized by lower ICV values. Based on this, the mudstones with ICV < 1 are interpreted as compositionally mature rocks that have probably undergone sediment recycling [60,61], which may not be suitable for the reconstruction of chemical weathering history.

In this study, the Shahezi Formation mudstone samples exhibit ICV values ranging from 0.88 to 1.54, with an average of 1.11 (Figure 2). Most samples have ICV values higher than 1, indicating their immature composition and first-cycle origin. In comparison, some samples from TST-1 and HST-3 are characterized by relatively lower ICV values, generally 0.88 to 0.99. Thus, the TST-1 and HST-3 may have been affected by greater sediment recycling than HST-1 and HST-2. However, despite the potential increase of CIA values by sediment recycling, the CIA values of the HST-3 samples are still lower than HST-1 and HST-2. Hence, the influence of sediment recycling does not change the decreasing trend of CIA from HST-1 to HST-3. Some samples of TST-1 have higher CIA values. However, their ICV values are higher than the values of clay minerals (0.03 to 0.78) and feldspars (0.54 to 0.87), and the correlation between ICV and CIA is weak (R² = 0.13, p-value = 0.149). It suggests that sediment recycling is not the dominant factor controlling the CIA trend. Weak sediment recycling controls, as indicated by the poor correlation between ICV and CIA, have also been documented in other Middle Permian and Early Cretaceous deposits [33,101].

5.3.4. Controls on CIA Indicator: Carbonate Abundance

The determination of CaO* in the CIA_corr formula makes corrections for the occurrence of Ca in phosphates and carbonates [59]. However, some samples from the Shahezi Formation contain substantial amounts of Ca in non-silicate minerals, such as sublittoral argillaceous mudstones rich in bivalve (calcite) and skeletal fragments (fluorapatite), as well as profundal calcareous mudstones and horizontally stratified calcite beds (see Figure 7, Figure 8 and Figure 12 in Wang et al. (2025)) [32]. Thus, the application of CIA_corr to these samples may introduce uncertainties in assessing chemical weathering intensity. To address this, the index Ln(Al₂O₃/Na₂O) was calculated to provide supplementary evidence (Figure 6) [64]. A significantly positive relationship between Ln(Al₂O₃/Na₂O) and CIA_corr occurs is measured across the entire Shahezi Formation mudstone succession (R² = 0.78, p-value < 0.001), indicating that the presence of Ca in non-silicate minerals has no substantial influence on the stratigraphic trends in CIA_corr values.

5.4. Paleoclimate Variation Recorded in the Lishu Syn-Rift Paleo-Lake

5.4.1. Chemical Weathering Proxies

Chemical weathering proxies are highly sensitive to climate change. During the chemical weathering of silicate minerals, labile constituents (e.g., Ca²⁺, Na⁺, K⁺, and Sr²⁺) are preferentially removed relative to more stable residual cations such as Al³⁺, Ti⁴⁺, and Si⁴⁺ [56,58,69,94]. As a result, the WIP and Rb/Sr values of the lacustrine sediments decrease, and other proxies increase as chemical weathering progresses. The intensity of chemical weathering is primarily controlled by both humidity and land surface temperature, with weathering rates increasing as these variables rise [102,103]. This is largely because weathering intensity is driven by the amount of humic and associated acids percolating through the weathering profile, which is regulated by rainfall [56,57,96,104]. Accordingly, high CIA values (>80) are interpreted to reflect hot and humid conditions with intense paleo-weathering conditions; medium CIA values ranging from 60 to 80 indicate warm and humid climates with moderate weathering intensity; and low CIA values (<60) correspond to cold and/or dry environments with a near absence of chemical alteration [58].

According to the relationship among CIA values, paleo-weathering intensities, and paleoclimate conditions, the high CIA_corr values of the Shahezi Formation mudstones indicate a hot and humid climate (Figure 6 and Figure 8b). For the stratigraphical changes, TST-1 and HST-1 are characterized by higher CIA values (72.7 to 79.2, averaging 75.9) than those of HST-2 and HST-3 (69.6 to 75.1, averaging 72.4), indicating intense chemical weathering during the lake expansion stage. The gradual decrease of CIA in HST-2 and the sharp decrease of CIA in the late period of HST-3 represent two transient arid and cooling events, which correspond to low lake levels and fan delta deposition. Overall, during the Early Cretaceous (Middle Aptian–Early Albian), the climate conditions were hot and humid with intermittent cooling and drying climate fluctuations in the watersheds of the Lishu paleo-lake, which is consistent with the paleoclimatic interpretations in previous studies [11,105,106].

5.4.2. Sedimentological and Palynological Evidence

The multi-scale evolution of lakes, driven by alternating wet and dry climatic cycles, can be recorded in the long-term stratigraphic variations among deltaic, siliciclastic/carbonate lacustrine shoreline, and nearshore/offshore lacustrine deposits. Furthermore, fluctuations in terrestrial vegetation exhibit a sensitive response to climate changes, making the analysis of facies associations and palynomorph assemblages a widely utilized approach for reconstructing paleoclimate conditions [107,108].

During the Shahezi Formation period, the Lishu paleo-lake deposition was characterized by siliciclastic shoreline strata, consisting of alternating siltstones and mudstones within pro-fan delta, lake beach, and shoreface settings [32]. The dominance of siliciclastics, coupled with the absence of dolomite in shoreline facies, indicates that continental runoff and fluvial input were sufficient to continuously deliver silt and clay to hydrologically open lakes, thereby maintaining low water salinity [32,108]. In a similar example, Wiggins and Harris (1994) identified two contrasting lithofacies assemblages and associated climate conditions based on the long-term lake evolution recorded in the Lower Green River Formation, southwestern Uinta Basin: a “wet-climate model” and a “dry-climate model” [108]. The siliciclastic-dominated shoreline deposits and high fluvial discharge in the Lishu Rift Depression suggest that its development occurred primarily under wet climatic conditions.

Some studies have investigated the Cretaceous climate of the Songliao Basin using palynomorph assemblages from the lacustrine mudstones of the Shahezi Formation. Yan et al. (2017) inferred a mid-subtropical climate based on algae, ostracods, and palynomorph data obtained from the LS1 and SW110 boreholes in the Lishu Rift Depression [109]. The vegetation during the Shahezi Formation period was reported to be mainly composed of herbaceous (~43%) and coniferous (~31.6%) plants. Similarly, Zhang et al. (2024) reported a comparable palynomorph assemblage in the Shahezi Formation through analysis of the SK-2e borehole in the Xujiaweizi Rift Depression [110]. Their study concluded that the vegetation consisted mainly of coniferous (~37.9%), evergreen broad-leaved (~26.1%), and herbaceous (~25.6%) plants, suggesting a predominantly temperate climate with a subtropical climate being subordinate. Overall, the wet, subtropical climatic conditions inferred from sedimentological and palynological data are consistent with the hot and humid climate indicated by the geochemical proxies in this study. Therefore, it can be confirmed that a hot and humid subtropical climate prevailed in the Songliao Basin, NE Asia, during the Middle Aptian to Early Albian.

5.5. Uncertainty Analysis

5.5.1. Inaccurate Assessment of “Excess K₂O” and Pre-Metasomatic CIA Values: Insights from the Longmaxi Formation Mudstones Affected by Hydrothermal Fluids

The metasomatic addition of potassium into sedimentary rocks has received considerable attention, as it can alter the primary bulk geochemistry, decrease the CIA values of affected strata, and consequently result in erroneous reconstructions of paleo-weathering conditions [57,58]. This diagenetic modification can be petrographically expressed by the illitization of kaolinite (Equation (9)) and the replacement of plagioclase with low-temperature phases of K-feldspar (e.g., adularia, Equations (10) and (11)) [58,100,111,112,113].

3Al₂Si₂O₅(OH)₄ + 2K⁺ = 2KAl₃Si₃O₁₀(OH)₂ + 2H⁺ + 3H₂O

(9)

NaAlSi₃O₈ + K⁺ = KAlSi₃O₈ + Na⁺

(10)

CaAl₂Si₂O₈ + 2K⁺ + 4SiO₂ = 2KAlSi₃O₈ + Ca²⁺

(11)

Given these considerations, as discussed in Section 5.3.2, a method for correcting excess K₂O in the CIA calculation was initially proposed graphically by Fedo et al. (1995) and later formalized mathematically by Panahi et al. (2000) [58,95]. However, the application of this correction introduces several issues. Algeo et al. (2025) systematically examined the underlying assumptions of this approach, specifically: (1) that the weathering trajectories of pre-metasomatized rocks follow a path parallel to the A-CN axis; (2) that deviations from the ideal weathering trend can be attributed to metasomatic K addition; and (3) that the potassium addition can be corrected by projecting data points away from the K apex on the A-CN-K diagrams [93]. As demonstrated by Algeo et al. (2025), these assumptions may not always be valid for all modern weathering profiles, and thus, potentially may lead to unnecessary or even erroneous CIA corrections [93].

In fact, beyond these assumptions, this approach has been suggested to be applicable only for correcting the K addition introduced by the illitization of kaolinite [58,111]. In previous studies, the transformation of altered plagioclase into K-feldspar is usually ignored when correcting the K addition [114,115,116], as it has been assumed to occur predominantly in the sand-sized fractions of siliciclastic sediments, with no effects on the CIA values [58,111]. However, increasing petrographic evidence indicates that the transformation of plagioclase into K-feldspar can also take place in silt-sized grains within mudstones [35,117,118,119]. As shown in Figure 7, albite particles in the Longmaxi Formation mudstones are partly replaced by K-feldspar and further rimmed by hyalophane. During this diagenetic process, the molar ratios of replaced CaO + Na₂O to added K₂O can vary widely, owing to differences in mineral solubility, diffusion rates, and the composition of fluids interacting with plagioclase of variable anorthite content [111,120]. Consequently, the resulting K-addition vectors may have different orientations in A-CN-K diagrams, which cannot be simply corrected by projection away from the K apex [93]. Therefore, if plagioclase grains in the Shahezi Formation mudstones have undergone alteration to low-temperature phases, a diagenetic process that has yet to be confirmed, the application of Panahi et al. (2000)’s approach may result in inaccurate estimates of pre-metasomatized CIA values [95]. This issue is usually overlooked in previous studies reconstructing paleo-weathering conditions [114,115,116].

The transformation of kaolinite to illite cannot be regarded as petrographic evidence of potassium metasomatism without a deliberate evaluation of the potassium source. In the Longmaxi Formation mudstones, hyalophane occurring in a grain-coating and pore-filling phase suggests hydrothermal activity (Figure 7), which can be supported by geochemical signatures (e.g., positive Eu anomalies, elevated excess Ba values, and ⁸⁷Sr/⁸⁶Sr ratios) and typical mineral assemblages (e.g., barite, anhydrite, and microcrystalline silica) [34,35,121,122]. In general, Ba-rich and K-rich hydrothermal fluids can provide an external source of potassium for plagioclase alteration [34,122]. Consequently, metasomatic introduction of K into plagioclase results in an overall increase in bulk-rock K₂O contents and K/Al ratios, both of which exhibit positive correlations with CIA and CIA_corr values (Figure 9b,d). In contrast, for the Shahezi Formation, the data array in the A-CN-K diagrams trends towards the illite pole (Figure 8b). The abundant occurrence of authigenic illite plates (Figure 10a–c) and pervasively altered plagioclase grains (Figure 10a,c,d) cannot be considered to be definitive evidence of potassium metasomatism. This is because authigenic illite may form via the transformation of smectite or kaolinite under increasing temperature and pressure [123,124,125], with the required potassium potentially sourced from the breakdown of detrital K-feldspar [126], as supported by the low K-feldspar content in the samples (less than 0.9%, averaging 0.2%) [32]. In such cases, intraformational addition of K to kaolinite does not necessarily result in an overall increase in bulk-rock K₂O. Both bulk-rock K₂O content and K/Al ratios in the Shahezi Formation display negative or insignificant correlations with the CIA and CIA_corr values (Figure 9a,c), in contrast to the positive correlations that are typically indicative of K metasomatism [93,100]. Therefore, only potassium addition from external sources, such as hydrothermal alteration or brine circulation [127], should be regarded as evidence for potassium metasomatism. In the absence of geochemical or petrographic evidence for such an external K addition, calculating “excess K₂O” and applying CIA corrections may be unwarranted and could potentially introduce significant errors in paleo-weathering intensity assessments.

5.5.2. Problems with Corrections to CaO Contents and CIA Corrections

In this study, CaO was corrected for phosphate content by subtracting 10/3 × P₂O₅ from the total CaO content [59]. For the majority of the Shahezi Formation samples (40 out of 50), the remainder of the CaO content was higher than the molar concentrations of Na₂O, so the Na₂O values were used to represent the CaO content within the silicate fraction [59]. In the remaining ten samples, where the corrected CaO was less than the Na₂O concentration, the corrected CaO value was utilized instead [59]. This method is effective for sediments derived predominantly from felsic sources and does not require any pre-chemical leaching procedures [93]. However, compared to sodium, calcium tends to be more susceptible to chemical weathering due to its larger ionic radius and weaker bonding within the plagioclase lattice [59]. Therefore, calculating CaO* likely results in an underestimation of CIA values [59]. On the other hand, a K₂O correction decreases the CIA discrepancy of the sample points. In the A-CN-K diagram (Figure 8b), the data points, located on the same line projecting backward away from the K₂O apex, have the same CIA_corr values, even though their original CIA values differ. Consequently, the use of CIA_corr may obscure stratigraphic variations in chemical weathering intensity.

6. Conclusions

In this study, the lacustrine-dominated interval of the Shahezi Formation (JLYY1 borehole, Lishu Rift Depression, southeastern Songliao Basin) was examined to reconstruct paleo-weathering intensity and paleoclimate conditions during the Early Cretaceous. Multiple chemical weathering indices, including CIA, WIP, Ln(Al₂O₃/Na₂O), PIA, and Rb/Sr, were integrated with the compositional variation proxy (ICV), alongside sedimentological and palynological evidence, to elucidate paleoclimate variations within a stratigraphic framework encompassing four stages of lake evolution, from TST-1 to HST-3.

• Tectonic background and sedimentary provenance. The Lishu paleo-lake developed within a tectonic regime of a continental island arc. The closure of the Mongol-Okhotsk Ocean and the subduction of the paleo-Pacific plate beneath the Paleo-Asian continent, during the Early Cretaceous, induced backarc extension and gravitational collapse, initiating the rifting of the Songliao Basin. REE signatures and Al₂O₃/TiO₂ ratios indicate that the Shahezi Formation deposits have a uniform provenance, primarily derived from felsic igneous sources.
• Chemical weathering intensity. A comprehensive dataset of the Shahezi Formation mudstones, collected from six boreholes across three rift depressions, was analyzed using a rectangular coordinate system and the least squares method to estimate the actual weathering trend, protolith composition, and pre-metasomatized CIA values on A-CN-K diagrams. The results indicate a dominantly tonalitic protolith (plagioclase/K-feldspar = 5.3). The CIA_corr values range from 77.6 to 85.3, with an average of 82.7, these are higher than the CIA values (69.6 to 79.2, averaging 75.0), and indicate a hot and humid paleoclimate. Furthermore, all weathering proxies exhibited stratigraphic variations from the transgressive systems tract (TST-1) to the highstand systems tract (HST-1 to HST-3). Intense chemical weathering occurred during periods of lake expansion, while two transient arid and cooling events, recorded in HST-2 and HST-3, correspond to phases of low lake levels and the rapid progradation of fan deltas.
• Paleoclimate conditions. Sedimentological and palynological analyses provide additional support for paleoclimate assessment. The predominance of siliciclastic materials and the absence of dolomite in shoreline strata of the Shahezi Formation suggest that the Lishu paleo-lake developed under wet climatic conditions. Assemblages of algae, ostracods, and palynomorphs further indicate a mid-tropical environment. Overall, a hot and humid subtropical climate, intermittently interrupted by episodes of cooling and aridity, prevailed in the Songliao Basin (NE Asia) during the Early Cretaceous (Middle Aptian–Early Albian).
• Uncertainty analysis. The method of calculating silicate CaO content, proposed by McLennan (1993), may result in an underestimation of CIA values [59]. Additionally, the K₂O correction approach, suggested by Fedo et al. (1995) and Panahi et al. (2000), could reduce the original CIA differences for certain data points [58,95]. Importantly, this correction method is specifically applicable to K addition resulting from the illitization of kaolinite; potassium enrichment derived from the replacement of plagioclase and/or K-feldspar cannot be ignored or adequately corrected using this method. Furthermore, metasomatic K₂O addition cannot be simply inferred from deviations of data arrays from the “ideal weathering trend” on A-CN-K diagrams, nor from the presence of abundant authigenic illite and altered plagioclase grains. The decomposition of intraformational detrital K-feldspar can also supply the potassium required for the transformation of smectite or kaolinite to illite, and this process does not necessarily correspond to increases in bulk-rock K₂O contents. Therefore, in the absence of robust evidence for external potassium addition, the calculation of “excess K₂O” and pre-metasomatized CIA (i.e., CIA_corr) values may be unjustified and could introduce significant errors in the assessment of paleo-weathering intensity.