In this study, the profile chronology data came from the paleomagnetic chronology measured by Zhang Weilin in the Institute of Qinghai–Tibet Plateau Research, Chinese Academy of Sciences (Zhang et al., 2020). The Nangqian profile is divided into NQ2, NQ, and NQ1 sections for sampling from bottom to top. There are 28 samples in the NQ2 section, 13 samples in the NQ1 section, 26 samples in the NQ section, and 67 samples in total.

In the NQ section, some samples are missing because of the unoutcroped stratum of about 150–200 m, the latest sample in this study is NQ1-242. The length of the section stratigraphic map is based on Zhang’s section length (Zhang et al., 2020). There is a wide range of red bed sequences in the sampled strata, and the lithologic variation trend is roughly represented as follows: paleosoil sandstone in the lower part, limestone−marl in the middle part, and siltstone−mudstone in the upper part with obvious gypsum layers, which also reflects a climate condition gradually changing from semi-arid to arid (Figure 2); the age range is approximately 51.8–37.5 Ma BP.

The sample was ground to more than 80 mesh using an agate mortar before testing for organic carbon content. 0.20 g powder samples were weighed using an electronic balance with an accuracy of 1/10,000 and placed in a quartz crucible. Before the sample was tested on the machine, the measured samples were soaked with 6% hydrochloric acid for 24 h and acidified to remove inorganic carbon. After rinsing with deionized water to neutralize it and then drying it in an oven at 40°C, the crucible containing the sample was placed on the combustion table of the instrument, and the CO₂ production was measured by using an infrared detector to calculate the total organic carbon (TOC) content of each sample. The samples were pretreated and tested in the Key Laboratory of Petroleum Resources Research of Gansu Province, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences. The instrument was LECOCS900 carbon and sulfur analyzer.

A total of 67 samples were collected in this study with an interval of about 10 m. Organic matter was extracted by a traditional Soxhlet extraction device, and the extracted organic matter was used for organic geochemical analysis. Before extraction, the sample was ground to 100–200 meshes, the extraction solvent was methylene chloride and methanol (v:v=9:1), and the extraction time was 72 h. The extracted organic matter and solvent mixture samples were evaporated and concentrated by nitrogen purging, then, n-hexane was further used as an extraction solvent, and the saturated hydrocarbon components were extracted several times using silica gel and alumina as fillers. The saturated hydrocarbon components were dried and diluted with CHCl₃ and analyzed by gas chromatography-mass spectrometry (GC-MS), and the n-alkane spectra of mass-nucleus ratio m/z=57 and m/z=85 were obtained. Then, the relative content of n-alkanes in the sample was obtained. The same procedure is used to check blank samples during testing to monitor for contamination.

Test instrument conditions: gas chromatography-mass spectrometer model HP6890GC/5973MS; the capillary column was an HP-5MS quartz capillary column (30mm×0.25mm×0.25 μm). The carrier gas was high-purity helium. The carrier gas flow rate was 1.2 ml/min, and the carrier gas linear velocity was 40 cm/s. The initial temperature of the column is 80°C, the temperature rises by 3°C per minute, the final temperature is 300°C, and the final temperature is kept constant for 20 min. The ion source was EI, the ionization energy was 70 eV, the ion source temperature was 280°C, and the interface between mass spectrometry and chromatography was 280°C.

The compound-specific carbon isotope was determined using a gas chromatography-isotope ratio mass spectrometry (Thermo Scientific MAT253). The experiments were completed in Gansu Key Laboratory of Oil and Gas Resources Research, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences.

The distribution characteristics of n-alkanes in 67 samples from the Nangqian section are shown in Figure 3. The carbon number of n-alkanes is distributed between nC₁₅ and nC₃₃. For the NQ2 section, their peak patterns can be roughly divided into two types. Among them, 21 samples from NQ2-15 to NQ2-360 show a bimodal distribution pattern, with nC₂₂ and nC₂₃ as the main peak carbon. A total of seven samples from NQ2-370 to NQ2-483 were distributed in a single-peak pattern, and the main peak carbon was nC₂₅. For the NQ section, the peak patterns were a single-peak distribution, and the main peak carbons are nC₂₅ and nC₂₆. For the NQ1 section, which was distributed in a bimodal pattern, nC₂₃ was the main peak carbon in the front peak part and nC₃₁ was the main peak carbon in the back peak part. The high carbon number part had an obvious odd carbon advantage.

In this study, nC₂₇/nC₃₁, average carbon chain length (ACL) of n-alkanes (C > nC₂₅), carbon dominance index (CPI) of n-alkanes (C > nC₂₃), and relative input index of n-alkanes of aquatic plants Paq were selected, as well as Pr/Ph, ∑ n-C₂₁ -/∑ n-C₂₂ +, etc., the corresponding calculation formula is as follows:

ACL_27-31=(27×nC₂₇+29×nC₂₉+31×nC₃₁+33×nC₃₃₎/(nC₂₇+nC₂₉+nC₃₁+nC₃₃) (Fang et al., 2021)

As shown in Figure 4 and Table 1, CPI values of Nangqian profile samples ranged from 1.02 to 3.54, with an average of 1.65. ∑ n-C₂₁ -/∑ n-C₂₂ + values range from 0.06 to 0.73, with an average of 0.27; Paq values ranged from 0.26 to 0.93, with an average of 0.61. ACL values ranged from 27.11 to 30.25, with an average of 29.05. nC₂₇/nC₃₁ ranged from 0.33 to 6.35, with an average of 1.55. nC₁₇/nC₃₁ values ranged from 0.01 to 4.57, with an average value of 0.59.

In this study, TOC data of 60 samples were obtained by measuring and removing outliers (Table 1), with a content range of 0.03–0.20% and an average value of 0.08%.

In this study, 28 n-alkanes samples from 51.8 to 37.5 Ma, were selected for carbon isotope analysis of n-alkanes. As shown in Figure 4 and Table 2, the carbon isotopes of n-alkanes nC₂₃ and nC₂₅ in the Nangqian Basin range from 39.0‰ to 29.5‰ and from 35.1‰ to 26.1‰, respectively. In long chain alkanes variation range of carbon isotopes were 34.0–23.7‰ (nC₂₇), 35.5–23.7‰ (nC₂₉), 31.9–24.8‰ (nC₃₁), and 31.6–24.5‰ (nC₃₃) (table 2). On the whole, the long-chain and middle-chain n-alkanes isotopes in the Nangqian Basin show a trend of gradual increase on a long-term temporal scale and a certain degree of periodic fluctuation in a short period. The carbon number range of carbon isotope results of n-alkanes obtained in this study includes nC₁₇-nC₃₃, but the experimental results of low carbon number part and high carbon number part were missing, and the GC-MS results of n-alkanes also indicate that the trend of long-chain n-alkanes (ACL) is relatively obvious, so we use the data to get more complete results. Long-chain n-alkanes nC₂₇−nC₃₁, represent the range of carbon numbers in higher plants as an indicator of climate change.

In 2010, Aichner analyzed the distribution and carbon isotopes of n-alkanes in aquatic plants and lake surface sediments from 40 lakes in the central and eastern Qinghai–Tibet Plateau (Aichner et al., 2010) and concluded that ¹³C produced by mid-chain alkanes from aquatic plants are richer than long-chain alkanes produced by land plants. This is because some aquatic plants in the lake can use HCO^3- for metabolism, and HCO^3- is more enriched in ¹³C than in dissolved CO₂.

However, in the sediments of fluvial and lacustrine facies profiles in the Nangqian Basin, the middle-chain n-alkanes (nC₂₃ and nC₂₅) of most samples are more depleted than long-chain n-alkanes (nC₂₇-nC₃₃) in ¹³C (Figure 4). This phenomenon has also been found in the surface sediments of Lake Challa in tropical Africa. The carbon isotopes of medium alkanes (nC₂₃ and nC₂₅) in the surface sediments of Lake Challa range from -30‰ to -48‰. Lake Challa is a crater lake with a steep slope at the bottom and no shallow water suitable for the growth of aquatic plants. Therefore, the authors believe that medium alkanes may not come from aquatic plants. However, the specific source of alkanes is not given (Damsté et al., 2011). In Xiaolongwan Lake in northeast China, the δ13C values of middle-chain n-alkanes in lake sediments are more negative than those of long-chain n-alkanes. This phenomenon is explained by the authors as follows: under the environmental conditions of anoxia at the lake bottom, methanogens decompose organic matter to produce methane and CO₂. With the seasonal flow of lake water, CO₂ enters the shallow water layer and is utilized by aquatic plants, resulting in the relatively depleted ¹³C in middle-chain n-alkanes (Sun et al., 2013).

The organic matter in lake sediments can be divided into two sources: one is from imported land higher plants and another is the endogenous aquatic organisms from the water body, which are mainly composed of lower bacteria, algae, and aquatic plants (emergent, submerged, and floating plants) (Dodd and Afzal-Rafii, 2000). The distribution range of n-alkanes carbon number in terrestrial higher plants is generally between nC₁₅ and nC₃₃, with nC₂₇, nC_29, or nC₃₃ as the main peak, which has an obvious odd-carbon advantage. The distribution range of n-alkanes carbon number in lower bacteria and algae is generally between nC₁₅ and nC₁₇, with nC₁₇ as the main peak and single-peak type distribution, without obvious odd-carbon advantage (Eglinton and Hamilton, 1967). The carbon number distribution of submerged/floating plants is generally between nC₂₁ and nC₂₅, while the distribution of n-alkanes in emergent plants and higher plants is similar, and the bimodal distribution is considered to be the mixed source (Dodd and Afzal-Rafii, 2000).

Previous studies have proposed that the sediment Paq value can indicate the proportion of n-alkanes input by submerged/floating plants in n-alkanes with high carbon numbers. It is generally believed that when the Paq value is less than 0.1, terrigenous higher plants are the main biological sources; when the Paq value is between 0.1 and 0.4, emergent plants are the main biological sources; when Paq value is between 0.4 and 1.0, the main biological source is submerged/floating aquatic macrophytes (Ficken et al., 2000).

In addition, for the two indices of ∑ n-C₂₁ ^-/∑ n-C₂₂+ and nC₁₇/nC₃₁, it is believed that they can also indicate the proportion of lower aquatic bacteria and algae to higher plants. nC₁₇/nC₃₁<0.5 indicates that higher plants are the main inputs, and nC₁₇/nC₃₁>2 indicates that bacteria and algae are the main inputs (Xie et al., 2005; Xie et al., 1999).

Therefore, according to the abovementioned indicators, we can divide the whole period into three stages to discuss the source of organic matter. In the first stage, 51.8–46.4 Ma, Paq is 0.8, nC₁₇/nC₃₁ is 0.97, ∑n-C₂₁ ^-/∑ n C₂₂+ is 0.32, and the source of organic matter is mainly submerged/floating aquatic macroplants, with a small number of bacteria and algae input. The main carbon numbers were nC₂₂ and nC₂₃. The distribution spectra of n-alkanes also show an obvious predominance of the front peak. Meanwhile, during 48.1–46.4 Ma, some n-alkanes with low content and high carbon number appeared back peak characteristics, which may be caused by the strong hydrodynamic conditions and the addition of small amounts of terrigenous organic matter.

The second stage is 46.4–42.7 Ma, in which the Paq value is 0.55, nC₁₇/nC₃₁ is 0.34, and the ∑ n-C₂₁ ^-/∑ n C₂₂+ is 0.18, indicating that the biological sources mainly changed from submerged/floating aquatic macrophytes to terrestrial higher plants, and the input of aquatic bacteria and algae is small. In the early period of this stage, the main carbon number is nC₂₅ and it gradually transients to nC₂₇ in the later period of this stage. The distribution spectra of n-alkanes also show an obvious unimodal pattern at this stage.

In the third stage, 42.7–37.5 Ma, Paq was relatively low, with an average of 0.44, and nC₁₇/nC₃₁ had an average value of 0.38, indicating that the biological sources were mainly mixed inputs of submerged/floating aquatic macroplants and terrestrial higher plants. The average value of ∑ N-C₂₁ -/∑ N-C₂₂ + was 0.26, indicating that there was less input of aquatic bacteria and algae. The distribution spectrum of n-alkanes also showed a bimodal pattern at this stage, indicating that the source of organic matter was the aquatic-terrestrial mixed source.

Modern molecular organic geochemical studies show that nC₂₇ or nC₂₉ is the main peak carbon in the distribution of n-alkanes when woody plants are dominant, and nC₃₁ is the main peak carbon when herbaceous plants are dominant. The change in the nC₂₇/nC₃₁ ratio can reflect the change in the relative content of woody and herbaceous plants. The increase in nC₂₇/nC₃₁ indicates the evolution of herbaceous plants to woody plants, and the decrease in nC₂₇/nC₃₁ indicates the evolution of woody plants to herbaceous plants (Xie et al., 1999). Different chain lengths correspond to the different plant sources; therefore, the average carbon chain lengths (ACL) of long-chain alkanes (carbon number > nC₂₆) from lake sediments can also be used to indicate different vegetation types (Xie et al., 1999). The ACL of herbaceous plants is greater than the woody plants(; Liu et al., 2018) and the ACL value change can reflect the change in the relative abundance of woody and herbaceous plants.

The carbon isotopic composition of leaf paraffin hydrocarbons is commonly used to indicate paleovegetation types (Garcin et al., 2014; Niedermeyer et al., 2017) and paleoecological environment changes (wu et al., 2019b) and has also been applied by some scholars in the paleoelevation reconstruction of the Qinghai–Tibet Plateau (Deng and Jia, 2018; Sun et al., 2015).

As for the emergence time of C₃ and C₄ plants, many previous studies have been carried out. Existing evidence shows that C₄ plants mainly appeared in Miocene around 23 Ma, the earliest to Oligocene about 32 Ma according to the sharp decline in atmospheric CO₂ concentration (Vicentini et al., 2008; Hackel et al., 2018; Bouchenak et al., 2010). So in this study, the latest deposition age of the Nangqian Basin section is around 35 Ma. Also, the growth environment of C₄ plants is in the low−mid latitude region at a lower altitude (Boutton et al., 1980). Although C₄ plants can also be seen on the Qinghai–Tibet plateau, the biomass proportion is very low (Deng and Li,2005). According to previous studies, the altitude of the Nangqian Basin in the Eocene reached 3000 m (Fang et al., 2020). Thus, before 35 Ma, existing evidence showed that there was hardly a large number of C₄ plants, so the source of Eocene sediments in the Nangqian Basin was mainly C₃ plants.

The nC₂₇/nC₃₁ value and ACL value (carbon nC₂₇-nC₃₃) of the sediments from the Nangqian section showed some correlation (n=67, R ²=0.47), According to the nC₂₇/nC₃₁ values, ACL and CPI values and the variation characteristics of carbon isotope δ¹³C of n-alkanes, the evolution of palaeovegetation in the Nangqian section can be divided into three stages, which are relatively consistent with the changes of sedimentary facies.

Stage I: 51.8–46.4 Ma, the main plant types were mid-chain woody aquatic plants. In this period, the depositional environments consisted of a braided river and an ephemeral lake, and the fluvial sedimentary facies were an open system with strong hydrodynamic conditions. In this system, the terrestrial plants carried by surface runoff may be an important factor affecting the input of terrestrial plants. Therefore, the phenomenon of high overall chain length shown by ACL and nC₂₇/nC₃₁ in this period may be caused by the input of exogenous terrestrial vegetation brought by rainfall and surface runoff.

Stage II: 46.4–42.7 Ma, during which the depositional environments gradually became stable and changed from braided rivers and ephemeral lakes to brackish lakes. As for compound-specific carbon isotope δ¹³C of n-alkanes, the value of δ¹³C of n-alkanes gradually showed a positive trend. At the same time, ACL values gradually increased, while nC₂₇/nC₃₁ gradually decreased, the Paq value decreased significantly, and the main carbon number gradually changed to a high carbon number, most of which was nC₂₅, indicating that the overall vegetation type changed from woody plants to herbaceous plants in this period. The TOC also increased significantly compared with the previous stage, indicating that plant productivity increased with the stable sedimentary environment.

Stage III: 42.7–37.5 Ma, in this period, the depositional environments mainly consisted of mudflats and salt lakes. In terms of n-alkanes, the ACL value kept a low trend and the CPI value increased significantly, while the Paq value and nC₂₇/nC₃₁ value also kept a low value on the whole. The proportion of high carbon number in main peak carbon increased significantly, indicating a rapid change in the proportion of terrestrial higher plants. With the further stability of the sedimentary environment, TOC content also increased further. In summary, herbaceous plants were still the main vegetation type in this period.

Previous studies have shown that the Paq value, the relative input index of n-alkanes of aquatic plants, can reflect the change in precipitation in geological history. A higher Paq value indicates a higher content of mid-chain n-alkanes from submerged/floating plants, indicating the expansion of the lake and higher humidity. On the contrary, the contribution of terrigenous higher plants and emergent plants increased, indicating low lake surface and relatively arid climate conditions. Relevant research conclusions have been confirmed and widely applied to the Qinghai–Tibet Plateau (Cranwell, 1973; Pu et al., 2011; Duan and Xu, 2012; Zheng et al., 2007; Cui et al., 2008). In addition, Eglinton and Hamilton considered that temperature is positively correlated with n-alkanes chain length (ACL). Plants tend to synthesize longer chains of n-alkanes when temperature increases (Eglinton and Hamilton, 1967). At the same time, Poynter and Eglinton proposed that the change of ACL value could reflect the relative change of paleotemperature in marine sediments and believed that the change of ACL value could reflect the relative change of paleotemperature in the source area of terrestrial sediments. The research results were also confirmed in the study of n-alkanes and temperature changes in modern plant leaves (Sorrel et al., 2017; He et al., 2018).

In addition, the carbon isotope of leaf paraffin in sediments is related to the types of vegetation photosynthesis, atmospheric CO₂ isotope and CO₂ concentration, temperature, precipitation, and other factors in the sedimentary basin.

Leaf paraffin alkane is formed by photosynthesis through the absorption of atmospheric CO₂ by plants. Therefore, the carbon isotopic composition of a single alkane is related to atmospheric CO₂ isotopic composition. However, the variation range of atmospheric CO₂ isotope was less than 0.8‰ during the Eocene (Tipple, 2010). Assuming that the degree of fractionation between vegetation and atmospheric CO₂ did not change, the variation of carbon isotope of CO₂ could only cause the variation of carbon isotope of leaf alkane to less than 0.8‰. In the Nangqian Basin, the variation range of a single carbon isotope can reach more than 10‰, which indicates that the single carbon isotope in the Nangqian Basin may be affected by atmospheric CO₂ isotope variation to some extent, but it is not important.

At the same time, the reduction of the atmospheric CO₂ concentration will reduce the degree of carbon isotope fractionation between leaf paraffin and atmospheric CO₂ and then affect the plant leaf paraffin isotope (Schubert and Jahren, 2012). However, the resolution of the study on the change in atmospheric CO₂ concentration is low and controversial. According to the available data, the global cooling of EOGM (Earliest Oligocene Glacial Maximum) events in the late Eocene may be related to the decrease in atmospheric CO₂ concentration (Barron, 1985; Pearson et al., 2009). If the alkane isotopes of leaf wax were mainly affected by the change of atmospheric CO₂ concentration in the Eocene, the carbon isotopes should show the same trend as atmospheric CO₂ concentration in the Eocene. However, the carbon isotope of the Nangqian Basin showed a positive deviation with the passage of time during the Eocene, which did not correspond well with the change of atmospheric CO₂ concentration, indicating that the change of atmospheric CO₂ concentration was not the main factor affecting the change of carbon isotope of n-alkanes.

Pollen fossil studies show that the vegetation of the YAL section (41–40 Ma) in the Nangqian Basin is dominated by mixed pollen of angiosperms and gymnosperms, of which the angiosperms are dominated by Nitraria and gymnosperms by Ephedra and Taxaceae (Yuan et al., 2016). In the RZ section, the dominant vegetation types of the Nangqian Basin did not change much in the late Eocene (41–38 Ma). Different types of plants have different carbon isotopes, n-alkanes in pine and cypress plants are more enriched in ¹³C than angiosperms. (Diefendorf and Freimuth, 2016; Pedentchouk et al., 2008). Therefore, if the coniferous and cypress plants increase in the basin, the carbon isotope of middle paraffin in river and lake sediments will increase. By comparing the carbon isotopes of n-alkanes from the late Eocene in the Nangqian Basin, it was found that the carbon isotopes of leaf paraffin were positively skewed with the decrease of coniferous and cypress gymnosperms, indicating that the change in vegetation type was not the direct factor causing the change of carbon isotopes of n-alkanes. The climatic factors leading to the change of vegetation types may be the main reason for the change of compound-specific carbon isotopes.

Recent studies on carbon isotopic compositions of plants at a global scale show that the carbon isotopic compositions of organic matter and n-alkanes in leaves of modern plants are obviously influenced by environmental factors, especially by annual average precipitation. A previous study (Diefendorf et al., 2010) summarized the carbon isotope data of 3310 leaves of 557 species of trees from 105 sampling sites around the world, revealing that there was a significant positive correlation between the carbon isotope fractionation among leaves of atmospheric CO₂ and the annual average precipitation, indicating that in a large spatial range, the regulation of leaf stomata on CO₂ exchange caused by water balance pressure is the main mechanism affecting carbon isotope fractionation in plants. Similarly, previous research (Kohn, 2010) collected carbon isotope data of C3 plants from 570 published sampling sites around the world and found that the carbon isotope of C₃ plants decreased with the increase in annual average precipitation. Most of the plants whose δ¹³C>-25.5‰ distributed in the region with annual mean precipitation less than 500mm, and the correlation between plant carbon isotope and precipitation gradually weakened when annual mean precipitation exceeded 1,000 mm. Liu and An (2020) collected and collated carbon isotopes of leaf paraffin from major plant types around the world and found that there is a significant negative correlation between carbon isotopes of n-alkanes from C3 plants and annual average precipitation, but a weak correlation between carbon isotopes and annual average temperature. However, carbon isotopes of leaf paraffin in C4 plants and monocotyledon plants have a weak correlation with annual average precipitation and an obvious positive correlation with annual average temperature. Therefore, the author believes that, on the basis of clear paleovegetation types, carbon isotopes of leaf paraffin can be used in the study of paleoprecipitation and paleotemperature. The vegetation types of the Nangqian Basin from the Eocene were dominated by C3 plants. Therefore, carbon isotopes of leaf paraffin may indicate the variation of annual average precipitation in the basin.

According to the results of this study, during the period of 51.8–37.5 Ma, the overall variation trend of n-alkanes in the Nangqian profile is relatively consistent (Figure 4), indicating that vegetation in the Nangqian Basin is sensitive to climate change, especially in terms of temperature and humidity. According to the index including the TOC, n-alkanes, and carbon isotope change trend in the Nangqian profile, there were obvious climate shifts at 46.4 and 42.7 Ma, which accord with the change of sedimentary facies; accordingly, the paleoclimate evolution of the Nangqian Basin from the Eocene to Oligocene can be divided into the following three stages.

Stage I: 51.8–46.4 Ma, the sedimentary facies were a braided river and an ephemeral lake, which suggests that this period kept relatively humid climate conditions; meanwhile, the carbon isotope δ¹³C of n-alkanes is generally negative, indicating a relatively humid sedimentary environment.

Previous studies on the chain length distribution characteristics of leaf paraffin of various plant types in 400 mm precipitation line in China showed that temperature in the summer (growing season) has an important influence on the chain length distribution of leaf paraffin in terrestrial higher plants. The higher the temperature in summer, the greater the ACL, but the change of ACL can only be a qualitative indicator of temperature change. It cannot be used as a quantitative indicator of temperature (Wang et al., 2018). However, such a conclusion also needs a stable sedimentary environment and vegetation type as the foundation support. In stage I, the sedimentary facies consisted of fluvial facies and ephemeral lake facies, which were relatively unstable on the whole. ACL was not an accurate indicator to reflect the trend of temperature change. At this stage, TOC has a reflection of the organic matter content in lake sediments and also has a relatively low value, it may also be affected by the unstable sedimentary environment and influenced the preservation of organic matter; therefore in stage Ⅰ period, the Nangqian Basin temperature change may need more evidence to explore further.

Stage Ⅱ: 46.4–42.7 Ma. In this stage, the ACL values of the samples increased significantly, while nC₂₇/nC₃₁ and Paq values decreased. Indicating climate dry trends because the ACL of plant leaf paraffin reflects the leaf epidermis of epicuticular wax on leaf water balance degree, the longer the carbon chain lock water ability, the stronger ACL exists in correlation with the temperature because of blades on the regulation of water in plants, and as the temperature increases, leaf transpiration of water also increases. In order to maintain water balance in the body, the upper epidermis of leaves needs to be more water-locked, so the carbon chain of leaf paraffin synthesized by plants is longer (Wang et al., 2018). Thus, in a relatively stable sedimentary environment, ACL values rise and Paq value decreases indicating a long-term climate drying trend gradually. As for temperature changes in this period, under the background of the overall global cooling, the ACL index cannot separately be used to measure temperature change; concrete temperature changes still need further discussion. During this period, the vegetation type changed greatly, from woody to herbaceous, in response to the climate characteristics of drying.

Stage Ⅲ: 42.7–37.5 Ma, during this period, nC₂₇/nC₃₁ and Paq values remain low, while ACL values remain high. As can be seen from the distribution characteristics of n-alkanes in this study, odd−even dominance is mainly concentrated in the range of high carbon numbers; accordingly, it may mean that the productivity of higher plant vegetation is significantly improved, as shown in Fig. 4.1, from 42.7 to 37.5 Ma, the distribution of NQ 1 segment samples gradually changed from unimodal to bimodal. In addition, compared with stage Ⅱ, the sedimentary environment of stage Ⅲ changed but was still stable. According to lithology, redox conditions were slightly elevated, but it is generally believed more reductive sedimentary environment is more suitable for the preservation of organic matter; thus, as for TOC changes, the preservation condition may not be the main factor. TOC increased during this period. It indicates that the climatic conditions became more suitable for herbaceous plants during this stage. ACL values also maintained high values in a relatively stable sedimentary environment, which may be some response to the temperature warming and correspond well to the Middle Eocene Climatic Optimum (MECO) event at around 40 Ma in the same period. They may be influenced by the MECO, the carbon isotope δ¹³C of n-alkanes was significantly positive in the whole during this period, indicating that climatic conditions had turned noticeably drier.

It was previously believed that a lower ratio of Pr/Ph (Pr/Ph < 0.6) usually represents a hypersaline environment of hypoxia, while a higher gammacerane also reflects higher environmental salinity Peters and Moldowen, 1993. We calculated the average values of Pr/Ph among the three different stages (Table 1) and found that the average value of Pr/Ph in stage Ⅰ is 0.30, stage Ⅱ is 0.38, and stage Ⅲ is 0.31, showing a certain increasing trend. Similarly, the maximum value of G/nC₃₀H was 2.19 and the minimum value was 0.006. The average value was 0.26, and the average value was 0.02 in stage Ⅰ, 0.13 in stage Ⅱ, and 0.43 in stage Ⅲ, during the whole 51.8–37.5 Ma period, the gammacerane increased generally and decreased in the late stage, which may be caused by the weakening of water stratification caused by the change of sedimentary environment. Therefore, the overall salinity of the Nangqian Basin also gradually increases during the whole long 51.8–37.5 Ma period. In particular, there is a relatively more obvious increase in salinity in stage Ⅲ, which corresponds to the changes in sedimentary facies and the obvious increase of gypsum layers in outcrops, indicating that the overall climate characteristics of stage Ⅲ are still drying.

Furthermore, the abovementioned compounds, especially Paq, showed some fluctuations during this period, which was consistent with the previous palynological results and clay mineral analysis results from the YAL profile, RZ profile in the Nangqian Basin, and Gongjue Basin dating from about 41.2 to 37.8 Ma (Bartonian Period) (Long et al., 2011; Miao et al., 2016; Yuan et al., 2016; Yuan et al., 2020). It was also consistent with the Eocene records of the Xining Basin (Figure 5), Qaidam Basin, and Hoh Xil Basin during the same period. Therefore, we believe that even under the condition of short-term warming and continuous drought, there was still short-term dry and wet alternation in this stage.