Figure 4. Comparison of carbon and oxygen isotopes between C. pulveratrix and M. yantaiensis from the same horizon. Note that the δ18O value of M. yantaiensis was significantly lower than that of C. pulveratrix, while the δ13C value of M. yantaiensis wasmostly higher than that of C. pulveratrix.

5 Discussion

5.1 Oxygen isotopes in land snail shells and changes in summer monsoon rainfall
Many studies have shown that oxygen isotope in land snail shell carbonate is positively related to oxygen isotope in atmospheric precipitation. (Gu et al. , 2009; Prendergast et al. , 2016; Wang et al. , 2016; Milano et al. , 2018; Padgett et al. , 2019; Wang et al. , 2019; Zhai et al. , 2019). Generally speaking, the δ18O values of C. pulveratrix were more positive than those of M. yantaiensis(Figure 4a). This is consistent with the eco-physiological habits of the two land snail species. The M. yantaiensis snails like to live in a relatively warm and humid environment and in seasons with more abundant rainfall. Due to the rainfall effect, the summer rainfall δ18O will be more negative, so δ18O in shell carbonate of M. yantaiensis ​​is also relatively low. In contrast, the active season of C. pulveratrix is relatively cool and dry with less rainfall (such as spring and autumn), so relatively more positive oxygen isotope of rainfall during this time can result in relatively high δ18O in shell carbonate of C. pulveratrix .
Snail shell δ18O can be combined with other paleoclimate indicators such as the median grain size (Md), magnetic susceptibility (SUS) of the loess and faunal assemblages of land snails to indicate the strength of the East Asian summer monsoon (Wu et al. , 2018). A previous study has shown that the shell δ18O of C. pulveratrix can be used as an indicator of summer precipitation to reflect the strength of the summer monsoon (Gu et al. , 2009). Specifically, the shell δ18O of C. pulveratrix in the monsoon region of China decreased when the summer precipitation increased. This is consistent with the δ18O record of stalagmites in Hulu cave in Southern China (Wang et al., 2008).
Generally, the shell δ18O of C. pulveratrixshowed a negative correlation with SUS and a positive correlation with Md in the Beiyao loess-paleosol section (Figure 3). This is consistent with the results of Gu et al. (2009). In the middle part of MIS7, the δ18O of C. pulveratrix exhibited a negative shift, with the minimum value being -8.13‰. Meanwhile, the Md value decreased, the number of cold-aridiphilous species C. pulveratrixdecreased, and the number of sub-humidiphilous species M. yantaiensis increased (Figure 3). It suggested that the East Asian summer monsoon intensified during this period, and the δ18O of precipitation became more negative due to large amount of precipitation.
At the beginning of MIS6, the δ18O value of C. pulveratrix experienced a positive shift, while the SUS value also became lower, indicating that the climate tended to be drier. Subsequently, the δ18O of C. pulveratrix showed a change to more negative value, with the most negative value reaching -7.5‰, and the Md value also became lower, indicating that there have been a significant increase in rainfall amount during the middle part of MIS6.
At the end of MIS5 and during MIS4, the δ18O values ofC. pulveratrix snails were generally more positive, with an average δ18OVPDB value of -4.2‰. At the same time, the SUS increased and the Md decreased. Collectively, it indicated a relative cold and dry climatic condition.
From MIS4 to MIS3, the δ18O of C. pulveratrixsnail showed a significant decrease, indicating that the climate has entered a humid and rainy mode. However, the oxygen isotope became more positive during middle MIS3, which corresponded to the decrease in SUS. This implied that the climate during MIS3 was variable and there was once a relatively cold and dry climate. Despite this, the δ18O of C. pulveratrix during the middle MIS3 was still more negative than that during MIS4, indicating a slightly drying middle MIS3. The δ18O values of C. pulveratrix during the late stage of MIS3 were -0.6‰ by average more negative than those during the early stage of MIS3, suggested a generally more humid climate during the late MIS3. But we acknowledged that the δ18O during the early MIS3 was highly variable and some negative extrema that are even lower than the late MIS3 δ18O also appeared during this period. This may reflect some transient stages with much humid condition also occurred during the early MIS3. The three-stage sub-division of MIS3 can be also envisaged on the SUS curve of our loess section (Figure 3). The average δ18O value of C. pulveratrix was -5.3‰ during MIS3 stage. In contrast, the average δ18O during MIS2 was much higher (-4.2‰) and it showed a clear trend of increase, suggestive of a climatic transition from wetness to dryness.
Within MIS2 stage, the δ18O values of C. pulveratrix increased up to -2​​‰ at about 21.6 ka, which marked extreme dryness during the last glacial period (LGM). Similarly, the δ18O of C. pulveratrix from Mangshan loess section in central China also showed an extremely positive value (approximately -1​​‰) around 22 ka (Gu et al. 2009). The two study sites are about 100 km away. Collectively, it manifested a synchronous regional drought in central China during the LGM.
The δ18O values of M. yantaiensis exhibited almost the same pattern of variation as those of C. pulveratrixdid. During late MIS7 stage, the δ18O of M. yantaiensis was more negative than that of C. pulveratrix and attained to the most negative of -9.71‰ when the δ18O of C. pulveratrix dropped to its most negative one (Figure 3). In the meantime, SUS also increased its peak value. These lines of evidences corroborated abundant rainfall brought by the intensified summer monsoon during the late MIS7. During the early MIS3, the δ18O of M. yantaiensis showed a gradually decreasing trend, which was synchronous with the changes in C. pulveratrix δ18O and SUS. This further confirmed climate shifted to more humid condition from MIS4 to early MIS3.
5.2 Carbon isotopes in land snail shells and vegetation changes
The carbon isotope of land snail shell is mainly related to carbon isotopes of dietary plants (Goodfriend and Ellis, 2002; Stott, 2002; Metref et al., 2003; Balakrishnan and Yapp, 2004). A previous study on modern land snails in China has shown that snail shell carbonate was enriched in 13C by 14.2‰ relative to snail body that has on isotopic difference from organic diet (Liu et al. , 2006). At the same time, C3 and C4 plants have far different carbon isotope compositions, i.e., the average δ13C of C3 plant is -27.1 ± 2.0‰ whereas the average δ13C of C4 plant is -13.1 ± 1.2‰ (Farquhar et al. , 1989; O’Leary, 1998; Cerling, 1999). Therefore, the proportion of C3 to C4 plants in snail food can be estimated based on the shell-diet carbon isotope fractionation and snail shell carbon isotope. Because there is a 1.3‰ decrease in the δ13C of atmospheric CO2 since the industrial revolution due to the combustion of 13C-depleted fossil fuels, so-called Suess effect (Marino et al. , 1992), the above two δ13C end-members for C3 and C4 plants should be adjusted to -25.8‰ and -11.8‰, respectively, during the last two glacial-interglacial periods in our study.
The maximum δ13C of C. pulveratrix was ​​-7.34‰ that occurred at MIS5. Considering shell-diet carbon isotope fractionation of +14.2‰, the converted dietary δ13C was -21.5‰ and the inferred proportion of C4 plant was about 31%. The minimum δ13C of C. pulveratrixwas ​​-9.71‰ that showed at MIS7. The estimated relative C4 abundance was about 14%. In contrast, the most positive δ13C of M. yantaiensis was -3.05‰ that occurred at MIS3, corresponding to a relative C4abundance of 61%. The most negative δ13C of M. yantaiensis was -5.03‰ that showed at MIS7, converting to 47% of C4 in the food. It can be seen that M. yantaiensis snails consumed more C4 plants thanC. pulveratrix . We acknowledged that the proportion of C4 plants in snail’s food was overestimated because land snails may also take in a small portion of soil carbonates that have more positive δ13C than C3 and C4 plants. However, this does not influence our assessing the relative changes in C4 abundances over different MIS stages.
To some extent, relative abundance of C4 plants can reflect the climate and seasonal changes. At seasonal level, C4 plants prefer to grow in the summer when there are more warmth and abundant precipitation whereas C3 plants grow in spring and autumn with relatively low temperature (Sage et al. , 1999; Huang et al. , 2012). At glacial/interglacial time-scale, C4 biomass tended to increase during warm/humid interglacial periods whereas C3 biomass dominated during the cold/dry glacial periods (Liu et al., 2005; Yang et al., 2015). As shown in Figure 4, the δ13C ​​ofC. pulveratrix was mostly more negative than that of M. yantaiensis at the same horizon. This may indicate that C. pulveratrix was more active in relatively cold/arid environments or seasons and accordingly ingested more C3 plants. This is consistent with the phenomenon observed by Huang et al. (2012).
In general, the δ13C curve ofC. pulveratrix has a positive correlation with the SUS curve and a negative correlation with the δ18O of C. pulveratrix . This indicates a linkage of C3/C4 abundance in dietary food of land snails to climate changes. Specifically, the δ13C values of C. pulveratrix snail shell during late MIS7 were slightly more positive than those during MIS6, and the δ13C of C. pulveratrix during MIS3 was more positive than MIS2 and MIS4 as well (Figure 3). Because the feeding habits of the same snail would not largely change, the above variation in C4 abundance in the snail’s food may reflect the changes of C4 biomass in natural vegetation along with climate, i.e., relative abundance of C4 plants increased during the warm/humid interglacial (or interstadial) periods. This is in accordance to the aforementioned conclusion reached by previous studies (Liu et al., 2005; Yang et al., 2015).
5.3 The relationship between snail numbers of two species and environment change
During late MIS7, the number of cold-aridiphilous C. pulveratrixsnail was relatively lower than that of sub-humidiphilous M. yantaiensis and the land snail M. yantaiensis had reached a peak amount. At this time, Md became finer, SUS value increased, and the shell δ18O values ​​of both C. pulveratrix andM. yantaiensis shifted to more negative. These multiple proxies uniformly suggested that the warm and humid climate prevailed, which was suitable to the growth of sub-humidiphilous M. yantaiensis . In addition, a large number of stone artifacts were found at the depth of 11-13 m (MIS7) in the Beiyao section (Du and Liu, 2014), indicating strong human activities. The inferred warm/humid climatic condition was conducive to the intensified prehistoric human activities.
After entering MIS6, the number of cold-aridiphilous species increased and reached the peak of the whole profile at 9.7 m whereas the sub-humidiphilous species almost disappeared, which implied the climate became much colder and drier than the previous stage. In the meantime, the δ18O of C. pulveratrix shifted to more positive value, i.e., ​​up to -5.3‰, reflecting less monsoonal rainfall as well.
During most MIS5, land snail fossils were not preserved due to the influence of strong pedogenesis and there were only a few sub-humidiphilous snails at the depth of 6.5-7 m. At the end of MIS5, a small number of cold-aridiphilous species began to appear, indicating that the climate started to be relatively cold and dry, in accordance to the Md and SUS records.
To MIS4 stage, the number of cold-aridiphilous species significantly increased, reaching a maximum of 58, while sub-humidiphilous species rarely existed and even disappeared. The cold/dry climate as seen from the δ18O of C. pulveratrix, Md and SUS accounted for the flourish of the cold-aridiphilous C. pulveratrix .
During MIS3, the numbers of C. pulveratrix and M. yantaiensis showed alternative increases, further testifying variable climatic conditions. It also indicated that the climate was of moderate conditions so that both cold-aridiphilous and sub-humidiphilous species co-existed. At the early MIS3 stage, the number of C. pulveratrixdecreased when M. yantaiensis reached its peak abundance. In contrast, both the numbers of C. pulveratrix and M. yantaiensis largely reduced at the middle MIS3. To the late MIS3,M. yantaiensis went further reduced but the number of C. pulveratrix increased. This assemblage change indicated that the climate was warmer and more humid at the early MIS3 than at late MIS3. A faunal assemblage study of land snails in central Chinese Loess Plateau also suggested that the temperature and humidity were higher during the early MIS3 (Chen and Wu, 2008). However, the δ18O ofC. pulveratrix was highly variable during the early MIS3 and was not as more negative as that during the late MIS3 (Figure 3). This reflected a variable summer monsoon and an overall less rainfall during the early MIS3.