The Qingtai site is located on the mound of Qingtai Village, Xingyang City, Henan Province, central China (Figure 1). In 1981, Zhengzhou Municipal Museum and Xingyang County Cultural Center excavated the Qingtai site, and ¹⁴C dates placed the time period between ca. 5,500–4,750 cal. BP, a large-scale settlement site in the political hinterland of the late Yangshao culture (Zhang and Zhao, 1987). In 2015, to explore the origin of Chinese silk, the Zhengzhou Municipal Institute of Cultural Relics and Archaeology conducted a systematic investigation, exploration, and excavation at this site again (Fang, 2018). The excavation uncovered significant relics, such as three-fold ring trenches, adobe houses, tombs, and astronomical sacrifice areas, representing the emergence of civilization in central China. In addition, jiandiping, jars, cups, and funnels, suspected to be related to alcoholic beverages, were also discovered. It was worth noting that a painted bottle was found in a tomb with niches, which was also the only complete painted pottery unearthed from Qingtai during this excavation period.

In this study, we selected 17 pottery samples from the Qingtai site for starches, phytoliths, fungi, and organic acids analyses, including seven jiandiping amphorae, one funnel, two lids, five flat-based jars, one cup, and one painted bottle (Figure 2).

Residue sampling of the pottery was completed in the field laboratory at the Qingtai site. First, each surface of the sample was washed separately with distilled water. Then, an ultrasonic tooth cleaner was used to clean the pottery inside; the ultrasonic water samples were then transferred to the 50 ml centrifugal tube as the experimental samples. To test potential contamination, we also collected the control samples by gathering dust on shelves in the field laboratory and soil on each pottery surface. Further extraction was conducted at the Bio-archaeology laboratory, University of Science and Technology of China.

Among the 218 starch grains, 19 starch grains could not be identified due to the damage or lack of typical characteristics. The remaining 199 starch grains were classified into nine broad categories based on their morphology and size (Table 1; Figure 5).

Type A: starch grains (n = 48, 8.72–20.41 μm) are mostly faceted, hilum is centric with a “−” or “+” or “Y” or star-shaped fissures, some parts of starch have depressions (Figure 5A). The extinction arms are crossed and vertical under polarizing light, and lamellae are invisible. This type of starch grains is mostly from Panicoideae plants (Ge et al., 2010; Yang et al., 2010), such as foxtail millet (Setaria italica), common millet (Panicum miliaceum), as well as Job’s tears (Coix lacryma-jobi). However, the size of a common millet is relatively small, while the size of Job’s tears is relatively large, and the end of the extinction arm of Job’s tears starch grains is mostly similar to a “Z” shaped curve. The size range and morphological structure of this type of starch granules are consistent with those of foxtail millet, so it is identified as Setaria italica.

Type B: starch grains (n = 27, 12.18–26.03 μm) are polygonal and nearly circular. Its characterized by a centric hilum, linear fissures, and invisible lamellae. The extinction arms are crossed and vertical under polarizing light, and the end of the extinction arms presents a “Z” shaped curve (Figure 5B). In modern plants, the starch grains of Panicoideae plants generally have a polyhedral structure, and the extinction arms cross vertically. According to our lab modern sample database (Figure 3B), the size of foxtail millet and common millet starch grains are relatively small, and the end of the cross-extinction arm of Job’s tears has a unique “Z” shaped curve (Ge et al., 2010). Thus, type B starch granules come from Coix lacryma-jobi.

Type C: starch grains (n = 92, 5.37–39.65 μm) are nearly round or convex lens-shaped, the hilum is centric and closed, the extinction arm is “Z” shaped; if turning over the starch after tapping the glass slide, a long “—” shaped fissure appeared. This type can observe two kinds of starch grains of different sizes. The larger starch grains (11.23–39.65 μm) have clear lamellae (Figure 5C), while the smaller ones (5.37–9.4 μm) are not obvious. Combined with related (Yang and Perry, 2013) research on the morphology of ancient and modern starch grains in Triticeae. We judged that this kind of starch grains come from the Triticeae plants.

Type D: starch grains (n = 10, 11.46–47.24 μm) are the circle (Figure 5D) and long oval (Figure 5E), which have a large size distribution range. Short “-” shaped or “X” shaped fissures can be observed at the hilum, and the extinction arms under polarizing light are cross. The lamellae of this kind of starch grains gradually spread outward around the hilum is obvious, and the hilum of the long oval starch grains is eccentric. Comparing our laboratory’s modern database (Figure 3D) and related analysis (Wan et al., 2011) of the starch grains in roots and tubers in modern southern China, this type of starch grains should come from lotus root (Nelumbo nucifera).

Type E: starch grains (n = 11, 17.35–23.18 μm) are round, semi-circular, and bell-shaped with a smooth surface and no fissure, the hilum is centric or eccentric, and the lamellae are invisible. Extinction arms are crossed vertically or curved under polarizing light (Figure 5F). Compared with the modern sample database of our laboratory (Figure 3E), the shape and length of this type of starch granules are similar to those of the root of snake gourd (Trichosanthes kirilowii), so this type of starch granule is identified as the Trichosanthes kirilowii.

Type F: starch grains (n = 3, 5.26–8.63 μm) are polyhedral structures, with the centric hilum and invisible lamellae and the cross-extinction arm is “X” shaped (Figure 5G). All these grains have small sizes of less than 10 μm, and most of them are closely arranged. Based on the morphological characteristics and the microscopic observation of starch grains of modern rice by Wei et al. (2008), we infer that Type F starch grains originate from rice (Oryza sativa).

Type G: starch grains (n = 2, 9.62–32.52 μm) are kidney-shaped and oval, and the hilum is invisible. One of them has a fissure through the long axis, and the other has no fissure (Figure 5H). According to our laboratory database and related study on legumes starch grains (Wang et al., 2013), this type of starch granules should come from Leguminosae.

Type H: starch grain (n = 1, 19.36 μm) is triangular oval, the hilum is closed and centric, the lamellae are invisible, and the cross-extinction arm is “X” shaped (Figure 5I). The starch grains of nut plants are mostly triangular oval and drop-shaped, such as Castanea sp. of the Fagaceae tribe and Quercus sp. However, the hilum of the Castanea sp. starch grains is mostly eccentric, and the Castanea sp. starch grains with a long axis greater than 10 μm can be observed with lamellae, while the Quercus sp. starch grains basically have no lamellae (Yang et al., 2009). Thus, these starch grains can be identified as Quercus sp.

Type I: starch grains (n = 5, 16.43–27.79 μm) are mainly oval-shaped, lamellae are unclear, cross-extinction arms are “X" shaped, hilum is eccentric, and the part of them have “-” shaped fissures (Figure 5J). According to the database of our laboratory, one of the typical characteristics of starch grains in roots and tubers is the eccentric hilum. These five starch grains extracted lacked further species characteristics and cannot be further distinguished, so this type can be classified as a roots or tubers plant.

Compared with the results of our simulation experiments, nearly half of the starches show signs of morphological alterations to some extent and were related to brewing behavior (n = 100, 45.9% of the total). These damage characteristics (n = 58) included small pits and cracks, missing edges, hazy extinction cross, and weakened birefringence which was the direct result of enzymatic hydrolysis of starch grains (Figure 6A,B). Simultaneously, there are also some other damage characteristics (n = 42) caused by fermentation and gelatinization, such as enlarged fissures radiating outward from the hilum, further expansion of the central depression, swelling, and melting (Figure 6C–H). Some starch grains only have the edge, and the central area almost disappears, forming a ring structure (Figure 6G). As the starch grains continue to decompose, the extinction cross arm gradually disappears until the crystal structure is lost (Figure 6G,H).

Of the three pottery samples (S3, S4, S17) yielding brewing damaged starch grains, 32 phytoliths were found only on the funnel S3, and most of which were from Poaceae plants (Table 2). Two double-peak phytoliths (Figure 7: g) from rice husk and dumbbell with scooped ends paralleled arrangement (Figure 7A) phytoliths from rice stem or leaf were present, corresponding to the presence of rice starch grains on the funnel. One Ω-type phytolith (Figure 7H) of foxtail millet husks was also found, but no η-type phytolith typical or starch of common millet was recovered. Other Poaceae families include two fan-shaped (Figure 7E), three cross (Figure 7F), 14 bilobate (Figure 7D), and two square saddle (Figure 7B) phytoliths, also found under the microscope. Further, hair cells (n = 1, Figure 7C) were found, mainly in the stem or leaf of eudicots.

Through the LC-MS analysis, 11 kinds of organic acids, including lactic acid, acetic acid, amber acid, fumaric acid, propanoic acid, butyric acid, citric acid, 3-hydroxybutyric acid, malic acid, pyruvic acid, and oxalic acid were detected from the funnel, lid, and painted bottle (The metrical data for organic acids remains is listed in Supplementary Table S1). Among them, eight kinds of organic acids were found in the funnel except for citric acid, 3-hydroxybutyric acid, and malic acid. Eight kinds of organic acids were found in the lid except for lactic acid, fumaric acid, and malic acid. Nine kinds of organic acids were found in the pottery bottle except for pyruvic acid and oxalic acid.

Among the 17 pottery samples selected from the Qingtai site, three pottery samples (funnel, lid, and painted bottle) with brewing damaged starch demonstrated that they were used as alcoholic vessels. The brewing materials at Qingtai included foxtail millet, rice, Triticeae, Job’s tear, legumes, lotus root, and Quercus, which correspond to the daily food sources as indicated by the microbotanical remains from Qingtai (Sun, 2018). This phenomenon also accords with the selection of fermentation ingredients in other Yangshao sites (Liu, 2021a).

The types of starch grains and phytoliths extracted from the funnel, lip, and painted bottle can match each other. The species of starch grains were mainly foxtail millet (22.02% of the total) and Triticeae (42.20% of the total). Correspondingly, Ω-type phytoliths from foxtail millet husks were found, and the total number of Poaceae phytoliths accounted for the highest proportion, up to 62.50%. Besides, small amounts of starch grain (n = 3) and phytoliths (n = 3) of rice were present. Although husk phytoliths were found in the funnel, it is hard to define whether the sprouting or chewing, or qu method was used, because simulation experiments showed that the husk phytoliths existed in these three saccharification methods of alcohol. What is surprising is that no fungi remain was seen in any of the three alcoholic vessels. This result is likely attributable to the preservation environment of the microorganisms. Studies have shown that the perishable nature of fungi makes it more difficult to preserve archaeological records than other plant or animal remains (Dugan, 2008; Berihuete-Azorin et al., 2018).

From the result of LC-MS analysis, four common alcoholic organic acids, including lactic acid, acetic acid, amber acid, and fumaric acid (Li, 2017; Coelho et al., 2018) were found in three pottery. Lactic acid and fumaric acid are produced by the fermentation of sugar contained in starchy raw materials such as rice and wheat owing to the action of lactic acid bacteria (Hao et al., 2021). Acetic acid is commonly found in various types of alcohol, and it is converted from different organic matters by Acetobacter through fermentation (Barata et al., 2012). Lactic acid is the most common organic acid in grain wine, followed by amber acid (Zhao et al., 2020). These results are consistent with data obtained in starch grains and phytoliths analyses, suggesting the three potteries were related to the cereal-based beverages. However lactic acid and amber acid are also present in fruit and honey wines (He, 2017; Zeng et al., 2018). Meanwhile, malic acid, citric acid, oxalic acid, and pyruvic acid are the main organic acids in fruit and honey wine; citric acid is also commonly found in honey wine (He, 2017; Li, 2017; Zeng et al., 2018). The macro-botanical remains of fruit such as jujube, grape, and hawthorn were recovered from the Qingtai site (data from Miaomiao Yang’s result, unpublished). Due to the absence of tartaric acid, which is the strongest biomarker of grape wine (Cheng et al., 2013), the addition of grapes might be ruled out. As the surfaces of fruits and honey contain yeast which can convert sugars to ethanol and carbon dioxide (McGovern et al., 2005), it can be seen that the Qingtai people were likely to add fruits and/or honey as fermentation agents.

Based on multiple lines of evidence in the current data, the wine from Qingtai was likely a mixed alcoholic beverage made of various cereals as main ingredients, fruit and/or honey as a starter, as well as supplemented by roots, tubers, beans, and nuts.

First, phytoliths were detected only on the funnel among the three wine vessels of the Qingtai site. Our simulation experiments show that filtration will affect the amount of phytoliths residues in wine. Given that silk relics have been unearthed from the Qingtai site, it is believable that people at the site were able to separate the dregs from the wine (Zhang and Gao, 1999). Combining the shapes of the three vessels, the combination of residues, and the brewing simulation experiment results, we speculated that the unhulled cereals were indeed used in the production of alcoholic beverages, and then the funnel was used to filter the dregs. Therefore, starch grains, phytoliths of husk, stems, and leaves, as well as organic acids were left on the surface of the funnel. The painted bottle was likely used to store filtered alcohol, so no phytolith was found in it. Because the sample size of the lid residue was too tiny to make a clear judgment, it might be covered on the wine storage container to prevent the evaporation of alcohol. What’s more, three pottery strainers were unearthed from the layers of the late Yangshao Culture in Dahecun, Huizui, and Yangshaocun sites in central and western Henan, and alcohol residues were found on one of the strainers (Liu et al., 2019a). It can be said that the filtering behavior had appeared in the political center of late Yangshao Culture, which represents the progress of brewing technology and people’s higher pursuit for the taste of alcoholic beverages.

Second, throughout the Yangshao period, the shape of fermentation and storage vessels gradually evolved from globular jars to pingdiqi, then to jiandiping and vats (Liu, 2021a; Feng et al., 2021). In our experiment, it is a remarkable fact that no starch grain was observed in seven jiandiping and five flat-based jars collected at the Qingtai site, suggesting that the functions of these vessels were probably not related to alcoholic beverages. The existing statistics data show that the Yangshao sites where jiandiping and pingdiqi were used as wine vessels were distributed in the Weihe River valley (Figure 1: Lingkou, Jiangzhai, Banpo, Yangguanzhai, Mijiaya, Xinjie) and the western Henan (Figure 1: Dingcun), so the findings at the Qingtai site might have reflected the inconsistency among different regions. In addition, no alcohol residue was found in 10 jiandiping or flat-based jars collected from the Shuanghuaishu site in central Henan (mid-late Yangshao Culture), but alcohol residues were detected in one jiandiping and one vat from Chengyan site in western Henan (early Yangshao Culture) base on residue analysis of other sites in our laboratory (data from Jingwen Liao’s master thesis, unpublished). These data further confirmed that the fermentation vessels in Yangshao Culture were probably not a single linear evolutionary model, but a more complex multi-mode between different regions and stages.

Yangshao Culture is famous for its painted images. This 36 cm-high painted bottle was painted with black patterns of zigzag, octagonal, and circular electric fans on its surface. The leader of the Qingtai excavation team believed the images were symbolic of the Sun and reproductive worship, and partly reflected the religious beliefs (Gu, 2018). Speculated that the geometric patterns on the Yangshao amphorae were likely to be the earliest symbol of alcoholic beverages in prehistoric China, which expressed a functional significance Liu (2021a). Whether as a religious or functional expression, such an exquisite wine vessel was a funerary object and carefully buried in a niche (the only niche in the Qingtai), highlighting the significance of alcoholic beverages in the late Neolithic.

Previous studies have pointed out that the drinking patterns in the Yangshao Culture were likely to communal siphoning drinking with group dancing (Liu, 2017b; Liu et al., 2018b; Liu et al., 2020a; Liu et al., 2020c). In the early Yangshao period, this kind of communal feast was mainly to maintain community solidarity. While in the late period, it was the competitive feasting for gaining and maintaining personal status (Feng et al., 2021). However, the phenomenon of siphoning drinking has not been observed in Qingtai. Funnel for filtering and small painted bottle for storage from Qingtai indicated a new way of drinking pattern, which might have evolved from communal siphoning drinking to drinking alone, or a few people using cups to drink the clear liquid poured from the bottle. Although no evidence of wine cups was found in this study, which may be related to the small sample size, scientific evidence has been provided for the use of wine cups at the Shuanghuaishu, the same period and area site as the Qingtai site (data from Jingwen Liao’s master thesis, unpublished). In general, the appearance of these new wine vessel types reflected the changes in drinking methods of prehistoric humans in the early civilized society of the late Yangshao culture. But it is unclear whether this change in drinking patterns implied a new form of social organization in the late Neolithic period.