Introduction

Qara Qorum (Karakorum) was the capital of the Mongolian empire since its foundation by Činggis Qayan (Genghis Khan) in the early 13th century A.D. (Barkmann 2002). Qara Qorum is situated at 47° 11.63′ N, 102°47.3′ E, 1495 m above sea level where the Orchon river leaves the Changai mountains and widens its valley to a broad plain. The town in the Orchon valley at the north-western border of the Changai mountains, located about 350 km west of Ulan Bataar, covered an area of more than 2 km2 and was constructed by Chinese craftsmen (Roth 2002) (Fig. 1). Wilhelm von Rubruk, a Franciscan monk, who came to visit Mongolia as an envoy of the French king Louis the Holy, described the town, its people and economy. The town lost its role as the empire’s capital A.D. 1264 when Qubilai Qayan became emperor of China and moved the residence to Beijing. During the long-lasting wars after the fall of the Yuan dynasty in China in A.D. 1368, the town was completely destroyed in the early 15th century and never rebuilt. In the 15th century the Buddhist monastery Erdene Zuu, which still exists today, was founded in its direct neighbourhood.

Fig. 1
figure 1

The geographical position of Qara Qorum, and the main source areas for imported food in eastern and central Asia (from Brent 1976, modified)

Full size image

Excavations took place at the site 1948/49 by a Russian/Mongolian team (Kiyselyev et al. 1965), a Japanese survey in 1997 (Shiraishi 2001), and since 1999 by a German/Mongolian team, organized by the Mongolian Academy of Science, the German Archaeological Institute (Kommission für Allgemeine und Vergleichende Archäologie), and by the University of Bonn. The present excavations are situated in the Qayan’s palace and in the town-centre where the craftsmen and tradesmen used to live (Roth 2002; Erdenebat and Pohl 2002). The excavation will be continued until 2006. Afterwards, the results will be presented in an exposition when Mongolia celebrates the 800th anniversary of the capital’s foundation by Činggis Qayan.

According to common opinion the ancient Mongolians ate mainly meat, sometimes even raw. On the other hand, the quality of the archaeological finds at Qara Qorum, the pottery, the coins, glasswork and porcelain for example, indicate a high state of civilization. Furthermore, Rubruk reports cereals available at the town’s market. This fact and the discovery of “celtic fields” near the town in aerial photographs (Roth and Erdenebat 2002, plate 2) were the reasons for botanical investigations.

The climate is continental and semi-arid with short, hot summers, long, very cold dry winters and short transition phases. The annual average temperature is about 0°C, the annual precipitation less than 300 mm. The wettest period is the summer.

The present vegetation is dominated by steppe with Artemisia and Stipa and by Larix sibirica and Betula platyphylla woods, especially on the northern slopes of the mountains (Walther and Breckle 1986, 296ff.). The patterns of the woodland-steppe distribution in connection with the partly extremely high grazing pressure and the unregulated wood exploitation suggest that the present vegetation is highly influenced by humans. In particular, much more woodland might be expected under natural climatic conditions. Husbandry is dominated by domestic animals—especially horse, goat and sheep grazing, connected with nomadic or semi-nomadic lifestyle. In some places, however, we could observe un- irrigated arable fields with Setaria italica, Triticum aestivum or Panicum miliaceum. But even in the wet summers 2003 and 2004 the growth of the individual plants was very moderate, they were low and had small ears. In all observed cases we estimated the expected yields to less than 10 dt/ha. To summarize, cereal growing without irrigation has reached its climatic limit here because the annual distribution of precipitation is not favourable and the risk is high since dry years would prevent any yield at all.

Under these circumstances the existence of local cereal production in the medieval period without or with irrigation is a central question of this investigation, as well as the possible import of food plants and the influence of people on vegetation. Therefore, two main questions need to be answered: Did people manage local agriculture without irrigation? Has the present steppe been formed by the climate or by humans and their grazing animals? Present observations of woodland-steppe patterns and grazing support the second hypothesis.

Material and methods

The excavations at the khan’ palace up till now have yielded only two soil samples with plant remains. From the excavation in the centre of the city with cultural layers of more than 5 m thickness containing the remains of houses formerly inhabited by craftsmen and tradesmen, about 100 soil samples were collected during the summers of the years 2002–2004. In these samples different distinctive features are represented: Charred layers and the contents of heating channels with a lot of charcoal but with a very low concentration of seeds and fruits; mixed cultural layers with gravel, charcoal, animal bones and charred as well as uncharred plant material; cesspits or similar features with fruitstones and nuts in high concentration; concentrations of charred cereals and other “closed find assemblages” such as pot contents or chaff concentrations with uncharred material, which are often very well preserved by copper salts, to mention the most important feature types.

About 90 of these samples have been, at least partly, studied. The sample size was normally between 10 and 20 l. In special cases, especially when dealing with small features like pot contents and so on, they are naturally smaller. The samples were wet sieved, using the river Orchon water by means of a motor pump and a four-piece sieve set with a diameter of 40 cm and a finest mesh width of 0.5 mm. The sieve residues were wrapped in cloth, dried in the air, and the plant remains sorted using binocular microscopes and 10-fold magnification. As identification help we could only use our European reference collection and the usual literature used in Europe. We started to build up a herbarium and macrofossil reference collection for Mongolia. This is a difficult and long-lasting process. As a further obstacle, the only Flora of Mongolia is in Russian. We used the English translation of the Flora of USSR that covers Mongolia, too. The nomenclature of the scientific names follows Zander (2002).

To our great surprise not only charred, but also uncharred, desiccated plant remains were found in spite of the fact that the site and its soil is not waterlogged, but on the contrary, due to low precipitation, rather dry. The main reason for the preservation of uncharred organic material—wood as well—is the very low soil temperature; in deeper layers this is constant during the year and corresponds to the annual average temperature of about 0°C. These are not suitable conditions for microbial soil activity. Accordingly in the uppermost 2 m, where the soil temperature and water content changes during the year, only charred material is preserved.

Coring for pollen analysis in the Orchon valley was not successful because the fillings of the old Orchon branches consisted of coarse gravel. Pollen analysis of a deep core from the centre of Lake Ugii Nuur (Walther 2002), about 50 km north of Qara Qorum, is in preparation. First results are presented.

Results and discussion

In total nearly 50000 plant remains were identified. The results are summarized in the tables. Table 1 shows presence and abundance of the food plants. The list is not shorter than in most European medieval towns like Köln, Konstanz, Freiburg, Braunschweig or Lüneburg (Knörzer 1987; Küster 1989; Sillmann 2002; Hellwig 1990; Wiethold 1995). At the moment, we can only differentiate between the palace and the town centre. The latter material is from a trench of 30×20 m2 and dates between the early 13th and late 14th century A.D. A differentiation according to features and settlement phases will be possible when the archaeological analysis is completed. Some plant remains are shown in Fig. 2

Table 1 Food plants from medieval Qara Qorum; abbreviations: sp: state of preservation, uc: uncarbonised, c: carbonised, mi: mineralised
Full size table
Fig. 2
figure 2

Fruits and seeds of medieval Qara Qorum: 1, Ficus carica, nutlet; 2, Vitis vinifera, seed; 3, Pinus sibirica, nut-shell fragment; 4, unknown fruit; 5, Ziziphus jujuba, seed; 6, Triticum spelta, caryopsis; 7, Cannabis sativa, seed; 8, Piper nigrum, fruit; 9, Triticum aestivum, caryopsis; 10, Hordeum vulgare, caryopsis; 11, Juglans regia, nut-shell fragment; 12, Prunus dulcis, fruit fragment; 13, Panicum miliaceum, spikelet; 14, Prunus persica, fruit; 15, Oryza sativa, glume; 16, Coriandrum sativum, fruit; 17, Lycium sinense, seed; 18, Cucumis melo, seed; 19, unknown seed; 20, Carum carvi, mericarp; the scale bar is usually 1 mm, for Ziziphus jujuba, Prunus dulcis and P. persica 5 mm

Full size image

In the palace, charred grains of Hordeum vulgare and Triticum aestivum/durum/turgidum and a charred rachis fragment of Triticum aestivum were found. Uncharred remains of Pinus sibirica, Ziziphus jujuba, Juglans regia, Prunus dulcis, Corylus avellana, Panicum miliaceum and charred remains of Triticum aestivum/durum/turgidum and Hordeum vulgare from another sample are younger and date back to the 15th century, already the early Buddhist period.

In the town, the most common cereal is Panicum miliaceum with a constancy of 40% in the samples for its uncharred glumes. More than three quarters of all plant remains are uncharred Panicum glumes. Several pot contents consist mainly of desiccated grain remains of Panicum, several uncharred chaff concentrations were found as well as concentrations of charred grains.

The next cereal to mention is Hordeum vulgare, present mainly as the hulled form, but also as naked barley (Fig. 2). More than 4000 charred caryopses and also some barley chaff was found. Charred caryopses of Hordeum vulgare are much more abundant than those of Panicum miliacum, so barley was the most important crop, bearing in mind also that its grains are much bigger.

The third important cereal is Triticum aestivum (Fig. 2). Its grains are less common than those of Hordeum, but its rachis fragments are more numerous. They allow the identification of hexaploid naked wheat (Triticum aestivum). But keeping in mind the huge amount of Hordeum grains and the well-known fact that Hordeum rachis fragments are rather badly represented in the archaeobotanical record (Maier 2001), we can suggest that barley was of higher economic importance than bread wheat. Setaria italica is much rarer, occurring regularly in combination with large amounts of Panicum miliaceum (Fig. 2).

All other cereals known in the medieval period of Europe are also present, but rare, as are Avena sativa, Triticum spelta, Secale cereale, T. monococcum and T. dicoccum (Fig. 2). As last, but perhaps an important cereal, Oryza sativa has to be mentioned, only observed in a few uncharred glume fragments (Fig. 2).

The four most common cereals are obligate or preferable spring or summer crops. Winter crop cultivation in Mongolia would be impossible due to low winter temperatures down to −40°C. The occurrence of cereal chaff, straw and grains of these most common cereals in the charred remains from stores, as well as this dominance of summer crops, seems to confirm the hypothesis of local production. On the other hand, traditional Mongolian nomadic people imported cereals as complete plants and did the crop processing themselves, using chaff and straw as fodder or for other purposes even in the 20th century. Therefore the presence of chaff and straw seems to be no sure proof of a producer site as often argued (discussion in Jacomet and Kreuz 1999, pp 146ff.). On the other hand, the “celtic fields” themselves indicate local agrarian production. Therefore we suppose local production of Panicum miliaceum, Hordeum vulgare, Triticum aestivum and Setaria italica as summer crops. The rather large size of the grains and the associated weeds—mostly sedges and Chenopodiaceae which indicate wet and fertile soils—indicate irrigation and fertilization of the fields. Fertilization with dung of lifestock should have been no problem and water for irrigation was available in the nearby river Orchon. Due to low precipitation and especially the dry winter and spring, even the cultivation of spring and summer cereals is difficult to impossible without irrigation. Therefore it is most probable that the local cereal production was in irrigated fields. It should be possible to find old irrigation channels using archaeological prospection methods. Arable fields with irrigation require a high state of social organisation (cf. Nesbitt and O’Hara 2000).

The other, rarer cereals should have been imported, although only in small quantities, or they grew locally and scattered in wheat and barley fields as “weeds”, perhaps because the seed-corn was not pure, but contaminated. Avena needs a rather wet climate. Secale cereale, Triticum spelta und T. monococcum are normally grown as winter crops. If these typical crops of medieval Europe were imported, then most probably from central or western Asia or even from Europe, as well as T. dicoccon. This cereal, mostly grown as a summer crop, would have been suitable for local cereal production as well. Why it was not chosen is not known. Perhaps because its yields were lower than those of Hordeum vulgare, Triticum aestivum, and Panicum miliaceum under the given conditions, and because its nutrient advantage of a high protein content, was not important for people who had enough meat to consume. Oryza sativa can be considered as a Chinese import. The rarity of rice is a little bit striking. On the other hand, rice is also very rare in medieval European towns where it is considered as an indicator of prosperity (Arndt and Wiethold 2001).

Among the oil and fibre plants, only Cannabis sativa is more or less common (Fig. 2). This species also occurs today rather frequently and in large amounts in Mongolia, especially on moderately moist soils in valleys where the grazing pressure is not too high. Past or recent usage, for example as a drug, is not known from written or other sources. But the number of identified seeds indicate use. If this medieval use was based on cultivated or wild gathered material and in which way the plant was used is unknown.

The other species, Camelina sativa and Brassica rapa, are rare. Keeping in mind their normally weak representation, the local cultivation of both of them must be considered as plausible. In the second case the plant was perhaps used as a vegetable (Chinese cabbage).

Also rather rare, compared with European medieval towns, are pulses, Lens culinaris, Pisum sativum and Vicia faba. A possible reason for the low importance of both, oil plants as well as pulses, is the ample supply of animal fat and protein.

Vegetables and spices are, due to the same taphonomic reasons as in European towns, rare (cf. Willerding 1991). Most common is Carum carvi, probably wild gathered, as well as Juniperus. The others were either grown in gardens or imported, most probably from China. To be mentioned are Coriandrum sativum, Apium graveolens, Beta vulgaris, Foeniculum vulgare, Humulus lupulus, Lycium chinense, an Allium species, and perhaps Petroselinum crispum (Fig. 2). Surely Piper nigrum must have been most certainly imported from southeast Asia.

Among the fruits and nuts imported material prevails, because most perennial fruit trees or shrubs cannot survive in the cold Mongolian winter. Only Fragaria vesca, Prunus padus, Sorbus and Pinus sibirica (Fig. 2) can be gathered from the wild, the latter also being imported, since the tree does not occur in Qara Qorum’s vicinity but has its nearest stands in the Changei mountains at a distance of about 100 km: This was also the case in the past as Pinus cembra type never exceeds 10% in the pollen diagram from Ogii Noor. P. sibirica seed shell fragments and Fragaria nutlets are the most frequent fruits/nuts.

The most common imported fruits are Vitis vinifera, Ficus carica and Ziziphus jujuba (Fig. 2). Their nearest occurrences are in Turkestan, 3200 km away, Cashmere, 2200 km, northwest India, 2500 km and northern China, 1600 km away. Prunus insititia and P. dulcis are also not rare (Fig. 2). Their nearest occurrences are in China and central Asia. The other fruits were only found sporadically: Cucumis melo and Prunus avium, imported probably from central Asia, P. persica, Malus, Pyrus, Morus, Hibiscus, Corylus and Physalis alkekengi from China, and Juglans regia from the Himalaya region (Fig. 2).

Some common or significant wild plants are presented in Table 2. We must concede that up till now we have not been able to identify all types of wild plants to species or genus level, due to the lack of a complete modern reference collection. Also some probably useful plants could not be identified (Fig. 2).

Table 2 Wild plants from medieval Qara Qorum (selection)
Full size table

The most frequent weeds are Chenopodium species (Table 2). Former use can not be excluded. They occur desiccated, charred, but also recent/subrecent as contamination, which is not always easy to distinguish. Well known weeds from the European context are Bupleurum rotundifolium, Chenopodium hybridum, Descurainia sophia, Galium spurium, Polygonum aviculare, P. convolvulus, Solanum nigrum, Thlaspi arvense, Thymelaea passerina, Vaccaria pyramidata and others. The both most frequent ruderals, Hyoscyamus niger and Urtica cannabina, are also very frequent in the recent vegetation. The species from footpath communities are well known from Europe. Plants from steppe/grassland and from shrubs and woods are weakly represented. More frequent are plants from wet habitats, probably to be situated in the Orchon valley in which the braided river follows several intertwining branches.

From the Ugii Nuur profile, in a distance of about 50 km north of Qara Qorum, the upper 6.5 of 8 m have been analyzed, although with wide sampling distances (Fig. 3). No precise time model exists for this profile because the radiocarbon dates are still in preparation, but we estimate that at least the middle and late Holocene is represented. The bottom of the profile probably is of Lateglacial age (Walther, personal communication). Throughout the whole profile the tree and shrub pollen proportions are only between 10 and 25%. This must not be proof of a very open landscape without or with only few woods because both most frequent herbaceous pollen types, Artemisia and Chenopodiaceae, which together always have records of about 75%, are very strong pollen producers, whereas Larix as the most frequent tree in the recent vegetation, is very poorly represented in the pollen record (Lang 1994) and has values of less than 2%. Pinus cembra type, a rather strong pollen producer, never exceeds 10%. Therefore, P. sibirica was never part of the local and regional vegetation. The changes of the AP/NAB relations indicate more woodland towards the middle Holocene, perhaps induced by more precipitation, and a—probably anthropogenic—woodland clearance afterwards. Regular occurrence of grains of the Cerealia type from this first decline of AP onwards (core depth 400 cm) let us suppose that the medieval Mongolians were not the first people to practise agriculture in this region, but that the origins of agriculture in the Mongolian Steppe must be of prehistoric age.

Fig. 3
figure 3

Pollen percentage diagram from Ugii Nuur; selected taxa, percentage calculation sum all taxa excluding water plants and spores

Full size image

Conclusions

The inhabitants of the medieval Mongolian capital Qara Qorum used many cultivated and to some extent also wild gathered plants as food. Most frequent and abundant were four cereal species, Panicum miliaceum, Hordeum vulgare, Triticum aestivum and Setaria italica, probably locally produced as summer crops in fertilized and irrigated fields. Oil and fibre plants as well as pulses were less important because they were partly substituted by meat. More than ten species of vegetables and spices indicate a high gastronomic level. Some of them were probably gathered in the wild, other perhaps grown locally in gardens, but most of them must have been imported. The same holds true for the about 20 species of fruits and nuts, most of which were also found in European towns, but rather rarely and more frequently only in later periods; they are regarded as indicators for a refined lifestyle (van Zeist 1991). Whereas in Europe imported plants came from a distance of normally not more than several hundred km or less, the production sites of the imported plants consumed in Qara Qorum were in China or central Asia to mention the most important areas; that is at distances of 1500–2000 km and more, separated from the consumer site by huge deserts. To surmount such obstacles and distances using horses or camels in order to provide better food requires a similar high state of organization as needed to move huge armies over distances of more than 6000 km which was also managed successfully by the Mongolians in the same period. The ancient town covered an area of more than 2 km2. The present excavation covers an area of about 1000 m2, about 0.5 per thousand of the total town area. Whether the results are representative for the whole town, if for example all inhabitants had such rich food, is uncertain. Differences between palace and town and inside the town will be discussed in a more developed state of the investigations.