Expérimentation d'agriculture néolithique

Salavert A., Toulemonde F., Auray R., Hoerni C., Huirorel G., Lafarge I. (2022) ‒ Présentation de l’expérimentation d’agriculture de type néolithique menée à l’archéosite du parc de la Haute-Île (Neuilly-sur-Marne, France) : mise en œuvre et résultats des cultures céréalières/ Presentation of the Neolithic-type agriculture experiment of the « parc de la Haute-Île » (Neuilly-sur-Marne, France): implementation and first results from the cereal plots, Bulletin de la Société préhistorique française, 119, 1, p. 49‑76.

http://www.prehistoire.org/515_p_57280/bulletin-de-la-spf-2022-tome-119.html

L’expérimentation agricole menée dans le parc de la Haute-Île (Département de la Seine-Saint-Denis) pendant trois années agricoles (2017-2020) a pour objectif de tester un système agricole de type néolithique (5500-4900 avant notre ère) impliquant un travail du sol minimal et un taux de fertilisation faible. Cet article présente la mise en œuvre de l’expérimentation et une analyse préliminaire des résultats concernant les rendements céréaliers et les relevés des cortèges d’adventices. L’amidonnier et l’engrain se sont montrés généralement compétitifs face aux plantes adventices herbacées. Les parcelles semées à l’automne avec la technique du sillon ont permis de récolter en moyenne 1350 kg/ha (ratio 1:15) d’engrain et 900 kg/ha (ratio 1:8,5) d’amidonnier. Cette moyenne est à nuancer en raison de la variabilité interannuelle des rendements pour les deux céréales. Néanmoins, le rendement est toujours supérieur à 1:10 pour l’engrain et en général inférieur à 1:10 pour l’amidonnier. Ainsi, contrairement aux données agronomiques qui indiquent que l’engrain est moins productif que l’amidonnier, l’expérimentation indique que certaines variétés d’engrain peuvent être plus performantes que celles d’amidonnier en culture d’hiver, plusieurs années consécutives, sous un système agraire engageant un très faible apport organique et un faible investissement humain. La flore adventice relevée sur les parcelles est dominée par les plantes vivaces probablement en raison de la faible intensité du travail du sol et de la courte durée de l’expérimentation. La majorité des taxons trouvé dans le lot de stock d’engrain (récolte traitée par battage puis vannage) provient de la prairie environnante. Les adventices résiduelles y représentent seulement un quart de la diversité enregistrée sur les parcelles et sont majoritairement des annuelles d’hiver ce qui ne correspond pas au ratio vivace/annuelle enregistré sur les parcelles.

Mots clés : Néolithique, Europe tempérée, expérimentation, blé vêtu, système agricole, adventice, rendement céréalier.

This article presents implementation choices and initial results on crop yields and weed assemblages from a 3-year Neolithic-type agricultural experiment (2017-2020) at the “parc de la Haute-Île” (Department of Seine-Saint-Denis), about 20 km east of Paris (France). The project aimed to experiment with the currently accepted farming system for the Early Neolithic in central Temperate Europe (ca. 5500-4900 cal BC), i.e. a system of winter or spring cereal cultivation, conducted on small permanent plots, whose fertility is maintained by a low input of organic amendment.

At the “parc de la Haute-Île”, the intensity of work on plots was minimal (no weeding, no irrigation, low fertiliser input) due to constraints related to labor and time availabilities of the experimental team. The plots were mainly cultivated with the two emblematic hulled wheats of the Linearbandkeramik period (LBK): emmer (Triticum turgidum subsp. dicoccon) and einkorn (Triticum monococcum). The initial state of the experimental site is a meadow surrounded by a wooded edge and the initial soil has an inherent high fertility. The total area devoted to the farming experiment was 180 m² (31.5 m² per plot). Each year, three plots were cultivated either in monoculture or in maslin, i.e. cultivation of several cereal species on the same plot, subsequently harvested together. The annual cropping plans included one plot of cereals that had never been fertilized, another amended with animal manure (equivalent of 2t/ha of animal manure were introduced), and another following a pulse cultivation (rotation). The aim was to evaluate the cereal yields following the different methods used (fertilization, seasonality), as well as the qualitative diversity of weed assemblages in the plots and in one processed einkorn subsample, sown in autumn and gathered low on the stem.

The main questions underlying the experimental project are:

-Does low organic input cultivation without weed control benefit one or the other of the two-hulled wheats in the monoculture and maslin plots?

-Does soil fertilization have an effect on the yields of the hulled cereals in the short term?

-What is the composition of the wild flora present in the cultivated plots, what is its origin, is it representative of the cultivation methods?

-Does the weed flora found in the harvest lots after their treatment reflect that found on the agricultural plots?

At the "parc de la Haute-île", emmer and einkorn were generally competitive with herbaceous weeds in the farming system experimented that includes low labor input and low fertilization intensity. The yields for the two cereals show inter-annual but also intra-annual variabilities. Einkorn was the best performing cereal in both the monoculture and maslin plots. The average yield for einkorn is 1350 kg/ha (ratio 1:15) and 900 kg/ha (ratio 1:8,5) for emmer. The yield is always higher than 1:10 for einkorn regardless of the seasonality of sowing and fertilization methods. For emmer, the ratio is generally less than 1:10. Regarding sowing seasonality, the spring crop tested in year 1 (2018-2019) performed less well than the winter crop for both taxa, with emmer even stopping his development during the agricultural season. On the unfertilized plot, the evolution of yields between the test year (2017-2018) and year 2 (2019-2020) is discontinuous and the two cereals show contrasting behaviors. For einkorn, the plot cultivated without fertilization in the test year delivered the highest yield (1:23) of the three experimental years. On the five plots fertilized in monoculture, yields were not systematically higher than on the unfertilized plots. For emmer, in year 2, the two fertilized plots grown (1:12 and 1:17) yielded more than the unfertilized plot (1:7) and even more than the reference value of the test year (1:10). Emmer seems thus to respond somewhat better to short-term fertilization successions. These results contradict the current agronomic data that indicate a better productivity of emmer. The experiment of the “parc de la Haute-Île” allows us to qualify the hypothesis that the better performance of einkorn during episodes of heavy rainfall could explain its preponderance on most of the LBK archaeological sites in Central European despite its supposed lower yield. Indeed, the experiment shows that currently, some varieties of einkorn can outperform emmer, several years in a row, under a low input winter cropping system, on a soil with high inherent fertility.

A total of 36 herbaceous species were observed in the cereal plots. Most (n=25) were species identified in the meadow. At least 5 taxa may have originated in the meadow and/or initial seedlings and 2 taxa originated exclusively in the initial seedlings of the test year. The distribution by biological type shows 50% annuals (mostly winter annuals) and 50% perennials (without vegetative reproductive organs in majority). The high presence of perennials can be explained by the short duration of the experiment that did not favor the domination of annuals, and the low intensity of weeding. In the einkorn stock after threshing and winnowing, 10 taxa were identified to the species level. The distribution by biological type indicates 80% annuals, which are mostly winter annuals. Among the perennials, those without vegetative reproduction dominate. The ratio of annuals/perennials observed in the einkorn stock is thus not representative of what was observed on the plot. The assemblage diversity is reduced compared to the one recorded in the plots. Furthermore, winter annual weeds are over-represented compared to the ratio recorded in the plots, where perennials dominate. This could be explained by the fact that annuals, such as poppy (Papaver rhoeas) or brome (Bromus sp.), produce more seeds than perennials and are therefore more likely to be present in the samples. This result will have to be verified and explained when the entire harvest subsamples of the three experimental years are processed.

Keywords: Neolithic, temperate Europe, experimentation, hulled wheat, farming system, weed, cereal yield.