"Slash and burn" or "weed and manure"? A modelling approach to explore hypotheses of late Neolithic crop cultivation in pre-alpine wetland sites

The record of prehistoric crop cultivation in central Europe dates as far back as 5500 BC. In the piledwellings of the north-western pre-alpine forelands, dating roughly from 4300 to 800 BC, favourable taphonomic conditions provide evidence for the ways of cereal cultivation and consumption in unmatched detail. Based on different sets of (bio-) archaeological and palynological evidence, different hypotheses of crop husbandry methods have been developed for the wetland settlements. During the late Neolithic, two partly antithetic ideas are discussed: On one hand Shifting Cultivation assumes frequently shifted crop fields and the use of fire to provide nitrogen for plant uptake; On the other hand Permanent Cultivation reconstructs longer-ranging use of the fields, to which nitrogen may have been provided by various means. From the Bronze Age onwards, most probably some form of extensive ard cultivation was applied. In this article, we explore the implications of the different hypotheses for the socio-ecological system of the wetland sites. We combine the capability of agent-based modelling to simulate dynamic processes with the benefits provided by geographical information systems and the possibilities provided by the use of modern agro-ecosystem modelling tools. First, we used a mechanistic crop growth model, MONICA, to evaluate the influence of important factors of prehistoric crop yield formation: the climatic conditions, the soil texture and the degree of nitrogen availability. Second, we applied an agent-based model (WELASSIMO_crops) to simulate the spatial and economic implications related to the different crop husbandry methods. Our results provide quantitative information on the extent of crop husbandry activities in the wetland sites and on the effect of natural and anthropogenic factors on prehistoric crop yields. Without manure application, initial average yields of 1.0 t ha-1 a-1 are shown to decrease rapidly to only 50 %after 10 years. A manuring rate of 10 t ha-1 a-1 allows for higher yields of 1.7 t ha-1 a-1 and a slower rate of fertility decrease, but requires high numbers of livestock per capita. In shifting cultivation, high yields of 2.7 t ha-1 a-1 are reasonable, while necessitating a very large area and high labour input. Using the model results and a case study, we argue that permanent cultivation is more likely to have been the standard method, while burning of the landscape may have had different objectives than crop husbandry. We find that the combination of agent-based social and process-based biophysical modelling is a powerful tool to study the complex interdependencies in human-environment systems in the past.