It has been proposed that agricultural expansion during the Roman Classical Period (RCP) was facilitated by a more humid climatic regime in the Mediterranean region compared with present. Certain proxy reconstructions of precipitation indicate that climate in the Mediterranean during the RCP was anomalously wet in the context of the late Holocene possibly owing to large-scale climatic processes. Independent of an anomalous wet signal, it is proposed that prior to anthropogenic deforestation greater vegetative cover maintained a more humid climate than present owing to increased evaporative fluxes between the land and atmosphere.

Here, we take a multidisciplinary approach to investigate these alternative hypotheses for a wet RCP. We present a composite analysis of precipitation proxies for the Mediterranean region that reveals weak evidence for a regionally consistent wet signal during the RCP. Using an Earth System Model forced with recently published reconstructions of RCP land use change we find that, at a regional scale, the Mediterranean climate was insensitive to prescribed changes in landcover. We supplement these findings with inferences on changes in water availability between the RCP and present based on the distribution of Bronze and Iron Age archaeological sites in the Fertile Crescent and North-East Africa as well as classical writings.

Although there is little evidence for a regionally consistent wet signal during the RCP, we do find evidence of a local wet signal in proxy reconstructions from the Fertile Crescent. Modern empirical studies present convincing evidence for increased early wet season convective rainfall associated with an expansion of irrigation in this region. Using a modelling approach we demonstrate that an expansion of irrigated agriculture during the RCP could indeed cause an increase in precipitation, providing a possible explanation for the anomalous wet signal exhibited in these proxies.