Climate at the onset of western Mediterranean agriculture expansion: Evidence from stable isotopes of sub-fossil oak tree rings in Spain

Abstract

Climate conditions during the early Holocene may have contributed to the spread of Old World agriculture from its area of birth, in the Fertile Crescent (Near East), west through the Mediterranean Basin. Reconstructing the specific environmental conditions existing in early agricultural settlements of the western Mediterranean may help to elucidate this point. The aim of this work was to gain information on past climate of one of the earliest agricultural settlements of the Iberian Peninsula, La Draga, a lacustrine site dating back to the 2nd half of the 8th millennium BP, in which post fragments of deciduous oak have been exceptionally well preserved in an anaerobic environment (sub-fossil wood). We studied the relationship between climate factors and carbon isotope discrimination (Δ¹³C) and oxygen isotope composition (δ¹⁸O) in wood α-cellulose from modern tree-ring series of Quercus humilis Mill. and Quercus petraea (Matt) Liebl. recovered from the area. Climatic responses observed in extant material were used to interpret the isotopic signal of wood α-cellulose extracted from sub-fossil wood. Results showed that water availability of late spring, early summer and September influences Δ¹³C and δ¹⁸O, which allows their use as proxies for palaeoclimatic reconstruction. Differences between sub-fossil and extant samples in Δ¹³C (19.35 vs. 18.02‰) and δ¹⁸O (26.32 vs. 29.28‰) records suggest slightly lower temperatures and higher plant water availability than at present during the establishment of agriculture at the site. These results seem to disprove the existence of an environmental limitation preventing continuation of hunting and gathering activities as a cause for the adoption of agriculture in this early agricultural site on the Iberian Peninsula.

Introduction

Old World agriculture emerged in the Near East about 10,500 years ago and spread in the subsequent millennia through Eurasia, reaching the western shores of Europe by the 8th millennium BP. The spread of agriculture during the Early Holocene seems to be the result of cultural and demographic diffusion of the agricultural package (Bánffy, 2000, Bosch et al., 2000, Cavalli-Sforza, 1996, Gronenborn, 1998), which includes cultivation practices and seeds established at the time agriculture was first adopted. Nevertheless, environmental factors may have also played a role in speeding this process, at least in the Mediterranean Basin, where aridity was already present at the beginning of the Holocene (Sherratt, 2007). Thus, the earliest agricultural sites all have in common their “oasis” pattern; that is, a set of spatially limited but highly productive environments where rivers or springs began to accumulate small alluvial fans that allowed “floodwater farming” (Hillman et al., 1989, Sherratt, 2007). Apparently this was the pattern for much of the early extension of cultivation systems during the following millennia. Thus, the spread of agriculture seems to have involved the replication of the environmental conditions where cultivation emerged, with sites tied to particular pockets of well-watered terrains (Sherratt, 2007, Van Andel et al., 1995). Nevertheless the archaeological record does not at the moment clearly support or counter the association of early cultivation sites with climatic stress.

The site of La Draga, an early Neolithic lakeside settlement dating back to 5300–5150 cal. BCE located in the northeastern Iberian Peninsula, provides the first evidence of the adoption of agriculture in Catalonia and constitutes one of the first examples in the western Mediterranean Basin (Bosch et al., 2000, Buxó, 2007, Buxó et al., 1997). Cultivation of cereals, mainly wheat (Triticum aestivum/durum)and barley (Hordeum vulgare L.), and legumes (Vicia faba L. and Pisum sativum L.), has been reported at this site (Buxó et al., 2000). This is a typical early agricultural settlement in the sense that it is placed in the shores of a seasonally enlarged lake basin that might have allowed floodwater farming (Bosch et al., 2000). According to pollen and anthracological studies (Pérez-Obiol, 2007, Piqué, 2000, Roberts et al., 2008), vegetation around the lake at that period was formed by a riparian forest near the shore line (with presence of Corylus, Alnus, Fraxinus and Populus species) and, further inland, a deciduous oak-dominated (Quercus spp.) hardwood community. This contrasts with the present situation, in which a mixed forest of evergreen and deciduous oak is dominant in the area (Vilar et al., 1991). Such a wide extension of temperate species may be indicative of the existence of a sub-humid Mediterranean climate prevailing in the northwestern Mediterranean during the Early Holocene (Jalut et al., 2000, Vernet and Thiebault, 1987). Based on pollen studies, a likely decrease of temperature and an increase in precipitation at the time of La Draga's settlement might have caused fir forests to spread over a much lower altitude than usual during the Holocene (Burjachs and Allué, 2003). Afterwards, the Mediterranean climate began its expansion, with evergreen oaks dominating over deciduous species owing to the impacts on vegetation of high temperatures and relatively low precipitation in summer (Aguilera et al., 2009, Burjachs and Allué, 2003, Jalut et al., 2000). Nevertheless, pollen data in La Draga must be interpreted with caution due to the karstic origin of the lake, which causes pollen spectra to reflect a potentially very large area including different altitudinal stages (Jalut et al., 2000). On the other hand, anthracological studies do not only reflect vegetation changes, but also human preferences in the use of certain species (Piqué, 2000). Complementary palaeoenvironmental proxies at the local scale are therefore needed to describe more precisely the climatic conditions prevailing at the beginning of agriculture in this area.

The interest on Holocene palaeoenvironmental reconstruction has steadily increased over recent decades, and several approaches including, among others, lake sediment (Colman et al., 1995, Vegas et al., 2001), palynological (Jalut et al., 2000, Pérez-Obiol and Julià, 1994), hydrological (Roberts et al., 2008) and dendrochronological records (Hunter et al., 2006, Loader et al., 2008) have been successfully used for this purpose. Tree rings in particular provide high potential for past climate reconstructions (Briffa et al., 1990, Tardif et al., 2003), as long time series from bog oaks (Quercus robur L. and Q. petraea (Matt) Liebl.) and pines (e.g. Pinus sylvestris L.) are available from Central Europe (Friedrich et al., 2004, Leuschner et al., 2002). However, the potential information present in tree-ring width data is very sensitive to the high variability often found among time series, both within- and between-sites (Loader et al., 2003, Sass-Klaassen et al., 2005). In recent years, a growing number of studies have used carbon and oxygen stable isotope records as tools for palaeoclimatic inference (Ferrio et al., 2006, McCarroll and Pawellek, 2001, Robertson et al., 1997). Stable isotopes usually require fewer tree-ring time series than ring width data to recover consistent climatic signals (McCarroll and Loader, 2004). Their implementation and interpretation is facilitated by an increasing knowledge of external (e.g. water precipitation isotopic composition) and internal (e.g. stomatal conductance and assimilation rate) factors driving changes in isotopic signatures of plants (Barbour et al., 2001, Farquhar and Lloyd, 1993, Roden and Ehleringer, 2000). These factors are in turn affected by environmental variables such as radiation, air temperature, relative humidity, vapour pressure deficit and precipitation (Ferrio and Voltas, 2005, McCarroll and Loader, 2004, Warren et al., 2001).

Sub-fossil wood allows extension of living tree chronologies back in time and is therefore routinely used in dendrochronological studies. However, changes in the relative abundance of the different wood fractions following decomposition of organic material may produce significant shifts in isotopic signatures. In particular, holocellulose is preferentially degraded relative to lignin, and this chemical taphonomy could modify the carbon isotope composition of buried sub-fossil wood (Schleser et al., 1999). Of the two holocellulose components, cellulose is a more stable polysaccharide than hemicellulose (Richter, 2006) and, as a result, α-cellulose is usually targeted for δ¹³C and δ¹⁸O analyses when reconstructing past climatic and ecological conditions.

This study aims to provide new insights on the long-standing debate about whether adoption of agriculture in the western Mediterranean Basin was mediated by environmental constraints, or if it originated as a consequence of a demic process of diffusion or even a much faster maritime pioneer colonization (Zilhão, 2001). Nevertheless, these are not mutually exclusive alternatives. The hunter–gatherer and farming economies in fact shared a common objective: a guaranteed food supply in a changing natural and social environment (Zvelebil, 1986). When faced with the dilemma of securing a continuous food supply, hunter–gatherers could have been impelled to develop or, alternatively, to adopt technological, economic and social means coping with a periodic absence of resources (Zvelebil, 1986). Adoption of agriculture and husbandry may have played a central role in such adaptive strategy (Fuller et al., 2010). The needs for extra labour expenditure on crop-processing and soil fertility maintenance associated with agricultural practices would be compensated with linked gains in terms of potential crop yields and stability of food production (Araus et al., 2001, Fuller et al., 2010).

Data generated in this study can assist in determining the reasons for the adoption of agriculture; in particular, whether or not the rationale for agriculture was directly driven by climatic stresses. The latter case would represent a more complex scenario as, even if cultural influx was a basic component of agriculture adoption, securing food resources from climate vagaries could have also played its own role in this process. To this end, climate conditions were inferred for one of the earliest agricultural settlements of the Iberian Peninsula (La Draga) dating back to the 2nd half of the 8th millennium BP. Firstly, relationships between present climate (including variables such as temperature, precipitation and evapotranspiration) and stable isotope signatures (carbon isotope discrimination, Δ¹³C, and oxygen isotope composition, δ¹⁸O) of wood α-cellulose of extant Quercus petraea and Quercus humilis growing nearby the settlement were determined. Afterward, relevant climatic factors driving changes in Δ¹³C and δ¹⁸O records were used to interpret the isotopic signal recovered in architectural elements from La Draga, a lakeside settlement in which post fragments of deciduous oak have been exceptionally well preserved in an anaerobic environment (sub-fossil wood). The usefulness of stable isotopes as proxies for palaeoenvironmental reconstruction in the Early Holocene and the putative role of climate facilitating the adoption of agriculture in the western Mediterranean Basin are discussed.