The Terramare were settlement communities of the central Po Plain whose villages were enclosed by palisades, sometimes reinforced with embankments and moats. In early phases, roughly 1650 to 1350 BC, villages covered about one to two hectares of pile dwellings arranged in rows; in later phases, around 1350 to 1150 BC, some settlements expanded to as much as twenty hectares while others were abandoned. Excavations have revealed wooden poles and palisades, charred tree trunks at the base of the village of Montale, wooden planks at Pragatto, and the spectacular wooden votive tank at Noceto, all testifying to systematic and extensive use of timber for construction, expansion and defence.

The key evidence comes from pollen. In managed woodlands, coppicing influences pollen production, because cutting catkin-bearing branches temporarily reduces the pollen a tree sheds, with recovery during regrowth. The researchers analysed short-term fluctuations in arboreal taxa using pollen sums that excluded Quercus, so that the dominant oak signal would not mask variation in subordinate woody taxa. What emerged were gentle but rapid oscillations: peaks in deciduous oak pollen often coincided with synchronous declines in hornbeam, hazel, hop hornbeam, elm and ash, and vice versa. The independent, alternating behaviour of these curves points to cyclical fluctuations that cannot be explained solely by intrinsic variation in pollen productivity from mast-seeding, since masting-driven oscillations would be expected throughout the record rather than being confined to phases of Terramare occupation.

The pattern matches what is known from modern coppiced woodlands. Oak species such as Quercus robur and Q. petraea produce slower, light-demanding regrowth, whereas hornbeam and ash rapidly occupy canopy gaps after cutting, and hazel responds almost immediately with vigorous shoot production. This functional contrast corresponds to a three-stage cycle: a post-cutting phase dominated by fast-growing shrubs and light-tolerant trees such as hazel, ash and brambles; a mid-cycle phase of increasing competition and canopy closure involving hornbeam, maple and lime; and a late-cycle phase in which oaks gradually regain dominance. The inverse fluctuations observed in parts of the Middle and Recent Bronze Age sequences closely mirror these dynamics, suggesting that woodland exploitation followed a recurrent coppicing cycle structured by forest regeneration.

The interpretation is deliberately cautious. Pollen data are not generally regarded as direct evidence of woodland management, and pollen percentages are compositional, so closed-sum effects can distort apparent relationships between taxa. The authors argue, however, that the repeated cyclic patterns observed across multiple independent sequences, together with their archaeological context and ecological coherence, are most parsimoniously explained by selective coppicing as a recurrent management strategy. They read the pollen record as the cumulative expression of recurrent selective management over decadal to sub-centennial timescales, and note that the practice was probably an early, experimental form of coppicing that may only partially overlap with later silvicultural traditions. Selective cutting was probably applied to particular species chosen for wood properties and rapid resprouting capacity, rather than uniformly across the woodland.

The spatial and temporal resolution of the records supports this reading. Thirty-three samples spanning roughly 350 years of occupation, from about 1600 to 1250 BC, were collected from a stratified trench at Montale, and thirty-one samples covering about 380 years, from about 1550 to 1170 BC, came from the moat at Santa Rosa di Poviglio, giving an average resolution of roughly ten to twelve years between consecutive samples. Additional sequences came from wells at Poviglio, profiles at Baggiovara, Pragatto and Caorso, the necropolis of Casinalbo and the Noceto votive tank. At Poviglio, the alternating pattern is less visible in the lower part of the diagram, where only hop hornbeam peaks alternate with oak, but becomes clearly expressed in the central portion, with oscillations of oaks, hazel and hornbeam mirrored by hygrophilous taxa such as alder and elm; the signals weaken or disappear towards the top of the sequence.

That fading signal at the top of the record marks the system’s limits. During the Middle Bronze Age phases, woodland resources were initially exploited mainly for oak construction timber and subsequently managed through coppicing as a subsistence strategy supporting settlement expansion. In the late Recent Bronze Age, however, the regular rhythm of the practice declined, suggesting that coppicing could no longer be maintained under the combined pressure of resource depletion and environmental-climatic stress. Regional palaeoclimate records document a cooling phase in the fourteenth to thirteenth centuries BC followed by renewed warming and increasing aridity, culminating in a pronounced drought peak around 1200 BC, corresponding to the 3.2 ka event, when aridity proxies indicate drier conditions between 1200 and 1150 BC.

The pressures were visible on the ground long before the collapse. At Santa Rosa di Poviglio, sediment cores record a marked decline in arboreal pollen to below thirty percent, the lowest forest cover of the Holocene sequence at the site, and forest cover in the moat fell from about forty-five percent in the Middle Bronze Age to about twenty percent in the Recent Bronze Age, with deforested zones expanding from roughly 500 metres to nearly four kilometres around the village. A broader regional signal of forest reduction coinciding with Terramare expansion appears in the Adriatic marine core RF93-30, which contains Po River sediments transported several hundred kilometres downstream, indicating a basin-scale impact of woodland loss. Radiocarbon dates on charcoal add a telling detail: in Middle Bronze Age contexts the dates are systematically older because of the old wood effect, meaning forests still contained centuries-old trees, whereas by the Recent Bronze Age charcoal and collagen dates coincide, showing those old trees had disappeared and been replaced by younger, regenerating secondary woodland.

Woodland management was one strand of a wider agro-silvopastoral system. Pollen and macroremain evidence indicate cereal cultivation within an area of influence of roughly fourteen square kilometres around each village, with about twenty-five to thirty-five percent of that land open and largely free of woody vegetation. Emmer, einkorn, common and durum wheat, barley, broomcorn millet, spelt, oat and rye appear among the macroremains. Pollen curves of cereals, legumes and Cichorieae, a plant group indicative of pasture, suggest practices including crop-pasture alternation, fire farming, crop rotation and field manuring, with largely synchronous trends in cereals and legumes pointing to rotation or intercropping of food and fodder plants. A principal component analysis of 175 pollen samples across the sites separated samples by site rather than by chronology, indicating that local environmental conditions played a stronger role than temporal variation in structuring land use. Woodlands also supplied wild foods: acorns, berries and nuts, with Cornus mas particularly important in earlier phases, and wild grapevines growing near Poviglio, where morphobiometric analysis suggests about seventy percent of grape pips at Montale belong to the domesticated subspecies.

The authors argue that these findings challenge the widespread assumption that many temperate lowland forests represent natural baselines. Instead, the structure, composition and regeneration dynamics of such forests may carry the deep ecological legacies of millennia of human management, with regeneration pathways and species relationships established in the Bronze Age potentially still influencing temperate woodlands today. The Terramare maintained oak woodlands through coppicing for several centuries, and the late Bronze Age collapse of woodland and arable productivity, after that long-maintained equilibrium, illustrates how even finely tuned land-use strategies can falter when resource demand and environmental stress converge and pressure on woodlands surpasses the regenerative limits of the landscape.

Subject of Research: Pollen evidence of selective coppicing and forest regeneration by the Bronze Age Terramare culture in northern Italy

Article Title: Legacy effects of selective coppicing on forest regeneration in Bronze Age Italy

Article References: Mercuri, A. M., Florenzano, A., Clò, E., Zappa, J., Bosi, G., Rottoli, M., Zerboni, A., Cardarelli, A., & Cremaschi, M. (2026). Legacy effects of selective coppicing on forest regeneration in Bronze Age Italy. Nature Plants. https://doi.org/10.1038/s41477-026-02425-2

Image Credits: AI Generated

DOI: 10.1038/s41477-026-02425-2

Keywords: Bronze Age, Terramare, coppicing, pollen analysis, forest management, Po Plain, woodland regeneration, palaeoecology, agro-silvopastoral system, archaeobotany, sustainability, Nature Plants