Tree rings reveal signs of Europe's sustainable forest management long before the first historical evidence

The study investigates whether dendrochronological signatures in oak tree rings can reveal historical practices of coppice-with-standards (CWS) forest management and whether such practices predate their first mentions in historical records. CWS is a two-tier silvicultural system combining an understorey of coppice harvested on short rotations for fuelwood with widely spaced overstorey standards grown for construction timber. While documentary evidence of CWS in Germany begins in the thirteenth century, the actual onset and prevalence of CWS earlier in the medieval period remain uncertain. By characterizing tree-ring growth releases in modern, actively managed CWS stands and scanning historical and archaeological oak samples for similar cyclical patterns, the study aims to provide empirical evidence of past CWS management, with implications for forest history, resource use, and interpretation of paleoclimate records.

Prior work shows CWS was widespread in medieval and post-medieval Europe and declined with the shift from fuelwood to fossil fuels in the nineteenth century. Coppice and CWS systems have been studied from historical, geographical, anthracological, and dendrochronological perspectives, but most dendroarchaeology has focused on coppice rather than CWS, and prior CWS dendro studies largely examined abandoned stands. Historical sources indicate varied rotation cycles (5–40 years) and evolving management aims; terms and practices differed regionally (e.g., taillis sous futaie, Mittelwald). Evidence of coppice dates back to prehistoric and Roman times, but explicit CWS documentation appears from the High Middle Ages onward, leaving earlier periods poorly documented and motivating dendro-based inference of management.

Study regions and datasets: Modern CWS datasets were collected from two forests near Uffenheim, northern Bavaria: Weigenheim (WEIG; 220 ha; documented rotation ~28 years) and Welbhausen (WELB; 170 ha; rotation ~34 years; sources from 1447 and 1929). A total of 161 oak (Quercus sp.) standards were sampled: WEIG (36 discs; felled winter 2017/2018), WELB1 (73) and WELB2 (52). Stem discs were cut close to lower crown height using a random sampling approach to avoid reducing merchantable value. Historical datasets comprise 2120 mature oak samples (≥100 rings) from archaeological and architectural timbers dated 300–2015 CE: Bavaria (995) and north-eastern France (1125). Reference groups included a modern regional sawmill dataset (443 oaks from Bad Mergentheim and Wittighausen, NE Baden-Württemberg) and a prehistoric collection (115 oaks) from central Europe dated to the 8th–6th millennium BCE. Sample preparation and measurement: Samples were prepared using standard dendrochronological methods. Tree-ring width (TRW) was measured to 0.01 mm with binocular microscopes and semi-automatic measuring tables. Crossdating used PAST software. Release detection: Understorey coppicing in CWS is expected to produce sustained growth releases in standards due to reduced competition. The study adapted the growth averaging (GA) method of Nowacki and Abrams, computing percentage growth change (%GC) from consecutive 10-year means: %GC = [(M2 − M1)/M1] × 100, where M1 is the preceding 10-year mean including the target year and M2 the subsequent 10-year mean. Major releases were defined as %GC > 50%, moderate as %GC > 20% and < 50%. For multi-year sequences, the year of maximum %GC was taken as the release date. Modern CWS TRW series were first analyzed at tree level to characterize release frequency and intervals. Historical and archaeological TRW series were then scanned for CWS-like signatures: at least three consecutive release events with an average interval of 26–36 years and a standard deviation < 5 years, targeting cycle lengths broadly consistent with modern CWS (mean ~31 years; SD ~10 years) and capturing cycles between approximately 21 and 41 years. Only samples with ≥100 rings were considered to allow detection of ≥3 cycles. Analytical notes: Mean TRW chronologies were also constructed and age-stratified for modern stands, but periodic stand-level releases were not prominent, reinforcing the single-tree approach. The study evaluated potential confounders (e.g., insect outbreaks, storm events, hydroclimatic extremes) by comparing the timing of releases to historical records and considering expected synchrony across trees.

- Modern CWS standards (n=161) each showed characteristic growth releases, up to nine per tree (mean 4.23), with an average interval of 31.2 years (SD 10.1). Release intensity generally declined with age, and few releases occurred in the last ~20 years prior to felling. - Historical Bavaria dataset (n=995): 12 samples (1.2%) met CWS criteria, averaging 3.2 releases with mean interval 30.9 years (SD 2.7). Seven of these date to the first millennium CE, predating the first historical mentions of CWS. - Historical north-eastern France dataset (n=1125): 20 samples (1.8%) met CWS criteria, averaging 3.2 releases with mean interval 30.1 years (SD 2.7). Three samples predate documentary evidence (6th century, early 10th century, late 12th century). Several originate from early modern to recent periods. - Modern reference group (n=443): 4 samples (0.9%) met CWS criteria, each with three releases; mean interval 31.1 years (SD 2.3), indicating possible inclusion of CWS-origin material. - Prehistoric reference group (n=115): No samples exhibited cyclical suppression-release patterns, supporting method specificity and low susceptibility to random release detection. - Potential confounders: Periodic insect infestations were considered unlikely due to shorter intervals; storm records since 1800 showed no alignment with release years. Some clustering in 1920–1930 may reflect management or broader conditions rather than synchronous climate extremes. - The detection probability in historical datasets (1.2–1.8%) exceeded chance expectation (~0.7%), despite strict criteria.

The study demonstrates that cyclical growth-release signatures in modern, actively managed CWS oak standards occur at approximately 30-year intervals, providing a diagnostic pattern to identify CWS-origin timbers. Applying these criteria to historical and archaeological oak TRW series revealed CWS-like growth patterns throughout the medieval period in Bavaria and north-eastern France, including several cases from the first millennium CE. This indicates that CWS-type management was practiced long before its first documentary mentions in the thirteenth century. The single-tree analytical approach is appropriate given the lack of stand-level context for historical timbers and the observed variability in individual tree responses and rotation timing. While only a small fraction of historical samples met strict detection criteria, this proportion is above chance and likely conservative, reflecting methodological safeguards against overinterpreting sporadic releases. The findings also suggest that management-induced growth patterns exist in historical oak chronologies and should be considered in paleoclimate reconstructions, though their low prevalence implies limited overall impact when replication is high. Variability in cycle lengths, disturbances (e.g., wars), and evolving practices likely influenced the periodicity and clarity of release signals over time.

This work provides the first explicit extraction of a CWS dendrochronological signal from actively managed standards and applies it to historical and archaeological oak timbers, revealing evidence of CWS management throughout the medieval period and potentially as early as the first millennium CE—well before the first written records. The study advances understanding of historical forest management, resource use, and rural landscape organization. The approach should be routinely incorporated into dendroarchaeological analyses to identify past management practices. Future research should broaden geographic coverage, refine detection methods to better capture shorter rotation cycles, integrate independent historical and ecological evidence (e.g., by-product uses such as grazing and litter raking), and assess implications for climate reconstructions by quantifying management signals within larger datasets.

- Sampling height for modern discs was near lower crown height rather than breast height; ring counts do not reflect actual tree age, though TRW ratios are largely unaffected. - Historical and archaeological samples lack stand-level context and provenance; analyses necessarily focus on individual tree signals. - Strict scanning criteria (≥3 releases; average interval 26–36 years; SD < 5) likely underdetect true CWS cases, especially where cycles varied. - GA method has low sensitivity to short rotation cycles (<10 years), so medieval short-cycle coppice events may be missed. - Release intensity declines with tree age; recent decades in modern stands show fewer detectable releases, potentially reducing detectability in older standards. - Potential confounding disturbances (insects, storms, climate extremes) cannot be entirely excluded, though analyses suggest limited influence on the observed periodicity. - Geographic focus is limited to northern Bavaria and parts of NE France for historical datasets; results may not generalize to all of Europe.