Diverse prehistoric cattle husbandry strategies in the forests of Central Europe

The study investigates how early Neolithic farmers of the Linearbandkeramik (LBK) culture in Central Europe managed cattle within diverse forested landscapes during the sixth millennium BCE. While LBK expansion is often portrayed as culturally homogeneous, the varied environments they occupied—from marshlands and gallery forests to mixed oak woodlands—raise questions about regional adaptations in animal husbandry. Forests could provide both pasture and supplementary leafy hay, potentially improving herd health and milk production. Prior models emphasizing slash-and-burn clearances have been challenged by archaeobotanical evidence for intensive manured cultivation within forests. The research aims to reconstruct cattle diets and seasonal management to evaluate the extent and manner of forest resource use for herding and foddering across regions, using stable isotope proxies integrated with palaeoenvironmental data.

Historical and ethnographic evidence documents European woodlands as dual-use landscapes for shelter and animal fodder (leafy hay). Archaeobotanical studies have refuted the dominance of slash-and-burn in LBK subsistence, indicating intensive farming within forests. Previous isotope work suggested LBK cattle may have relied more on forest resources (lower δ13C in cattle bone collagen) compared to sheep/goats, and sequential enamel isotope analyses have proven valuable for seasonal dietary reconstructions. The canopy effect lowers plant δ13C under dense cover, which transmits to herbivore tissues, but equifinality challenges exist because marsh/waterlogged plants can show similar depletion. Compound-specific amino acid δ15N (β = Δ15NGlx-Phe) can distinguish woody versus herbaceous plant inputs, and dairy lipid δ13C from pottery provides high-resolution dietary signals during lactation. These frameworks inform the present multi-proxy approach to LBK cattle husbandry.

• Data compilation: 2,418 isotopic measurements from 61 LBK (and ALP for some sites) farming sites across Central Europe combining new and published data. Proxies include δ13C of cattle bone collagen (N=292), δ13C and δ18O of enamel bioapatite from sequentially sampled third molars (N teeth 46; N measurements 877; 7 sites), compound-specific δ15N of amino acids from sequential dentine samples (12 teeth), and δ13C of C16:0 fatty acids from pottery residues identified as dairy lipids (N=352 extracts; 23 sites for δ13C summary; total 44 sites tested for lipids). Species attribution for dairy fats is not isotopically resolvable (could be cattle, sheep, goat), but site-level comparisons and slaughter profiles support a dominant cattle signal.
• Laboratory protocols: Pottery lipids extracted by acid methanolysis and analyzed by GC, GC-MS, and GC-C-IRMS; dairy identified via Δ13C (C18:0−C16:0) ≤ −3.1‰. Bioapatite δ13C/δ18O measured on Kiel IV–Delta V IRMS after enamel pretreatment. Dentine amino acids derivatized to N-acetyl isopropyl esters; δ15N by GC-C-IRMS with internal standards and QA/QC.
• Diet inference: Converted tissue δ13C to diet using enrichment factors: Δcollagen–diet = −5.1‰; Δlipid–diet = +1.5‰ (based on Δlipid–collagen = −6.6‰); Δbioapatite–diet = −14.5‰. Deer (red/roe) bone collagen provided a forest-dwelling herbivore baseline δ13Cdiet = −27.7‰ (s.d. 1‰; N=35).
• Seasonality: Used δ18O cycles to infer seasonal timing; compared intra-tooth δ13C variations and dentine β values (dietary Δ15NGlx−Phe corrected by −4‰ trophic offset) to differentiate woody vs herbaceous inputs and assess winter leafy hay foddering. Adjusted β positions by ~6 months relative to enamel mineralization for integrated plots.
• Environmental context: Mean forest cover (MFC) extracted from interpolated Holocene reconstructions at 7500, 7250, 7000 cal BP (averaged). Palaeoclimate proxies (summer/winter temperature and precipitation) interpolated from pollen-based models (~7100±100 cal BP). Hydrology variables included river drainage systems, Strahler stream order, and distance to nearest river.
• Statistics and mapping: Site medians interpolated (IDW) to visualize δ13Cdiet distributions. Correlations (Pearson/Spearman) evaluated relationships among isotopic proxies, longitude/latitude, MFC, climate, and hydrology with multiple-testing checks (Bonferroni/FDR).

• Overall δ13C-diet ranges: Across all proxies, diet δ13C spanned −31.2‰ to −19.8‰ within C3 plant ecosystems. No significant difference among the three primary datasets (ANOVA, d.f.=2, P=0.1).
• Tissue-specific summaries (Table 1): • Bone collagen (N=292; 45 sites): raw avg −20.9‰; inferred diet avg −26.0‰ (range −29.1 to −22.1‰). • Dairy lipids C16:0 (N=352; 23 sites): raw avg −27.7‰; inferred diet avg −26.2‰ (range −31.2 to −21.1‰). • Bioapatite enamel (N teeth 46; 7 sites; N=877): raw avg −11.5‰; inferred diet avg −26.1‰ (range −30.1 to −18.8‰).
• Geography and forest cover: • MFC varied regionally (~85% in Poland, Czechia, Slovakia, S Germany to ~60% in the Paris Basin). • Significant negative correlation: MFC vs δ13Ccoll (r = −0.29, P = 4.6×10⁻⁷), consistent with increased forest-based feeding lowering δ13C in collagen. • Unexpected significant positive correlation: MFC vs δ13Cdairy (r = 0.50, P < 2.2×10⁻¹⁶), implying cattle rearing in forested regions even where open pasture existed and/or grazing in waterlogged environments with similarly low plant δ13C. • Strong longitudinal trend: δ13Cdiet correlated with longitude (r = 0.66, P < 2.2×10⁻¹⁶), with more 13C-depleted values to the west; deer baselines showed no significant correlation with longitude/MFC (small N).
• Comparison to forest herbivore baseline: Western sites’ δ13Cdiet (collagen and dairy) often fall within/below the deer-based forest baseline (−27.7‰ ± 1‰), supporting substantial forest resource use.
• Seasonality and foddering: • Bioapatite intra-tooth δ13C range (max–min) per tooth: 0.3‰ to 3.5‰; amplitudes significantly correlated with MFC (Pearson r = −0.70, P = 3.84×10⁻⁷), indicating diverse plant sources/seasonal variability especially in certain regions. • Several individuals (e.g., Bischoffsheim; one from Chotěbudice; APC2) had δ13Cbioap below a forest reference threshold (−13.2‰ enamel), particularly coinciding with low δ18O (winter), consistent with winter feeding on 13C-depleted plants. • CSIA-AA dentine β values distinguished plant types: Bischoffsheim individuals showed woody plant consumption; Balatonszárszó borderline; others (LUD1, APC1, BAL5) primarily herbaceous. No significant correlation of β with MFC (Spearman rho = −0.2, P = 0.6). • Integrated δ18O/δ13C with β confirmed winter leafy hay foddering at Bischoffsheim; BAL3 suggested similar seasonal foddering.
• Dairy production implications: Sites with foddering (e.g., Bischoffsheim, Cuiry-lès-Chaudardes) exhibited large intersite δ13Cdairy variability, potentially reflecting canopy density shifts during lactation, mixed species milking, and the influence of leafy hay on milk let-down and quality; such practices may have promoted out-of-season calving and extended milk availability.
• Climate effects: Palaeoclimate correlations indicated winter proxies (precipitation and temperature) had the strongest relationships with δ13Cdiet, suggesting climate-mediated pasture δ13C variability that could influence foddering strategies.

The integrated multi-isotope datasets demonstrate that LBK cattle husbandry was regionally diverse and frequently leveraged forest resources. Lower δ13C values in collagen and dairy lipids—especially in western regions relative to a deer forest baseline—indicate substantial forest grazing and/or consumption of leafy hay. Intra-tooth enamel δ13C combined with δ18O seasonality and dentine β values provide direct evidence for winter foddering with woody-derived leafy hay at specific sites (e.g., Bischoffsheim), while other sites show predominantly herbaceous inputs, revealing varied regional strategies. The positive correlation between MFC and δ13Cdairy suggests that, in forested regions, cattle still fed in contexts that can elevate δ13C in dairy (e.g., waterlogged habitats), highlighting equifinality and complex microhabitat use. The strong east–west gradient in δ13Cdiet and the significant associations with winter climate proxies imply broader environmental controls on pasture plant isotopic baselines and husbandry responses, such as adopting winter foddering to buffer forage scarcity and optimize milk production. Collectively, the findings address the central question by demonstrating that early farmers not only exploited forests for pasture and fodder but also adapted seasonally and regionally to landscape and climate variability to sustain cattle fertility, calving patterns, and dairy yields, with cascading effects on settlement dynamics and woodland transformation.

This study synthesizes bone, tooth enamel, dentine amino acid, and pottery lipid isotope evidence to reveal distinctive regional and seasonal LBK cattle management strategies across forested Central Europe. Key contributions include: (1) robust evidence for forest use in cattle diets, including winter leafy hay foddering in the northwest; (2) identification of significant geographic (east–west) and environmental (forest cover, hydrology, winter climate) structuring of δ13Cdiet; and (3) linkage of foddering practices to potential increases in fertility, out-of-season births, and enhanced milk availability, with implications for community resilience and forest ecosystem modification. Future research should refine temporal resolution of palaeoclimate and land-cover reconstructions, expand species-resolved lipid analyses, increase the number and geographic spread of sequential tooth and CSIA-AA datasets, and develop models to quantify the ecological impacts of herding on forest structure and dynamics.

• Equifinality: Similar δ13C depletions can arise from dense canopy shade or waterlogged/marsh environments, complicating unambiguous attribution of forest feeding.
• Forest cover proxy: Mean forest cover (MFC) reflects percentage cover, not canopy density/structure; localized woodland use may be underrepresented by interpolated reconstructions.
• Dataset biases: Collagen preservation varies by microregion; teeth availability limited (7 sites) due to recovery and excavation practices; dairy lipid datasets have uneven site distribution and recovery biases.
• Species attribution: Pottery dairy lipids cannot be species-differentiated (cattle vs sheep/goat), introducing potential mixing effects.
• Palaeoclimate proxies: Interpolated and time-averaged pollen-derived climate reconstructions may not capture short-term variability contemporaneous with site occupations.
• Comparative baseline: Deer dataset is relatively small, potentially limiting baseline robustness for forest herbivores.
• Temporal alignment: Assumptions in aligning enamel and dentine formation (e.g., 6-month offset) introduce uncertainty in seasonality reconstructions.