1 Introduction

With a long tradition in the natural and social sciences (Ericson, 1985), strontium (Sr) isotopic ratios (⁸⁷Sr/⁸⁶Sr) have been widely applied in historical (Ryan et al., 2021; Van Ham-Meert & Daly, 2023; Vobejda et al., 2024), archeological (English et al., 2001; Hajj et al., 2017; Ostapkowicz et al., 2017; Pinta et al., 2021; Rich et al., 2012, 2016), and environmental (Hole et al., 2022) studies to identify the source and transport of wood; in ecological studies to establish nutrient cycles (Reynolds et al., 2012); in forensic studies to detect illegal logging (Aguzzoni et al., 2025) and trace the origin of agricultural products (He et al., 2025); and in societal studies to reconstruct past human and animal mobility, origins and diet (Bentley, 2006; Britton et al., 2022; Crowley & Godfrey, 2019; Depaermentier, 2023; Depaermentier et al., 2020; Guiserix et al., 2022; Heddell-Stevens et al., 2024; Holt et al., 2021; Montgomery, 2010; Neil et al., 2020; Thomsen & Andreasen, 2019; Wang & Tang, 2020; Wang et al., 2023; Wooller et al., 2021). Often applied in combination with other proxy and/or model evidence, the assumptions of the Sr fingerprinting approach are that (a) abundant Sr from the underlying bedrock is incorporated into a tree's annual growth rings via its root system rather than leaves (Evans et al., 2010), and (b) any isotopic fractionation occurring during biological uptake and ion exchange is negligible (Capo et al., 1998; Johnson et al., 2022). However, the validity of these assumptions, especially the first, has been questioned (Erban Kochergina et al., 2021; D’Andrea et al., 2023; Van Ham-Meert & Daly, 2023).

Among the recognized challenges are post-depositional Sr isotope fractionation processes that might, for instance, accompany the preservation and waterlogging of timbers or the effects of marine inundation (Domínguez-Delmás et al., 2020; Snoeck et al., 2021), as well as the need for robust statistical analyses of large data sets (Drake et al., 2014). Moreover, long-range aeolian dust transport can shift bioavailable ⁸⁷Sr/⁸⁶Sr away from the underlying lithology (Boethius et al., 2022; Hartman & Richards, 2014). Another constraint is the degree to which wood Sr isotopic signatures might be biased toward the atmospheric Sr signal in precipitation (Drouet et al., 2005; Erban Kochergina et al., 2021; Montgomery, 2010). Meteoric water contains a range of soluble ions, including Sr, sourced from sea spray (Göhring et al., 2023), continental dust and anthropogenic aerosols (Négrel et al., 2007), which can be incorporated by vegetation from soil water.

Despite recognition of the limitations of Sr isotopic fingerprinting (see Depaermentier et al., 2025 for a review on the paleoenvironmental potential of bioarcheological isotope data), and the contributions of both geological and atmospheric inputs on bioavailable Sr (Capo et al., 1998; Graustein & Armstrong, 1983; Naiman et al., 2000), there have been few investigations of different wood samples from different tree species at different forest sites.

Here, we compare the Sr isotopic signature in large-scale precipitation with local to regional signatures of modeled Sr trace elements in bedrock against 24 newly measured Sr signatures in soil and forest floor samples, as well as 120 newly measured Sr signatures in different wood samples from five tree species at four contrasting forest sites in the Czech Republic. Our aim is to assess the relative influences of bedrock geology and precipitation on the Sr isotopic composition in wood.

2 Materials and Methods

All four sampling sites are located in mixed forest habitats between 250 and 630 m above sea level (asl) in the central and western part of the Czech Republic (Figure 1). While sites have different bedrock geology, their climatology is broadly similar (Figure S1 in Supporting Information S1), influenced by prevailing westerlies from the Atlantic Ocean, especially between autumn and spring (Brázdil et al., 2022). In contrast, high-pressure cells over the eastern European landmass and anti-cyclonic blocking promote warm summers and cold winters during which precipitation totals are relatively low.

Doupov is the coldest and wettest site, and Hradec Králové and Svatoslav are both the warmest and driest sites. Annual temperature means range from 6.9 to 9.1°C and annual precipitation totals range from 490 to 610 mm (based on E-OBS v30.0e; updated from Cornes et al., 2018). This is consistent with nearby station recordings from the Czech Meteorological Service (6.2–9.1°C and 610–670 mm). The coldest and warmest months are January and July (−2.5 and 18.9°C). The lowest and highest monthly precipitation totals are recorded for February and July (circa 22 and 110 mm).

Our four sampling sites are characterized by different lithologies of volcanic, plutonic, metamorphic, and sedimentary rocks of varying ages (Figure 1). We applied a mechanistic model (Bataille & Bowen, 2012; Bataille et al., 2014), using a geochemical database available for the Czech Republic to estimate Sr concentrations for the Cenozoic basalt in Doupov, the Mesozoic sediment in Hradec Kálové, the Variscan granite in Svatoslav, and Precambrian gneiss in Javornik. For igneous and metamorphic rocks, we assumed that ⁸⁷Sr/⁸⁶Sr ratios reached approximately 0.7013 around 3 Ga ago with subsequent independent evolution of ⁸⁷Sr/⁸⁶Sr in the mantle and crustal reservoirs, reflecting their distinct Rb/Sr ratios, and decay of ⁸⁷Rb (decay constant, λ = 1.42 × 10⁻¹¹ yr⁻¹). For all silicate rocks, we considered knowledge of the geochemistry and available ⁸⁷Sr/⁸⁶Sr values from the Czech Republic. Of the total 69 lithologies, 21 and 15 classes are igneous and metamorphic rocks (excluding marbles), respectively. The ⁸⁷Sr/⁸⁶Sr isotope values from the literature cover half of these 36 classes. For the other half, we estimated ⁸⁷Rb/⁸⁶Sr_parent based on lithology as follows. The Rb/Sr_parent of terrestrial sedimentary rocks was estimated by assuming a uniform source with a constant value of 0.35, corresponding to the average Rb/Sr_parent of the silicate rocks of the Bohemian Massif. For Cenozoic and Mesozoic sediments with mixed terrestrial and marine origins, we estimated ⁸⁷Sr/⁸⁶Sr based on the CaO abundance. Our model-based estimates of bedrock Sr concentrations also form a basis for possible high-resolution isoscape mapping across the Czech Republic and beyond.

For each calendar date, parameters for t₂, (Rb/Sr)_parent, (Rb/Sr)_rock and the Sr concentrations in meteoric water and rocks were defined by their respective medians and standard deviations, which were transformed from coefficients of variation where necessary. A total of 10,000 simulations was conducted for each entry to propagate uncertainties. For each simulation, Sr isotope ratios and F_prec were calculated based on the generated stochastic samples. Results report the average, standard deviation, and 95% confidence range of the bedrock-specific Sr isotope ratios.

The value of atmospheric ⁸⁷Sr/⁸⁶Sr input was taken as the average of 55 published ⁸⁷Sr/⁸⁶Sr values in open area precipitation in the Czech Republic measured between 1990 and 2021. The span, 0.7094–0.7164, with a mean/median and standard deviation of 0.7118/0.7113 and 0.0020, respectively, indicate broad spatial and seasonal coherency. Seventy-four throughfall precipitation (62 runoff) ⁸⁷Sr/⁸⁶Sr values from the Czech Republic with means/medians of 0.7125/0.7118 (0.7187/0.7187) corroborate the reliability of the precipitation Sr signature.

At each sampling site, we collected and analysed soil samples from three separate soil pits, representing the upper forest floor horizon and the underlying first 40 cm of mineral soil. While the highest rate of nutrient recycling from litter occurs at the forest floor, the ⁸⁷Sr/⁸⁶Sr values of the lower mineral soil are likely affected by the abundance of Sr in the bedrock and its isotopic composition. Nutrient uptake of all four tree species analysed in this study is restricted to the upper forest floor horizon and the underlying first 40 cm of the mineral soil (Úradníček et al., 2010). The resulting 24 soil samples were air dried and sieved prior to isotopic analysis (see details below). We also drilled 5 mm increment cores at circa 1.3 m stem height from three mature oak (Quercus spp.), beech (Fagus sylvatica), spruce (Picea abies), pine (Pinus sylvestris), and larch (Larix decidua Mill.) trees that were growing near each other on similar site-specific bedrock (Table S1 in Supporting Information S1). The individual tree ages ranged from 39 to 122 years (with no substantial differences between sites and species). All trees had well-developed root systems, and none showed any sign of abiotic disturbance, including insect defoliation. While spruce and pine often develop shallow roots, oak, beech and larch form deep roots. Standard dendrochronological techniques were applied for sample extraction, preparation, ring width measurement, cross-dating, and chronology development. Each core was split into an inner and outermost section of juvenile and mature wood (i.e., young and old cambial ages), respectively. While the two halves include heartwood and sapwood (i.e., the inner- and outermost parts of the stem), beech and spruce lack visually distinct heartwood/sapwood proportions.

All 24 soil and 120 wood samples were analysed for bulk radiogenic isotopic composition and Sr concentration at the Czech Geological Survey facility. Organic forest floor and soil samples of circa 100 mg were burned at 800°C for 16 hr before silicate rock decomposition with HF–HNO₃–HClO₄–HCl solution (Erban Kochergina et al., 2021, 2022). The soil phase was isolated by using the first step of the BCR sequential extraction procedure (Kochergina et al., 2017; Sutherland & Tack, 2002), and the resulting samples of 1 ± 0.01 g were leached in 40 ml of 0.11 M acetic acid for 16 hr. Wood samples were immersed in de-ionized water and sonicated repeatedly to remove dust and other contaminants. Samples were then dried at room temperature and burned at 800°C for 16 hr. While the dried wood samples weighed 0.4–0.7 g, the ash mass ranged from 0.006 to 0.020 g. The ash was dissolved in 3 ml of concentrated HCl with a few drops of HClO₄ at 140°C for 48 hr, before the solution was evaporated to dryness. Three repeated dry-down steps were followed using 0.5 ml of concentrated HNO₃.

The final residue was completely dissolved in 4 ml of 6 M HCl. Sr was separated using Sr-spec resin (Erban Kochergina et al., 2022), and all samples were measured on a single Ta filament with a Triton Plus thermal ionization mass spectrometer (TIMS; Thermo Fisher). Sr isotopic ratios were measured in static mode, that is, the abundance of all isotopes was acquired simultaneously by Faraday collectors connected to 10¹¹ Ω amplifiers (for 100 ratios and an integration time of 4 s). The measured isotopic ratios were subsequently corrected for mass-dependent instrumental fractionation using an exponential law and following an internal standard with a presumed natural ⁸⁷Sr/⁸⁶Sr ratio of 0.1194 (Steiger & Jäger, 1977). The isobaric interference of Rb at mass 87 was corrected, assuming the natural value of 0.3857 for ⁸⁷Rb/⁸⁵Rb (Steiger & Jäger, 1977). Reference material was measured together with our samples. External precision was established by repeated analysis of international reference materials [NBS987: ⁸⁷Sr/⁸⁶Sr = 0.71024 ± 7, 2σ; n = 4 (Pin et al., 2014); ⁸⁷Sr/⁸⁶Sr = 0.71024 ± 12]. The carbonate reference material EN-1 yielded an ⁸⁷Sr/⁸⁶Sr value of 0.709170 ± 18. The basalt BCR-2 reference material gave an ⁸⁷Sr/⁸⁶Sr value of 0.705015 ± 8, and the granodiorite reference material JG-1a gave an ⁸⁷Sr/⁸⁶Sr value of 0.711048 ± 12. The soil NIST 2709a reference material gave an ⁸⁷Sr/⁸⁶Sr value of 0.708153 ± 6 (n = 2). Sr concentrations were measured by ICP-OES (Agilent 5100) with a detection limit of 0.01 μg L⁻¹ (for details see Novak et al., 2020).

3 Results and Discussion

The four sites are characterized by significantly different, model-based bedrock Sr isotopic ratios (p < 0.001) (Figure 2). The Cenozoic basalt at Doupov has the lowest ⁸⁷Sr/⁸⁶Sr ratio of 0.7039, followed by the Mesozoic sediments from Hradec Králové (0.7171), the Variscan syenogranite from Svatoslav (0.7231), and the pre-Cambrian metamorphic gneiss at Javorník (0.7384). The standard deviation of the modeled bedrock Sr isotopic ratios ranges from 0.0005 to 0.0056, with the lowest and highest variance found for Doupov and Javorník, respectively.

The bioavailable Sr isotopic ratios in the soil and forest floor samples deviate from the corresponding bedrock values (Figure 2). At Doupov where the bedrock Sr isotopic ratio is lowest, the mean ⁸⁷Sr/⁸⁶Sr ratios of the soil and forest floor samples are slightly higher and closer to the atmospheric Sr signature in precipitation (0.7118). The span of all isotopic ratios is most narrow for Doupov. At Svatoslav and Javorník, the forest floor ⁸⁷Sr/⁸⁶Sr ratios are lower than those of bedrock, and also closer to the atmospheric value. This trend is less pronounced for the soil samples and not evident for Hradec Králové. Significant differences in soil ⁸⁷Sr/⁸⁶Sr ratios are evident across most sites (except Svatoslav and Hradec Králové) and all soil horizons (p < 0.001). The mineral topsoil samples exhibit higher ⁸⁷Sr/⁸⁶Sr ratios than the organic forest floor samples (p < 0.001).

Compared with the ⁸⁷Sr/⁸⁶Sr ratios in bedrock (0.7035–0.7441, modeled) and soil (0.7043–0.7552, measured), a much narrower range of ⁸⁷Sr/⁸⁶Sr ratios is found in the 120 wood samples (0.7041–0.7245). The Sr in wood is systematically closer to the large-scale atmospheric Sr isotopic signature in precipitation (0.7118), and furthest from the corresponding site's bedrock values (Figure 2). The ⁸⁷Sr/⁸⁶Sr ratios measured in all 30 wood samples at Doupov are slightly higher than those of rock and soil. However, at all other sites the ⁸⁷Sr/⁸⁶Sr ratios are much higher for the modeled rock and measured soil samples than the calculated atmospheric-derived precipitation baseline; the ⁸⁷Sr/⁸⁶Sr ratios in wood are lower and much closer to the precipitation Sr signal.

Closer inspection of the species- and stem-specific wood samples reveals insignificant differences in ⁸⁷Sr/⁸⁶Sr ratios at the site level (Figure 3). At Doupov, where the bioavailable Sr isotopic ratios are significantly (p < 0.001) lower than the precipitation Sr baseline value, the mean ⁸⁷Sr/⁸⁶Sr ratios of the 15 innermost (juvenile = young in terms of cambial age) and 15 outermost (mature = old in terms of cambial age) wood samples are 0.7045 and 0.7046, respectively. At Svatoslav, where the bioavailable Sr isotopic ratios are highest, the mean ⁸⁷Sr/⁸⁶Sr ratios of the 15 innermost and 15 outermost wood samples are 0.7220 and 0.7213, respectively. The mean ⁸⁷Sr/⁸⁶Sr ratios of the 15 innermost and 15 outermost bulk wood samples at Javorník are 0.7187 and 0.7185, respectively. The mean ⁸⁷Sr/⁸⁶Sr ratios of the 15 innermost and 15 outermost wood samples at Hradec Králové are 0.7126 and 0.7121, respectively.

We found no systematic differences in Sr isotopic ratios between tree species, cambial ages and root structures within and between the four sampling sites (p < 0.001). However, some offset in Sr isotope ratios of individual trees is evident at sites with generally low Sr concentrations, such as Hradec Králové, and the ⁸⁷Sr/⁸⁶Sr isotopic signals in wood and precipitation are comparable. Despite small influences of root architecture (Snoeck et al., 2020) and historical changes in groundwater levels (e.g., Hradec Králové is located on the floodplain sediments of the Elbe River), the observations emphasize that the Sr isotopic ratio in tree wood is influenced by both the abundance of Sr in the bedrock and its isotopic composition.

While our results suggest that plant physiological processes have negligible effects on the Sr isotopic composition of wood, our two-endmember mixing model reveals a wide range of relative Sr contributions from precipitation to the final ⁸⁷Sr/⁸⁶Sr ratio in wood (Figure 4a). While possible effects of dust are considered negligible for central Europe, atmospheric-derived rainfall Sr is recycled in the topsoil horizon, where it can be absorbed by vegetation (Figure 4b). As the geogenic bedrock Sr concentration increases, so does the weathering Sr flux, and the relative atmospheric Sr contribution from precipitation to wood decreases. The strongest impact of meteoric water on wood Sr is found at Hradec Králové and Javorník, while the lowest contribution of precipitation Sr is found in the wood samples from Doupov and Svatoslav. The atmospheric Sr contributions are highest at sites where bedrock and soil Sr concentrations are lowest. This systematic trend challenges the widespread assumption that rock weathering is the primary source of bioavailable Sr (Figure 4b). The finding also corroborates previous reports on the importance of atmospheric Sr (Blum et al., 2008; Drouet et al., 2005; Kennedy et al., 2002; Montgomery, 2010). The extent of atmospheric Sr contributions to bioavailable ⁸⁷Sr/⁸⁶Sr might be strongest when precipitation totals and/or proximity to sea spray source are high (Erban Kochergina et al., 2021; Göhring et al., 2023; Montgomery, 2010). In line with previous findings for grasses, shrubs and trees (Snoeck et al., 2020), our results indicate that different tree species, cambial ages and root systems do not significantly, or systematically, influence the Sr isotopic ratio in wood. This plant physiological finding is encouraging for Sr isotopic fingerprinting studies.

However, our study demonstrates that the bioavailable Sr signature at any given site—and so the final Sr signature in wood—is likely to be a complex mixture of the underlying bedrock and the meteoric water, determined by the rate of weathering, lithology, and the amount of precipitation. Since large-scale atmospheric Sr concentrations can mask geogenic signatures at the landscape level and wider spatial scales, isoscape models must be adjusted for Sr-based provenance studies in archeological, environmental and forensic research. Insights from biogeochemical cycling should be included in the development of bioavailable Sr maps at high spatial resolution, aiming to capture even local variations in Sr concentrations. Furthermore, we argue for the routine use of the available empirical proxy and theoretical model evidence to reinforce interpretations.

Conflict of Interest

The authors declare no conflicts of interest relevant to this study.