1 Introduction

Pedogenetic processes and the resulting soil types are strictly dependent on existing climatic conditions (Schaetzl and Anderson 2005; Clark et al. 2012; Walker et al. 2012; Meier et al. 2014; Mauri et al. 2015; Tabor and Myers 2015; Binney et al. 2017; Pérez-Lambán et al. 2017; Zhou et al. 2021). The prominent role played by climate change over time, particularly between the end of the Pleistocene and the early Holocene, in conditioning soil-forming processes through vegetation cover change has been observed in various case studies (Tallón-Armada et al. 2014; Malkiewicz et al. 2016; Pérez-Lambán et al. 2017; Chendev et al. 2018; Armas-Herrera et al. 2019). In northern Italy, the majority of the sites studied from a paleoenvironmental perspective are hilly and mountain settings, mainly lacustrine environments characterized by continuity of sedimentation (e.g., Vescovi et al. 2010; Joannin et al. 2013, and references therein). On the other hand, case studies performed in wide alluvial plains with high discontinuous sedimentation rates are rare (Ravazzi et al. 2006; Marchesini et al. 2017; Bruno et al. 2020; Marcolla et al. 2021). In this framework, during recent decades, excavation for building and engineering activities in the eastern Po Plain (northern Italy) targeted stratigraphic sequences that were well constrained from archeological and chronological points of view (Bianchini et al. 2014, 2019; Cacciari et al. 2017; Vittori Antisari et al. 2011, 2013a). In these pilot studies, buried soils dating back to the early Holocene and up to the late Middle Ages, i.e., the regional Holocene pedocomplex, were recognized (Fig. 1). The significant thickness of the sedimentary cover is effective in sealing and preserving the main soil features. These sequences often contain black soils, which, although already described by previous studies in the Po Plain (e.g., Amorosi et al. 2014a, 2017; Bruno et al. 2020; Marcolla et al. 2021), have not been interpreted in terms of paleoenvironmental conditions. Notably, in other European countries (e.g., Spain), black soils have been ascribed to specific climatic conditions, vegetation types, and the occurrence of fires (Tallón-Armada et al. 2014; Armas-Herrera et al. 2019), and related hypotheses need to be tested in northern Italy.

Fig. 1
figure 1

Typical example of a buried black soil in the areas surrounding Bologna. In particular, the picture refers to the studied site of Salara (site 3-SAL)

In this paper, we investigated three stratigraphic successions that contain black soil horizons by integrating new pedological, geochemical, chronological, palynological, and black particle data. The results were compared with similar investigations recently carried out in surrounding areas (Vittori Antisari et al. 2016, and references therein) to provide new insights into the ecological and physiographic evolution of the area and to highlight the specific genetic conditions of black soils.

2 Regional settings

2.1 Location and present-day climate

The studied sites (and those used for comparison) are located in Bologna City and in its periphery (Fig. 2). The present-day climate of this area is characterized by the following mean annual (1960–2008) parameters: air temperature ranging between 13 and 14 °C, 700–1000 mm/year of rainfall, 800–900 mm of potential evapotranspiration, 0.65–1.00 of the FAO-UNEP aridity index (subhumid type), xeric soil moisture regime (80–115 day/year), and mesic soil temperature regime (Costantini et al. 2013).

Fig. 2
figure 2

Geographic location and physical settings of the studied sites (1–3) and those taken for comparison (4–10). A Geomorphologic map of site 1-MRZ. B Essential geomorphologic elements of the Bologna city center and location of stratigraphic sites 2 and 3. Sites 4 to 10 (see Supplementary Table 5) are recalled in the discussion

In this area, affected by a very severe human influence, the natural vegetation is pedunculate oak and common hornbeam. The latter wood, characterized by Quercus robur, Carpinus betulus, and Acer campestre, represents the potential vegetation, today almost completely destroyed. Moreover, there is riparian vegetation with remnants of riparian natural woods (Salix alba, Salix purpurea, Populus nigra and more rarely Alnus glutinosa, and Ulmus minor; Puppi et al. 2010).

2.2 Geology and geomorphology

The three studied sites (Fig. 2) are located in the middle of the Reno River mountain valley (site 1) and close to the Apennine chain foothills at the apex and lower margin, respectively, of the minor Aposa stream alluvial fan (sites 2 and 3).

The Apennine is a very young and still active mountain belt bordering the southern margin of the Po River sedimentary basin (Fantoni and Franciosi 2010), whose topographic evolution mainly developed after the Messinian (Ghielmi et al. 2010, 2012), during the late Pliocene and Pleistocene (Martelli et al. 2017a, b), inducing a generalized uplift of the chain (Muttoni et al. 2003; Gunderson et al. 2014). These processes generated the Pedeapenninic Thrust Front (Boccaletti et al. 1985) lying beneath the city of Bologna and currently uplifting at a rate of ca. 1 mm/year (D’Anastasio et al. 2006; Carminati and Vadacca 2010), while the alluvial plain is subsiding at a rate of 0.5–2 mm/year (Carminati and Martinelli 2002).

Site 1-MRZ (Marzabotto), lying in the core of the mountain chain, approximately 30 km upstream of the valley termination, contains an alluvial terrace system carved in a Tertiary marly formation (Marne di Cigarello Fm) (Fig. 2A), which was generated by the Reno River during the middle to upper Pleistocene. Terrace Ft2 was still flooded during the nineteenth and twentieth centuries AD. Terrace Ft3, hosting the studied stratigraphic succession of site 1, lies approximately 25 m above the present river talweg and dates back to approximately 19 ky BP (Picotti and Pazzaglia 2008). The Ft3 terrace hosts the Etruscan Age town structures of Kainua (sixth to fourth centuries BC: Bertani 2010) and was settled by at least the Chalcolithic and the Bronze Age (Cattani and Govi 2010).

Site 2-SMA (Bologna, S. Mamolo) lies at the valley end of the Aposa small stream (Cremonini 2002) at the transition of its apical fan area (Fig. 2B). The Aposa alluvial fan coincides with the Bologna historical settlement starting from the eighth century BC (Malnati 2010). Such a long-lasting period of settlement generated a 2- 4-m-thick anthropogenic deposit prism (Giorgi 2002) and makes recognition of the pristine, natural fan morphology difficult (Cremonini 1991). The most prominent morphologies in the fan area are those of two stream trenches (Fig. 2B) corresponding to the active Aposa stream course (Cremonini and Bracci 2010) and a second artificial course lying near site 3. On the southern side of the medieval town, near site 2, a thick series of leveling deposits masks the morphological transition between the Aposa valley termination and the subsequent fan trench.

Site 3-SAL (Bologna, Salara) lies near the lower boundary of the Aposa stream alluvial fan, where the overbank silty deposits prevail with respect to the channel sandy-gravelly facies (Amorosi et al. 2014b).

3 Methods and materials

3.1 Field survey and sampling

The three investigated sites (1-MRZ = Marzabotto; 2-SMA = Bologna-San Mamolo; 3-SAL = Bologna-Salara) were selected in correspondence with engineering excavations and were studied from stratigraphic and pedological points of view (Table 1; Fig. 3). The field observations allowed the recognition of stratigraphic units identified on the basis of distinctive characteristics (layer geometry and lateral relationships, mineral groundmass, coarse-grain-sized natural elements, colors, physical stress surfaces, postdepositional disturbances, human artifacts, and related ages). The morphological characteristics of each soil horizon were described according to Schoeneberger et al. (2012). Each soil horizon was sampled by collecting approximately 1 kg of material for analytical investigation.

Fig. 3
figure 3

Stratigraphic logs and related succession of buried soil profiles recorded at the three studied sites (1–3). Radiometric datings are expressed at the 2σ interval

3.2 Analytical methods

3.2.1 Grain size and physicochemical analyses

Soil samples were air dried and dry-sieved with a 2-mm mesh sieve. The particle size distribution was determined by the pipette method after dispersion of the sample with a sodium hexametaphosphate solution (Gee and Brauder 1986). The pH was determined potentiometrically in a 1:2.5 (w/v) soil:distilled water suspension with a Crison pH meter. The carbonate content was measured by volumetric analysis of the carbon dioxide released by a 6 M HCl solution (Loeppert and Suarez 1996).

3.2.2 Carbon speciation and isotopic analysis

Soil samples were finely powdered using an automated agate mortar. Carbon speciation was carried out with an innovative method recently defined as “smart combustion” (Zethof et al. 2019) using an Elementar SoliTOC analyzer, which allows oxidation-temperature-dependent differentiation of distinct carbon fractions. The analytical run takes 1600 s and involves a three-step heating of the samples at 400 °C, 600 °C, and 900 °C coupled with holding times of 230, 120, and 150 s, respectively. The formed CO2 was detected by an infrared detector. Accuracy and precision, evaluated by repeated analyses of samples and of soil standards, were better than 5% of the measured concentration (Natali et al. 2020). Carbon isotopic analysis of the TC, TOC, and TIC was carried out with an elemental analyzer vario-microcube coupled with an isotope ratio mass spectrometer (IRMS) Isoprime 100, according to the protocol defined by Natali and Bianchini (2015) and Natali et al. (2018). Data are reported in Supplementary Table 1 and summarized in Table 2.

3.2.3 Geochemical analyses by XRF and ICP-OES

X-ray fluorescence (XRF) analysis was carried out for the measurement of major elements according to the following procedure. Approximately 4 g of powder was pressed with the addition of boric acid by a hydraulic press to obtain powder pellets. Simultaneously, 0.5–0.6 g of powder was heated for approximately 12 h in a furnace at 1000 °C to determine the LOI (loss on ignition) values. This parameter measures the concentration of volatile species (e.g., H2O, CO2, and halogens) contained in the sample. The analysis of the powder pellets was carried out using an ARL Advant-XP spectrometer, which was properly calibrated analyzing certified reference materials, as described by Di Giuseppe et al. (2014). Precision and accuracy were calculated by repeated analysis of international standards having matrices comparable with those investigated, i.e., various typologies of sedimentary rocks, and were generally better than 3% for Si, Ti, Fe, Ca, and K, and 7% for Mg, Al, Mn, and Na. Data are reported in Supplementary Table 2 and summarized in Table 3.

Inductively coupled plasma optical emission spectrometry (ICP-OES) on aqua regia leachates was carried out for trace element analysis according to the following procedure. Approximately 0.25 g of dried soil was treated with aqua regia (6 mL 37% HCl and 3 mL 65% HNO3 Suprapur, E. Merck, Germany), following the procedure proposed by Vittori Antisari et al. (2013b); mineralization was carried out in Teflon bombs in a Milestone 1200 microwave oven and element concentrations were determined using a Spectro Arcos Ametek spectrometer. The accuracy of the instrumental method and analytical procedures was checked by repeating (three times) the analysis of samples and international reference materials (BCR) and laboratory internal standards, as described by Vittori Antisari et al. (2014). Data are reported in Supplementary Table 3 and summarized in Table 4.

3.2.4 Radiocarbon dating

The sample radiocarbon dating (14C) was performed at the University of Lecce Laboratory (CEDAD) through the AMS technique according to the method described by Calcagnile et al. (2005) and Fiorentino et al. (2008). This included a multistep sample pretreatment in order to remove sources of contamination and to turn the material into graphite. The carbon isotopic ratios (13C/12C expressed as δ13C) were then analyzed by comparing the 12C and 13C ion beam currents, and the sample 14C counts were in turn compared with those obtained for standard reference materials of known isotopic composition (e.g., the fossil wood IAEA C4). Samples of oxalic acid of known concentration given by the National Institute of Standard and Technology were used to perform the quality control of the results. The conventional radiocarbon ages were calculated according to Stuiver and Polach (1977), and then converted to calendar ages by using the latest internationally accepted calibration dataset (Reimer et al. 2004; Blackwell et al. 2006) and the OxCal 3.10 software (Bronk Ramsey 2001; Reimer et al. 2013, 2020). Data are reported in Table 5.

3.2.5 Palynological analyses

Seven samples were investigated for pollen plus Pteridophyta spores (pollen) and nonpollen palynomorphs (NPPs) to understand the vegetation landscape and to retrieve information on the climatic conditions that existed during pedogenesis. Samples, including buried black soils, were collected at three sites: two samples from site 1-MRZ (AC and 2A2b horizons), two samples from site 2-SMA (6A1b and 6A2b horizons), and three samples from site 3-SAL (2AC2b, 3A1b, and 6Ab horizons). Sample locations are shown in Fig. 3.

Pollen and other record extraction and slide preparation were carried out by D. Arobba at the Archaeobotanical Laboratory-Archaeological Museum of the International Institute of Ligurian Studies (Finale Ligure Borgo, Savona-Italy) by applying a methodology already tested for recent pollen substrates with some minor modifications (Traverse 2007; Giuffra et al. 2011). The method includes the following phases: adding a tablet of Lycopodium spores resuspended in 10% HCl for the calculation of pollen concentration (number of palynomorphs/gram); treatment with 2% hot sodium pyrophosphate solution for dispersing fine clays and sieving through a 5-micron nylon screen; maceration with 10% hot KOH solution to remove humic acids; gravity separation with heavy liquid (Thoulet d = 2.1); vacuum filtration on glass fiber papers (Schleicher & Schuell GF/52); removal of silicates with 40% HF for 24 h; and staining with safranin-O and preparation of glycerine fixed slides with Histolaque LMR.

Analyses were carried out under a light microscope (400 and 1000 magnification), and 400–800 pollen/sample was counted. Pollen identification was based on modern and Holocene pollen slide collection, current atlases and pollen keys (e.g., Andersen 1979; Faegri and Iversen 1989; Moore et al. 1994; Reille 1992, 1995), and a large morphopalynological bibliography. The basic pollen terminology was based on Berglund and Ralska-Jasiewiczowa (1986); pollen type names refer to the relevant pollen keys. The main data are synthetically reported in Table 6, whereas the complete dataset is reported in Supplementary Table 4. The term “taxon” is used to indicate both the systematic categories and the morphological pollen types. Plant names refer to Pignatti et al. (2017). Concentrations were reported for pollen and NPPs. Percentage pollen spectra were calculated from the pollen sum, including the total pollen. Pteridophyta percentages were calculated from the pollen sum plus themselves. Table 6 also reports some sums and indices that are significant for the vegetational-ecological interpretation, e.g., trees + shrubs, deciduous broadleaves, animal and human indicators e.g., wild plants (Artemisia, Amaranthaceae, Asteroideae, Cichorioideae, Cirsium, Galium, Plantago, Potentilla type, Ranunculaceae, Rumex, Stellaria type, Urtica dioica) suggesting animal and human presence (Mazier et al. 2006; Mercuri et al. 2013; Florenzano et al. 2015; Deza‑Araujo et al. 2021), as well as Hydro-Helo-Hygrophytes, i.e., plants in humid environments. Pollen clumps were also observed and related to disturbance effects from grazing and trampling, thus representing a proxy for herbivore load (Miehe et al. 2009). NPP identification was based on atlases and keys and relevant bibliography (see Torri, 2011; Miola, 2012). The terms “type” and undiff. (undifferentiated) reported in the spectra were omitted in the text for pollen.

3.2.6 Black microparticle analyses

Black microparticles were investigated on the same slides examined for the pollen analyses, and the results are reported in Table 6, and Supplementary Table 4.

Three categories of particles were counted: (1) black particles = wood charcoals ≥ 125 µm (maximum 300 µm); (2) black particles = nonwood charcoals with round edges or sometimes perfectly round ≥ 125 µm (maximum 290 µm); and (3) black particles < 125 µm undiff. (impossible to separate wood and nonwood). Particles were counted with the method outlined by Torri (2011) with few modifications. More precisely wood and nonwood particles ≥ 125 µm were counted on the whole slide separately; black particles undiff. < 125 µm were counted together in random rows up to at least 1000 particles. Categories 1 and 2 were interpreted as evidence of local and nearby fires. Category 3 was interpreted as a possible suggestion of fires affecting little shrubs, litter, and surface soil, but they could also derive from the breakup of large-sized particles or be transported from far away.

3.3 Statistical analysis

To compare the characteristics of buried black horizons with those of the other horizons for each study site and therefore within each stratigraphic sequence, a nonparametric statistical analysis was performed. In particular, a Kruskal–Wallis test was carried out. Within each stratigraphic sequence, the buried black horizons were considered different compared to the other horizons for P values less than 0.05. The statistical test was performed using the R software.

3.4 BHs in Emilia, literature research, and study approach

To better constrain whether and how the studied buried soils were influenced by forcing factors, the new data presented in this study were compared with other data retrieved from the literature concerning other black soil occurrences in surrounding areas. These additional data have been recovered at sites 4 to 10 (Fig. 1), and their physiographic, stratigraphic, and chronological constraints are listed in Supplementary Table 5. As described by the relevant authors (properly quoted in the supplementary section), they are “black soil horizons” that are morphologically similar to the BHs studied in this paper, and their thicknesses rarely exceed the value of 50–60 cm. The geomorphological setting of these additional sites is the same as that of sites 1 to 3. The radiocarbon dating from site 6 was newly acquired along with that from sites 1 to 3 to provide a time constraint for the related stratigraphic data. Particular emphasis is given to the radiocarbon ages of these BHs to evaluate the possible correspondence between their formation and climatic events or pulses, particularly cold/cool pulses.

Considering the lack of suitable paleoclimatic datasets for the studied geographic area, the BH dataset (1 to 10) was compared with the N-GRIP δ18O curve as paleoclimatic reference. Additional constraints were provided by the Δ14C residual curve (Stuiver et al. 1998), which is another paleoclimate proxy for the Holocene (e.g., Magny 1993, 1995; Blaauw et al. 2004). In the δ18O curve, lower (more negative) values correspond to cold (glacial) periods or cool (Holocene) short climatic pulses, whereas higher values correspond to warmer time. In contrast, Δ14C residual curve shows antithetical behavior with positive peaks corresponding to cold periods and vice versa, with the exclusion of 8.2 climatic event having a peculiar origin (Alley and Agustsdottir 2005).

4 Results

4.1 Field observations

The observations carried out in the field surveys at sites 1-MRZ, 2-SMA, and 3-SAL are summarized in Fig. 3, where radiocarbon dates (see Sect. 4.2.3) are also reported. The subsequent short stratigraphic outlines for each of the three sites are given according to a geographic location ordered from the highest elevation to downstream.

At site 1-MRZ, the bottom deposits are made of gravel representing the bar facies of the Reno riverbed during the Last Glacial Maximum. The uppermost part of the deposit contains slightly weathered marly gravels in a darkened loamy matrix and can be interpreted as a lateral equivalent or the erosional remains of a Bølling-age reddened soil (Cremaschi 1979; Gasperi et al. 1989; Cremaschi and Nicosia 2012; Ravazzi et al. 2012; Samartin et al. 2012) outcropping on the same terrace in a site located 300 m to the south. Hence, although no chronometric age for the 3ABb horizon is available, the morphostratigraphic setting of fluvial terrace Ft3 (see Sect. 2.2) also makes the site 1-MRZ a key pedosequence for constraining the reddened soil whose weathering was allowed only by the peculiar climatic conditions (Samartin et al. 2012; Obase and Abe-Ouchi 2019) recorded during Greenland Interstadial GI-1e, i.e., the Bølling chronozone (Reimer et al. 2020).

Black soil 2 of site 1-MRZ contains charcoal fragments all along the profile together with mm-sized fragments of fired earthy material, and its uppermost horizon records many slickensides, suggesting a vertic character for this horizon. Soil 2 is capped by a thin, mainly silty, sequence (AC horizon) dating back to between the Chalcolithic and Bronze Ages. Considering the location on the high alluvial terrace, both soil 2 soil and the AC horizon were generated by local splash and sheet erosion of the terrace surface itself together with the nearest hillslope.

For site 2-SMA, it is worth stressing that the survey was performed at the core of a courtyard that is part of a religious complex. The top of the log lies at an elevation 50 cm higher than the mean floor of the complex (76 m asl) and consists of modern deposits. The underlying stratigraphic units up to a depth of 160 cm contain ceramic artifacts dating back to the sixteenth to seventeenth centuries AD, and they must be interpreted as artificial leveling deposits linked to various building phases that occurred between the fifteenth and fourteenth centuries AD. No reason exists for supposing that excavations erased pre-existing stratigraphic units; thus, soil 3 of site 2-SMA is the last natural soil containing first Iron Age sherds (ca. seventh century AD) and was possibly exposed during the Roman Age and beyond. This long-lasting soil outcropping and its late artificial cover are consistent with both the soil 3 thickness and the site physiographic setting, which is a left-hand alluvial terrace lying at a significant elevation above the Aposa stream talweg. Instead, the silty deposits that are the parent material of buried soils 4, 5, and 6 of site 2-SMA can be interpreted as overbank deposits of Aposa or distal colluvial deposits delivered by sandy hillslope degradation. In these soils, well-developed burrowing activity was preserved, thus indicating the lack of agricultural disturbance. For the abovementioned reasons, the main black soil laying at a depth of 3.13 m was buried at a depth of only 1.5 m in the past, before the fourteenth century AD.

Site 3-SAL is located near the left shoulder (upper slope-break) of a wide human-made trench created in the twelfth century AD to accommodate a commercial harbor for the growing medieval city. These activities removed a massive amount of the natural sedimentary succession (Cremonini et al. 2007). The local refilling of the trench slope degradation is recorded between depths of 2.15 and 3.42 m, whereas the uppermost 2 m of the sequence is missing due to recent construction activities. The preserved part of the sedimentary succession is almost homogeneously clayey silt (silty loam) and is located at or just beyond the lower margin of the Aposa alluvial fan. Notwithstanding this fine character, and suggesting prominent overbank facies, the local sedimentary aggradation is mainly perceivable thanks to the darkened horizon of the buried soil profiles marking the sedimentary hiatuses linked to the mainstream entrenchment or, more likely, to the lateral shifting of the active depositional fan lobe or alluvial ridge.

4.2 Results of the analytical investigation

4.2.1 Morphological and physicochemical characteristics and carbon speciation

Within the abovementioned stratigraphic sets at distinct sites, 15 soil cycles were detected, comprising 35 horizons; the related pedological and physicochemical properties are reported in Table 1.

Notably, within the distinct soil cycles, black horizons (BHs) were observed, as they are characterized by a dry Munsell’s chart color from 2.5Y3/1 to 2.5Y6/1 (from “very dark gray” to “gray”). In particular, the following nine BHs were recognized: MRZ_2A1 and 2A2 at depths of 75–100 and 100–155 cm at site 1-MRZ; SMA_3A at depths of 135–179 cm; SMA_6A1, 6A2, and 6AC at depths of 313–330, 330–342, and 342–362 cm, respectively, at site 2-SMA; SAL_3A1 and 3A2 at depths of 405–427 and 427–460 cm, respectively; and SAL_6A1 at depths of 738–800 cm at site 3-SAL. Below these black horizons, other horizons enriched with both nodules and concretions of Ca-carbonate (BCk/Ck) are usually recorded.

The BH at sites 2-SMA and 3-SAL, compared with those associated with the respective stratigraphic sequences, showed more distinctive physicochemical characteristics than those at site 1-MRZ. Generally, they had lower pH values, CaCO3, and sand contents (Table 1). MRZ_2A1, despite having the typical Munsell color of BHs, appears to be affected by CaCO3 contamination that would derive from soil decarbonation processes of the overlying MRZ_1 soil. Consequently, only MRZ_2A2 showed full characteristics envisaged in other BHs; for this reason, the statistical test for the MRZ sequence was not significant. As shown by the statistical elaboration reported in Table 2 (based on the data fully reported in Supplementary Table 1), BHs differ from the associated horizons, being significantly depleted in CaCO3. For this reason, the correct speciation of total carbon (TC) in its distinct organic and inorganic fractions (TOC and TIC, respectively) is crucial for understanding the considered case study. Both the total composition and fractions were characterized from elemental and isotopic points of view.

Table 1 Soils morphological and physicochemical characters of the site 1 (MRZ), site 2 (SMA), and site 3 (SAL) stratigraphic successions. Black horizons are highlighted by gray background
Table 2 Contents of carbonates (CaCO3), total carbon (TC), total organic carbon (TOC), total inorganic carbon (TIC), δ13C of TC, TOC, and TIC in black horizons (BH) and in non-black horizons of the site 1 (MRZ), site 2 (SMA), and site 3 (SAL) stratigraphic successions. * indicates significant differences between black and non-black horizons within each site according to Kruskal–Wallis test (P < 0.05)

The TOC contents of BHs did not show significant differences with respect to the other horizons of the stratigraphic sequences; therefore, organic matter concentration does not appear to be the key explaining the BH peculiarity. In contrast, the C pools containing inorganic C (e.g., TIC and TC) were significantly depleted in BHs. Analogously, BHs displayed a more negative δ13CTC compared to overlying and underlying horizons. These differences are highlighted in Fig. 4, which shows the relationship between the TOC/TIC ratios and δ13CTC. In this diagram, the BHs are well separated by the other horizons of the respective stratigraphic succession. An exception is represented by site 1-MRZ, where MRZ-2A1 was contaminated by a concentration of younger carbonates leached from the overlying soil horizons. The comparatively more negative δ13CTC of BHs does not merely reflect the paucity of carbonate, because the isotopic compositions of organic and inorganic fractions, δ13CTOC and δ13CTIC, respectively, are also peculiarly negative in BHs (see Supplementary Table 1). Therefore, 13C/12C differences in the organic fraction favor peculiarly distinct types of vegetation during BH formation; the signature of vegetation could also have been transferred to the IC fraction during the formation of pedogenic carbonates.

Fig. 4
figure 4

Binary diagram reporting the TOC/TIC ratio vs. δ13CTC, emphasizing the peculiarity of BHs. An arrow has been placed in correspondence with sample MRZ_2A1 to suggest that the pristine features of this horizon have been altered by the late precipitation of carbonate, as testified by field observations

4.2.2 Major and microelements

XRF analyses (fully reported in Supplementary Table 2) are effective in differentiating the geochemical nature of BHs with respect to the other soil horizons within each stratigraphic succession. Table 3 underlines, for major elements, the geochemical differences characterizing the BHs. The BHs in site 2-SMA and site 3-SAL were characterized by significantly higher amounts of SiO2, TiO2, Al2O3, Fe2O3, and K2O and a lower amount of CaO than the overlying and underlying horizons. On the other hand, statistically, less significant are the differences found in site 1-MRZ.

Table 3 Contents of SiO2, TiO2, Al2O3, Fe2O3, CaO, K2O, MgO, MnO, Na2O, and P2O5 in black horizons (BH) and in non-black horizons of the site 1 (MRZ), site 2 (SMA), and site 3 (SAL) stratigraphic successions. * indicates significant differences between black and non-black horizons within each site according to Kruskal–Wallis test (P < 0.05)

Moreover, as shown in Fig. 5, the concentration of elements hosted by silica-alumina minerals (e.g., aluminum) is inversely correlated with that of calcium hosted in carbonate.

Fig. 5
figure 5

CaO vs. Al2O3 diagram emphasizing the peculiarity of BHs. An arrow has been placed in correspondence with sample MRZ_2A1 to suggest that the pristine features of this horizon have been altered by the late precipitation of carbonate, as testified by field observations

The microelement data analyzed by ICP-OES on aqua regia leachates are fully reported in Supplementary Table 3, while statistical tests are reported in Table 4 to emphasize BH peculiarities. It appears that trace elements are also useful to discriminate BHs from the associated horizons. Many trace elements, such as B, Be, Cr, Li, Ni, Sn, V, and Zn, have higher concentrations in BHs than in other horizons of the respective stratigraphic succession (Table 4). In contrast, in BHs, Sr, which varies with Ca in carbonates, is comparatively depleted.

Table 4 Contents of As, B, Be, Cd, Co, Cr, Cu, Li, Mo, Ni, Pb, Sb, Sn, Sr, V, and Zn in black horizons (BH) and in non-black horizons of the site 1 (MRZ), site 2 (SMA), and site 3 (SAL) stratigraphic successions. * indicates significant differences between black and non-black horizons within each site according to Kruskal–Wallis test (P < 0.05)

4.2.3 14C dating

The radiocarbon dating performed on the selected soil horizons is listed in Table 5. In the table, both the 1σ and 2σ values are reported, but in the subsequent discussion, 2σ intervals are taken into account to encompass the maximum age probability, allowing comparison with other chronological data.

Table 5 Black soils chronology in the studied sites

At site 1-MRZ, black soil 2 dates back to 6880–6642 calibrated years BP; thus, in terms of northern Italy cultural chronology, it is attributable to the middle Neolithic Age (Pessina and Tinè 2008). At site 2-SMA, black soil 6 shows a date of 15,557–14,872 calibrated years BP that fits with the Oldest Dryas Chronozone (see Sect. 5.2) or simply the end of glacial stadial GS-2.1 (Rasmussen et al. 2014). Finally, at site 3-SAL, black soil 3 dates back to 25,687–25,083 calibrated years BP, i.e., it is attributable to the Last Glacial Maximum (Shakun and Carlson 2010; Scapozza et al. 2014), which is glacial stadial GS-3 of the last glacial period (Rasmussen et al. 2014).

4.2.4 Palynological and black particle analyses

In the examined samples, the degree of pollen preservation, diversity, and concentration was high enough to perform a detailed analysis, as synthetized in Table 6 and fully reported in Supplementary Table 4.

Table 6 Palynological and black particle analyses

Results recorded at site 1-MRZ are listed below.

Horizon AC – pollen assemblage – the forest cover was very low (trees/shrubs = 16.8%) and represented by conifers (8.4%, mainly Pinus sylvestris) and deciduous broadleaves (6.0%, mainly Corylus avellana and Quercus deciduous). Herbs (83.2%) are mainly represented by Cichorioideae, which are interpreted as animal and human indicators (Florenzano et al. 2015; Deza‑Araujo et al. 2021), and by pollen clumps indicating disturbance effects from trampling (Miehe et al. 2009).

Hydro-Helo-Hygrophytes are low. NPPs include mainly dinoflagellate algae (see Supplementary Table 4). Black particles are relatively high (53,800/g) and are mainly represented by undiff. black particles < 125 µm. The vegetal landscape was an open, steppe-like land with scattered trees/shrubs. Humid environments were scarce. The climate was cool.

Horizon 2A2 – pollen assemblage – the arboreal pollen increases (trees/shrubs = 61.8%) and is mainly represented by conifers (50.0%, especially Pinus and Abies).

Deciduous broadleaves were low (7.8%) and are mainly represented by Corylus. Herbs (38.2%) are mainly represented by Cichorioideae. Animal and human indicators (see Sect. 3.2.5 and Supplementary Table 4) are moderate and are mainly represented by Cichorioideae and pollen clumps.

Hydro-Helo-Hygrophytes are low. NPPs show a prevalence of dinoflagellate algae (see Supplementary Table 4). Black particles are very abundant (148,734/g). Among them, undiff. black particles < 125 µm prevail, but there is also a good presence of large particles ≥ 125 µm, indicating local and nearby fires.

The landscape was shaped by a dense forest, mainly made of conifers with some broadleaves. Humid environments were scarce. The climate was cool and dry.

Results recorded at site 2-SMA are listed below.

Horizon 6A1b – pollen assemblage – the forest cover was moderate (47.2%). Conifers were dominant (30.9%, mainly Pinus sylvestris), and deciduous broadleaves were low (7.4%, mainly Quercus). Herbs (52.8%) are mainly represented by Asteraceae. Animal and human indicators are moderate and are mainly represented by Cichorioideae with some pollen clumps. Hydro-Helo-Hygrophytes are low. NPPs are few. Black particles are relatively high (88,458/g). Small particles ≤ 125 µm are always the most abundant. The vegetal landscape was a park-like forest with conifers (pines) and some deciduous broadleaves. Humid environments were scarce. The climate was cool and dry.

Horizon 6A2b – pollen assemblage – the forest cover was high (72.1%) and mainly formed by conifers (65.1%, especially Pinus sylvestris). Deciduous broadleaves were very few (3.8%). Herbs (27.9%) are mainly represented by Asteraceae. Animal and human indicators are very low. Hydro-Helo-Hygrophytes are low. NPPs are few. Black particles are very high (147,076/g) and are mainly represented by small particles < 125 µm. The largest black particles ≥ 125 µm are moderate. The landscape was shaped by a dense forest mainly made by conifers. Humid environments were very scarce. The climate was cold and dry.

Results recorded at site 3-SAL are listed below.

Horizon 2AC1b – pollen assemblage – the forest cover was high (trees/shrubs = 71.6%) and mainly represented by conifers (64.9%, especially Pinus sylvestris); broadleaves were less represented (10.1%, mainly Quercus). Herbs are mainly represented by Cichorioideae and Poaceae.

Animal and human indicators are moderate and mainly include Cichorioideae. Pollen clumps are absent. Hydro-Helo-Hygrophytes are very low. NPPs show mainly fungi. Black particles are low (21,611/g). Among them, small particles < 125 µm prevail and are accompanied by some of the largest black particles ≥ 125 µm. The vegetation landscape was shaped by a mixed forest with dominant conifers with some deciduous broadleaves. Humid environments were very scarce. The climate was cool and dry.

Horizon 3A1b – pollen assemblage – the forest cover was moderate (trees and shrubs = 39.6%) and mainly represented by conifers (30.4%, especially Pinus and Abies). Broadleaves were low (6.7%, mainly Tilia). Herbs were 60.4% and mainly represented by Asteroideae. Animal and human indicators are very high and are mainly represented by Cichorioideae, with high pollen clumps. Hydro-Helo-Hygrophytes are very low. NPPs are very low, mainly fungi. Black particles are relatively high (57,858/g) and mainly represented by small particles < 125 µm, while the largest black particles ≥ 125 µm are very few. The vegetal landscape was a park-like forest possibly cleared by humans, with conifers and some deciduous broadleaves. Humid environments were very scarce. The climate was cool and dry.

Horizon 6Ab – pollen assemblage – forest cover was very high (trees/shrubs = 90.9%), mainly represented by conifers (88.7%), especially pines. Deciduous broadleaves were scarce (0.9%). Herbs were sporadic. Animal and human indicators are very low. Hydro-Helo-Hygrophytes are very few. NPPs are moderate and mainly include green algae (Zygnemataceae; see Supplementary Table 4). Black particles are low (23,225/g). Among them, the undiff. black particles < 125 µm prevail and are accompanied by a significant amount of the largest particles ≥ 125 µm. The landscape was shaped by a thick conifer forest mainly made of pines. Humid environments were very scarce. Climate was cold and dry.

The resulting paleoclimatic picture is also supported by new paleobotanical evidence from the central-eastern Po River alluvial plain (Cacciari et al. 2020). In Fig. 6, a good fitting of the dated site samples with cold or cool episodes can be perceived. In the LGM and Late Glacial, the general climatic conditions were persistent over time and widely encompassed dating uncertainty (sites 2, 3, 5, 6, and 7). In the Holocene, from the Preboreal to Atlantic (sites 1, 4, 8, 9, and 10), it is more difficult to state if a sample fits well with each time interval of the δ18O curve. The better definition of the time-lapse duration of each worsening episode suggested by the residual Δ14C curve allows us to recognize a more evident correlatability. On the eastern side of the city area, the BH samples from sites 5, 6, and 7 have similar ages (Allerød/Younger Dryas transition), and they could truly be mutually correlatable. Site 9 is not represented by a single sample but by a set of samples and hence possesses only a descriptive value for pedogenetic episodes and a consequently low reliability. The site 4 sample can be associated with the 9.3 ky BP pulse as well as the site 8 sample, whereas the site 10b sample is consistent with the 8.2 ky BP (boreal) pulse. The site 10a sample and possibly site 1-MRZ appear to be correlatable to the Atlantic chronozone pulse (Early Neolithic).

Fig. 6
figure 6

Black soil chronology and paleoclimatic setting relationships. The horizontal black bars suggest the 2σ radiocarbon age for each site sample, as reported in Supplementary Table 5. A Plot of the δ18O variation from the N-GRIP core dataset: raw data after Svensson et al. (2008). The gray color is the envelope of the raw data; the central solid line is the ten-unit running mean of the raw data. Data refer to 1950 and not to 2000. B Details of panel A showing the original raw dataset between 16 and 6 ky BP. C Variation plot of residual Δ14C concentration from the GISP dataset (http://www.ncdc.noaa.gov/paleo/icecore/greenland/summit/document/gripisot.htm.). The solid line is the ten-unit running mean of the original raw data

5 Discussion

5.1 Timing of BH development in Emilia

As reported by Fig. 6, the cool pulses appear to have a 150–200-year duration according to the δ18O curve, or 700–800 years according to the Δ14C curve. The duration of these time lapses seems suitable for BH genesis (Duchaufour 1968). The two samples from site 10 (/a and /b) can support this idea; in fact, they are horizons of the same soil profile but show an age difference of approximately 600 years without any appreciable sediment addition, thus suggesting the existence of a top-down time gradient depending on the SOM turnover and mean residence time (e.g., Vysloužilová et al. 2016). The site 1-MRZ sample shows an age falling at the youngest limit of the Atlantic worsening pulse, or just beyond it. In this case, its true inception age could predate the available 14C age. Regardless, this soil seems to be the last that display morphological characteristics of the profile similar to that of the more ancient site samples (thickness, color, and CaCO3 concentration). Possibly, after the Chalcolithic Age, both climate characteristics and ancient human impact on the Reno River valley (Eppes et al. 2008) did not ever allow the development of similar BH soils.

5.2 Vegetation and climate inferred by palynological and black particle records

The site 1-MRZ series shows the transition from a dense forest, mainly conifers in a cool climate with important local and nearby fires, to a steppe-like land that was likely due to human clearance in a temperate climate. The black particles are more abundant in the buried black soil (Table 6), indicating significant episodes of biomass burning.

The site 2-SMA series shows the transition from a thick forest, mainly represented by conifers in a cold and dry climate, to a park-like forest that was probably derived from human clearance, during a cool and dry climate. Black particles are more abundant in the deepest buried black soil, indicating significant episodes of biomass burning.

The site 3-SAL series shows first the transition from a very thick conifer forest with some local and nearby fires in a cold and dry climate to a park-like forest probably due to human clearance, with dominant conifers and some deciduous broadleaves in a cool and dry climate. Subsequently, there has been a re-increase in forests with dominant conifers and deciduous broadleaves in a cool and dry climate.

Based on palynological and black particle analyses, buried black soils appear to be characterized by variously dense forest with prevalent conifers, scarce humid environments, and a cold or cool dry climate. Considering the recorded 14C ages of the studied BHs, the hypothesis is compatible with findings observed in sediments and paleosols from other localities in Emilia Romagna (Accorsi et al. 1996, 1999; Amorosi et al. 2001; Vittori Antisari et al. 2016) and more generally in northern Italy (Pini 2002; Serandrei-Barbero et al. 2005; Ravazzi et al. 2012; Guido et al. 2020; Marcolla et al. 2021). The high presence of undiff. black particles < 125 µm was probably due to fires affecting shrubs, litter, and surface soil.

However, the above reported landscape reconstructions should be taken with caution. In fact, direct comparisons with other pollen diagrams or spectra of northern Italy were not possible because pollen diagrams or single spectra also reporting wood and nonwood black microparticles were not available.

5.3 Black horizon genetic process

The chronological framework resulting from sites 1 to 10 indicates that the recorded pedological features are not related to a single pedogenetic phase, but rather they are repeatedly triggered during the Late Glacial and the Holocene in connection with worsening climatic pulses characterized by temperature decreases, as also confirmed by deep coring (Bruno et al. 2020; Marcolla et al. 2021).

Notably, the key BH character, with respect to the overlying and underlying horizons, is the paucity of carbonate (also measured in terms of TIC and Ca), which seems to be related to both sedimentological and subsequent pedological processes. The scarcity of carbonate was related to the interaction with acidic–CO2-rich–water (Ulery et al. 1993) due to the leaching processes operated by rainfall water infiltrating the acidic soil typical of pine forest biomes (Nilsson et al. 1982; Hornung 1985; Binkley 1996; Godbold et al. 2003) during cold periods and cool phases.

Carbon isotopes of BH are in fact characterized by δ13C values that are more negative with respect to the overlying and underlying horizons. This is in primis related to the scarcity of the inorganic carbon fraction, which is always characterized by δ13C values that are less negative than the associated organic fraction. However, the isotopic analysis of the organic fraction also shows more negative δ13CTOC values for BHs, a feature necessarily implying a difference in pristine vegetation. This means that during the genesis of BHs, the vegetation was richer in C3 plants (having an average δ13C of − 25‰) than in C4 plants (having an average δ13C of − 15‰), as usually noticed during cold periods (Meier et al., 2014). Thus, during BH formation, an arboreal association prevailed with respect to grassy cover, as corroborated by pollen investigation invariably showing pine-rich vegetation. Moreover, in BHs, the concentration of most trace elements (except Sr, which covariates with Ca) is always comparatively enriched but with variable enrichment factors, probably depending on a distinct involvement of each element in leaching processes.

The black particle investigation also revealed a significant concentration of microcharcoal in BHs. This means that during BH formation, the recorded pine forest cover was affected by natural fires (e.g., Badino et al. 2020), as still observed nowadays (Sofronov et al. 1998), or even fires of human-induced origin, as envisaged in many other case studies (Piquè et al. 2020; Aranbarri et al. 2020). In the European framework, the charcoal frequency severely increased during the Holocene (Power 2013), already dating back to pre-Neolithic times (Gerlach et al. 2012; Guido et al. 2013; Brandt et al. 2014) when the vegetation burning practice could have also been due to the hunter-gatherer community feeding approach (Tolksdorf et al. 2014). In the case of site 1-MRZ, dating back to the Neolithic and showing the maximum concentration of black particles, human activities (Rösch et al. 2002; Cremaschi and Nicosia 2012; Robinson 2014; Jacomet et al. 2016) can be suggested. In such a case, this could be the first evidence in the region of the human impact on the environment. In fact, according to previous investigations, in Emilia, only the Chalcolithic recorded the beginning of a severe human impact on the landscape (Cremaschi and Nicosia 2012; Zanchetta et al. 2013). In the case of the older site 2-SMA and site 3-SLA, the drivers are more plausibly triggered by natural causes (Zhao et al. 2021; Resco de Dios et al. 2021), even if some studies speculate that anthropogenic activities influenced the fire regime even in pre-Holocene times (Kaplan et al. 2016; Sorensen 2017). Regardless of the origin of fires, the effect of combustion on soils was the partial transformation of organic matter and organometallic compounds in recalcitrant black carbon (the microcharcoal particles), a process that plausibly contributed to enriching many trace element concentrations in BHs. Accordingly, Giovannini et al. (1988) proposed that distillation and volatilization processes of soil organic matter occur between 100 and 200 °C, whereas higher temperatures induce carbonization and structural variation in humic and fulvic acids (Almendros et al. 1990), with the formation of pyromorphic compounds that are less affected by microbial degradation. Fire would also lead to destruction of colloids, collapse of organomineral aggregates, and a subsequent increase in density, with loss of C and N (Natali et al. 2021) and an increase in metals such as Al, Fe, Ti, Cr, Ni, Sn, V, and Zn. Therefore, the dark color of BHs is due to both the microcharcoal concentration and the effect of organic matter burning and its consequent transformation.

6 Conclusions

BHs are recurrent in soil sequences in the areas surrounding Bologna at various depths. Their dark color is not related to an excess of organic matter, but to low concentration of carbonate and the peculiar presence of microcharcoal.

Our review of BHs in the areas surrounding Bologna provides a tool to evaluate climatic changes that occurred during the Late Glacial-Holocene period.

Geochemical, isotopic, and palynological/black particle evidence demonstrates that BHs invariably form over a time lapse of a few hundred years during cold/cool periods. This is corroborated by the comparison of the available BH radiocarbon age with climatic curves based on isotopic evidence (e.g., δ18O and residual Δ14C). In fact, these ages generally fit well with the cold periods retrieved by isotopic datasets. The studied BHs and their microcharcoal contents also provide interesting insights into the peculiar processes of vegetation fires, which seem to have been recurrent during the Late Glacial and the Holocene. Such a recurrence could have been of natural origin, but possibly also facilitated by early human activities, in particular from the Neolithic onward. Our research will continue tackling the same approach for more BHs in the surrounding areas to constrain the delineated hypotheses with more data.