Geochemical and mineralogical proxies for grain size in mudstones and siltstones from the Pleistocene and Holocene of the Po River alluvial plain, Italy
Published:January 01, 2007
- PDF LinkChapter PDF
Enrico Dinelli, Fabio Tateo, Vito Summa, 2007. "Geochemical and mineralogical proxies for grain size in mudstones and siltstones from the Pleistocene and Holocene of the Po River alluvial plain, Italy", Sedimentary Provenance and Petrogenesis: Perspectives from Petrography and Geochemistry, José Arribas, Mark J. Johnsson, Salvatore Critelli
Download citation file:
Recent studies carried out on fine-grained sediments recovered from boreholes in the eastern plain of the Po River demonstrate that significant mineralogical and geochemical changes in the provenance of sediments occurred in coincidence with the Pleistocene-Holo-cene transition. An increase in ultramafic-sourced sediment, related to more important inputs from the Po River, is evident at the beginning of the Holocene. The effects of grain-size distribution and provenance variation were investigated on recent unconsolidated sediments, mainly silts and clays. Sediments were collected from ten boreholes in the area, and the geochemical and mineralogical data were compared to the grain-size data. Among the chemical indexes, Zr/V, Y/Rb, Y/V, SiO2/Al2O3, Fe2O3/SiO2, Na/Al increase from pure clay to fine sand together with some mineralogical ratios, including quartz/interstratified illite-smectite and feldspar/interstratified illite-smectite. Some provenance indexes, both mineralogical and geochemical (Ni/Al, Cr/Al, serpentine/sheet silicates), were found to be independent from grain-size and are therefore valid for a wide textural range of sediments. Several geochemical and mineralogical proxies for grain size were identified. In the present case, all these indexes are independent from provenance influence and can be used as direct proxies for the grain size of the sediment, as confirmed by the multiple regression analysis performed to evaluate median and sorting. The equations included the most significant ratios and work well for median values <30 μm.
The chemical and mineralogical composition of sediments and sedimentary rocks is influenced by many factors, including source rock composition, climatic conditions in the watershed, the length and energy of sediment transport, redox conditions in the depositional environment, and grain size (e.g., Bhatia and Cook, 1986; Jarvis and Higgs, 1987; Argast and Donnelly, 1987; McLennan et al., 1990, 1993; Pearce and Jarvis, 1992; Johnsson, 1993; Jones and Manning, 1994; Garver and Scott, 1995; Fralick and Kronberg, 1997; Thomson et al., 1993; Lopez-Buendia et al., 1999; Dypvik and Harris, 2001; Ohta, 2004). Postdepositional processes such as weathering and diagenesis may alter the initial composition, so the study of recent unconsolidated sediments can provide an ideal situation to discriminate between grain-size and provenance signals using chemical and mineralogical parameters. In particular, changes in grain size in the composition of sediments and sedimentary rocks occur to the same extent as changes in provenance, weathering, and diagenesis (Bhatia, 1983; Johnsson, 1993), and their discrimination can represent a key step in the interpretation of provenance changes (see for example Ohta, 2004). Sedimentary particles are sorted according to dimension, shape, and specific gravity, creating variations in the chemical and mineralogical composition that can interfere with the signals derived from the source area, as documented, for example, by McLennan et al. (1993) for zircons in turbidites, which are in some cases affected by intense sedimentary sorting. Sorting effects are expected to be even greater in other depositional settings such as river channel deposits or beach sands, as reported for many elements (Zr, Ti, P, Cr, V, Y, Ce) by Garcia et al. (2004). For many of these elements, association with fine-grained sandstone and silt can be expected (e.g., McLennan, 1989; McLennan et al., 1993; Fralick and Kronberg., 1997; Dypvik and Harris, 2001), whereas other elements, such as V, Cr, Co, Ni, Cu, and Zn, are generally enriched in fine-grained sediments (Turekian and Wedephol, 1961; Horowitz, 1991).
The aim of this study was to investigate the relationships among compositional data (geochemistry and mineralogy) and grain size and sorting, with a focus on silt- and clay-sized sediments, even though some sands were included in our sampling. We discuss the significance of some geochemical ratios derived from bulk compositional data, as proxies for grain size, as well as possible combination of several parameters to quantitatively evaluate descriptive parameters of grain size (e.g., median and sorting). We selected for this study late Pleistocene to Holocene unconsolidated alluvial, transitional, and marine deposits in the eastern Po plain, northern Italy, the composition and evolution of which have been described in a previous paper (Amorosi et al., 2002). The study showed that changes in the composition of the sediment are related to different sediment sources, and a secondary aim of this work is to evaluate if the markers of provenance (e.g., Cr/Al2O3, Ni/Al, serpentine/sheet silicates) are influenced by grain-size variations.
The study area is located in northern Italy, in the eastern part of the Po River plain (Fig. 1), which is the largest alluvial plain in Italy and is drained by the east-flowing Po River. The Po drainage basin covers an area of ∼75,000 km2, and the river receives tributaries that drain both the Alps to the north and the Apennines to the south. Several minor rivers today flow directly into the Adriatic Sea from the Apennines, but during Pleistocene sea-level lowstand phases, they were tributaries of a major river system that flowed southward into the central Adriatic (Vai and Cantelli, 2004). Recent geological investigations associated with an extensive drilling campaign promoted by the Geological Survey of Regione Emilia-Romagna as part of the geological mapping project of Italy, have provided data for a better characterization of the late Quaternary deposits in the area. Amorosi et al. (1999b, 1999c, 2003, 2004) described in detail the architecture and evolution of the late Pleistocene–Holocene deposits (younger than 125 k.y.). They are characterized by a cyclic alternation of alluvial plain deposits of Pleistocene age overlain by a transgressive-regressive cycle that is made up of littoral to shallow-marine deposits of Holocene age. This cyclic evolution of sedimentary environments is fully exposed in the boreholes in the central Po River plain (cores labeled 187-S1, 204-S8, 205-S3, 205-S5, 205S-10), whereas in those cores closer to the Apennines chain (cores labeled 221-S19, 239-S2, and 240-S9), sedimentation is restricted to alluvial plain deposits.
MATERIALS AND METHODS
A total of 185 samples was collected from ten continuously cored boreholes in the eastern Po plain (Fig. 2), which have already been discussed in other works (Amorosi et al., 2002, 2007). Sampling was intended to characterize the fine-grained portions of the cores; it is irregularly spaced, but it describes the main facies recognized in the cores. The cores were sampled from the center in order to reduce possible contamination due to the coring operations. Slices of ∼2 cm were collected and processed, except for some heterogeneous intervals that required restricted thickness. Chemical analyses on all the samples were obtained by X-ray fluorescence spectrometry (Philips PW 1480) on pressed powder pellets with methods described by Amorosi et al. (2002). The estimated precision and accuracy for trace-element determinations were better than 5%, except for those elements at 10 ppm and lower (10%–15%). Semiquantitative mineralogical data were obtained by X-ray diffraction (XRD); details of the method are in Amorosi et al. (2002).
For the grain-size analyses, 0.5 g of sample were dispersed in 50 mL of deionized water and 5 mL of H2O2 (30%) and shaken until effervescence stopped. Particle-size analyses were performed using a Malvern Master Sizer/E granulometer equipped with a monochromatic beam of heliumneon laser (λ = 633 nm) and an optical design that enabled a measurement range between 0.1 and 600 μm. To evaluate the possible effect of calcite, 20 samples with variable carbonate content were reacted with diluted HCl (7%) until effervescence stopped, and then they were analyzed with the same analytical conditions. The use of cold diluted HCl removed mainly the calcite from the sample and did not affect dolomite.
In the statistical elaborations, samples with sand content >10% were not included, as well as all the samples from core 187-S1. The grain-size data from core 187-S1 will be presented in the following section and are used as a test for the proposed model.
Mineralogical, geochemical, and grain-size data for each sample are not reported here, but are available upon request from the authors.
RESULTS AND DISCUSSION
Grain-Size Data and Carbonate Effect
The investigated samples were for the most part mixtures of silt (63–4 μm) and clay (<4 μm) particles (Fig. 2). Only 4 of the samples had sand fraction (63 μm–2mm) > 10%, whereas 17 of the samples had a clay fraction >50%; this combination was present in every borehole but was more frequent in core 204-S1 (Fig. 3) 302. Samples from core 187-S1 were mostly silts and also included many sands (7 samples out of 14).
The descriptive parameters of the grain-size distribution (median and sorting) are presented in Figure 3 302. Most of the median values range between 4 and 16 μm (83% of the data), and only 9% percent of the samples have finer median values, and 8% have median values >16 μm. Sorting, evaluated following Inman (1952) as (16 percentile – 84 percentile)/2, directly follows the median values due to the occurrence of variable amounts of coarse-grained sediments that create wider dispersion in the grain-size distribution. There is no systematic association of these parameters with facies changes, which depends on the sampling strategy. However coarse-grained intervals occur as channel sands in the alluvial plain deposits and gravels in the cores closer to the Apennines (e.g., cores 221-S19, 240-S9, 239-S2 in Fig. 3 302). Other coarse-grained deposits are the beach sands in the transgressive barrier facies and marine sands in the delta front deposits, which characterize the cores located near the coastline (e.g., cores 205-S3, 205-S5, 205-S10 in Fig. 3 302). Part of these coarse-grained intervals were considered in core 187-S1 and are characterized by high median values and lower sorting.
Calcite is common in all the analyzed samples (see data in Amorosi et al., 2002), and it is particularly abundant in the cores close to the Apennines (221-S19, 240-S9, and 239-S2). Its removal produced contrasting results: in samples with low calcite content (∼20%), the acid attack did not produce macroscopic variation of the grain-size curve (Figs. 4A and 4B); other samples, with higher calcite abundance (up to 43%), showed more dramatic changes in size distribution (Figs. 4C–4F). All these samples, from core 239-S2 and characterized by high carbonate contents, showed a clear depletion of the finer grain sizes after carbonate removal.
Actually, the role of carbonate is twofold: it can be distributed randomly through all the grain sizes (leaving no effects after HCl digestion), and it can be concentrated in the clay fraction. The latter behavior, which could be surprising in cemented rock, is rather obvious for unconsolidated sediments. In our case, most of the calcite was detritic, originating from the sedimentary formations of the Apennines (Cavazza et al., 1993; Dinelli and Lucchini, 1998, 1999; Amorosi et al., 2002), and could be abundant in fine-grained fractions as a result of mechanical erosion and fragmentation during fluvial transport.
Grain-Size Relationships with Chemical and Mineralogical Data
Although better discrimination of the relationships between grain size and element distribution could be achieved through the analysis of separate grain-size fractions (Horowitz and Elrick, 1987; Weltje and Prins, 2003; Whitmore et al., 2004), the comparison of bulk compositional data with grain size has been applied. In the analysis of the trace-element distribution in sediments, both grain-size class abundance and statistical descriptive parameters of the grain-size distribution were considered (Horowitz, 1991; Zhao et al., 1999; Zhang et al. 2002; Viscosi-Shirley et al., 2003). We used these approaches to evaluate the possible effects of grain size on the distribution of minerals and chemical elements. A correlation analysis was carried out on the whole data set, and correlation coefficients were calculated among element concentrations, mineral abundances, grain size classes (sand, coarse and fine silt, and clay), and descriptive parameters of the grain-size distribution (Table 1).
For the large number of samples (n = 166), correlations are significant at the 0.01 confidence level at absolute values of 0.22. Among the variables positively and significantly correlated with the clay fraction, although their correlation coefficients do not reach absolute high values, there are Fe2O3, V, Cu, Zn, Rb, La, Ce, interstratified illite-smectite and kaolinite. Only V among the chemical elements, and interstratified illite-smectite and kaolinite among the mineralogical parameters have lower scattering. Also positively correlated to the clay fraction are Al2O3, K2O, TiO2, Co, Nb, and Th. These associations are expected, relating the more abundant sheet silicates in the <4 μm fraction to the chemical elements that usually concentrate in shales.
Amphiboles K- and Ca-Na micas, and Zr, Na2O, and P2O5 are positively correlated to the silt fraction, in particular to a coarse-silt fraction of the sediment (63–16 μm). The association of micas with silt agrees with the presence of coarse crystals within the sediment.
The association of Zr, P2O5, and Y is interpreted to be related to the occurrence of heavy minerals (zircon, garnet, monazite, apatite, hornblende) in fine-grained sediments, a feature which has been observed in other studies (Fralick and Kronberg, 1997; Dypvik and Harris, 2001; Garcia et al., 2004). These heavy minerals were not revealed by bulk XRD but are common in the heavy mineral fraction of alluvial sediments of the area (Gazzi et al., 1973; Gandolfiet al., 1982; Marchesini et al., 2000).
Quartz, feldspar, amphibole, and K-mica, and SiO2 and Na2O positively correlate to the sand fraction of the samples.
The parameters descriptive of the grain size (median and sorting) positively correlate to the compositional variables associated with the coarse fractions of the sediment, with particular high correlation with Zr and Na2O.
The results of these analyses open the possibility to propose and test geochemical and mineralogical proxies for bulk grain size of the sample that will be discussed in the following section. In particular:
-Na2O, SiO2, and amphiboles are representative of sands;
-amphiboles, Zr, K-mica, Na- and Ca-mica are representative of coarse silts;
-Zn, Fe2O3, Cu, Co, and V are representative of fine silts; and
-kaolinite, interstratified illite-smectite, V, Zn, Rb, and Fe2O3 are representative of clays.
Elements such as Cr and Ni did not correlate to any of the grain-size fractions or grain-size parameters (Table 1). It has been suggested that these elements were useful in the interpretation of sediment provenance in the area (Amorosi et al., 1999a, 2002) and clearly discriminate two populations in binary diagrams with Al2O3 (e.g., Figs. 5A and 5B). One population is influenced by sediments transported by the Po River (the black symbols in Fig. 5), and the other is mostly composed of sediment derived from the Apennines (open symbols in Fig. 5). These two populations cannot be clearly distinguished when other elements, such as V (Fig. 5C), are considered. Actually, if the data are subdivided according to the provenance attribution, the correlations became positive and significant (Figs. 5D and 5E), indicating that Cr and Ni are also controlled by the abundance of fine-grained material. Cr and Ni are controlled by the abundance of typical ophiolite-derived minerals, such as serpentine (Cr-serpentine correlation coefficient: 0.862; Ni-serpentine correlation coefficient: 0.856), or by the presence of ultramafic rock fragments.
Proxies for Grain Size
The results presented in the previous section show that there is strong partitioning of certain elements and minerals into different grain-size fractions, and so it is possible to derive several geochemical and mineralogical values that are useful proxies for grain size. We present some mineralogical proxies, but we are aware that published quantitative mineralogical data on fine-grained sediments are sporadic, even if the abundance of clay minerals by itself is a good indicator of the grain size of the sample. We concentrate the discussion mostly on geochemical proxies, based on both major and trace elements, which basically compare two or more elements that might have contrasting distribution compared to grain size. Some of the proxies have already been proposed and discussed in the literature (Argast and Donnelly, 1987; Herron, 1988; McLennan et al., 1993; Dypvik and Harris, 2001; Ohta, 2004); others have been derived from the results presented in Table 1, and their significance is evaluated here (e.g., Y/V, Zr/ V, Y/Rb). Moreover, those ratios that were particularly effective in the discrimination of sediment provenance (Cr/Al, Ni/Al, Mg/ Al) were tested to evaluate if their variation can be influenced by changes in grain size.
Some ratios significantly correlate with the sand content, such as SiO2/Al2O3 and SiO2/Fe2O3, and to a lower degree, Na2O/Al2O3, (Na2O + K2O)/Al2O3, (Zr + Ba)/Rb, and Ba/Rb (Table 2). The ratios involving major elements reflect the presence of quartz (SiO2) and feldspars (Na2O and K2O) compared to Al2O3, which is expected to be mostly controlled by the fine-grained sheet silicates. K2O can be present also in different types of sheet silicates, so its significance cannot be precisely defined. Trace-elements ratios are less correlated to the sand content, although the ratios (A, B, C, respectively) and versus clay (D, E, F), Two populations with different provenance (open involving Ba, particularly Ba/Rb, seem to correlate to the sand content. Ba can substitute for K in feldspars and might be representative of a relatively coarse fraction, whereas Rb is mostly concentrated in the clay fraction, substituting for K in illites (Dypvik and Harris, 2001). This location of Rb is consistent with its covariance with interstratified illite-smectite (r = 0.666), which represents the smallest mineral phase. These proxies, particularly those involving Na and K, but also Ba and Rb, might be sensitive to the degree of feldspar weathering that might cause remobilization of the elements and possibly concentration in the residual product. In this case, feldspars are common in the sediment, suggesting that weathering is not so strong and the proxies might be valid. The limited number of sandy samples prevents confident extrapolation of these proxies to our data set; however, many of them (SiO2/Al2O3, SiO2/Fe2O3, and Na2O/Al2O) are reported in the literature to be high in sands.
SiO2/Al2O3 and SiO2/Fe2O3 are positively correlated with median and sorting, whereas the other ratios involving major elements have lower significance.
Those ratios involving Zr and Y are representative of the coarse fraction (coarse silt) of fine-grained sediments (Table 2), particularly when normalized to elements associated to a fine-grained fraction (e.g., Al, Rb, V). Ratios such as Zr/Rb, Zr/V, Zr/ Al, Y/Al, Y/Rb, and Y/V correlate to high degree with the coarse silt fraction (Table 2) and reflect the occurrence of heavy minerals (zircon and garnet) in the sediments (Gazzi et al., 1973; Gandolfi et al., 1982; Marchesini et al., 2000). A similar explanation has been given for Zr/Rb by Dypvik and Harris (2001), but other ratios such as Zr/V, Y/Rb, and Y/V have the same significance and display slightly higher correlation coefficients with the coarse silt fraction. A slight difference is the higher coefficients of Zr/Rb and Zr/V to the sand fraction compared to the ratios involving Y. Ti is reported to be enriched in the coarse fraction of sediments in Aeolian sediments (e.g., Pye, 1987), in coarse sections of tur-bidites (e.g., Wehausen and Brumsack, 1999), or in extremely sorted clastic sediments (Garcia et al., 2004), so the ratio Ti/V is another good indicator of a coarse silt fraction of the sediment. Also some mineralogical indexes correlate to this fraction: amphibole/(interstratified illite-smectite), feldspar/(interstratified illite-smectite), and quartz/(interstratified illite-smectite). The first is controlled by the occurrence and abundance of a heavy mineral, whereas the others suggest that feldspars and quartz are preferentially concentrated into a coarse silt fraction.
Silt is the more important grain-size class in these samples, so all these ratios are important proxies. Zr/V, Zr/Rb, Y/V, and Y/Rb display large correlation coefficients (r > 0.750; Table 2) with median grain size, which is mostly in the silt range. The same ratio are also highly correlated with sorting of the sediment (Table 2).
Possible problems in the use of these ratios as grain-size proxies arise if the concentrations of the elements are altered by particular paleoenvironmental conditions, the occurrence of particular rock types, or if sorting of heavy minerals is particularly effective (Garcia et al., 2004). Under anoxic conditions and in sediments rich in organic matter (black shales and sapropels), V and Fe are significantly enriched (Vine and Tourtelot, 1970; Nijenhuis et al., 1999); furthermore, V has been used, in combination with Cr and Ni, as a paleoenvironmental indicator for redox state (Hatch and Leventhal, 1992; Jones and Manning, 1994). Fe concentrations could be influenced in reductive environments by precipitation of iron sulfides or carbonate in association with sediments rich in organic matter, but can be concentrated also in oxidizing conditions due to diagenetic remobilization (Froelich et al., 1979). Occurrence of alkaline volcanic rocks in the source area might alter the concentration on Zr, Y, and to a lesser degree K and Rb (Best and Christiansen, 2001), basaltic rocks might influence Fe, Ti, and V (Turekian and Wedephol, 1961), and ultramafic rocks can influence the distribution of Cr, Ni, and Mg, as already noted in the previous paragraph.
In the studied boreholes, Ni/Al, Cr/Al, and Mg/Al are the most significant provenance indicators, but as indicated in Table 2 and shown in detail in Figure 6, they do not show any correlation with the grain-size fraction or grain-size parameter. The negative correlation in the Mg/Al versus clay (Fig. 6C) shown by samples influenced by the Po River is related to the occurrence of detritic dolomite in samples poor in clay, which are more frequent in many of the cores close to the coastline that have a mixed provenance (Amorosi et al., 2002).
The identification of close relations between some geochemical indexes and grain-size classes and parameters describing the grain-size distribution, such as median and sorting, opens the possibility of using some geochemical indexes as direct proxy for grain size (Table 2). Cautions to the use a direct relation between a geochemical index and the median have been posed by Weltje and Prins (2003) because the grain-size distribution might be complex, and the median grain size might not fully represent the complete grain-size distribution observed. Sorting can help in the description of the sample heterogeneity and thus may account for complex grain-size distribution.
In order to further constrain the relation between median and sorting and geochemical indexes, we applied a stepwise multiple regression analysis to the data set, using the geochemical ratios as independent variables and median and sorting for dependent variables. In the model, all the ratios presented in Table 2 were initially introduced, including those more sensitive to provenance changes, which, however, did not increase the variance. The two resulting equations for median and sorting are:Fig. 7A), whereas for a larger median value, the model fails because it is not constrained. The same applies also to sorting: Equation 2 is calculated using samples with sorting values (expressed in μm) <30 μm, whereas in many samples of 187-S1, far larger values are observed.
We are aware that this approach is at the moment qualitative, and more data, particularly on coarse-grained sediments, are needed to provide more reliable estimates, but the data presented suggest that at least for fine-grained sediments (silts and clays) some geochemical ratios might be an effective and consistent proxy for the bulk grain size. In general, these results could provide useful insight into grain-size variations, particularly when working on siltstones and shales, for which lithification makes direct grain-size analyses difficult and petrographic tools can hardly help.
Effects of different grain-size distribution on mineralogical and geochemical parameters can be better analyzed in recent unconsolidated sediments that have not yet suffered postde-positional transformations related to diagenesis, compaction, and lithification. The comparison of bulk chemical and mineralogical composition with different grain-size classes in silts and clays has allowed the correlations between compositional and textural parameters to be outlined. The effectiveness of some mineralogical and geochemical ratios in describing the grain-size changes has been evaluated. Some indexes, such as quartz/interstratified illite-smectite; feldspar/interstratified illite-smectite; and SiO2/Al2O3, Na/Al, Zr/Rb, Zr/V, Y/Rb, Y/V, and Ti/V are more strongly related to a coarse silt fraction and have higher values in more coarse-grained samples. Multiple regression analysis combining median and sorting with the geochemical ratios was applied and tested with a subset of samples from an auxiliary core and proved to be particularly effective in the quantification. The model is limited to silt and clays, and it does not work well for more coarse-grained samples, for which further work is necessary. The indications derived from the combined evaluation of these proxies can provide important clues in the evaluation of textural features of the sediment when direct analysis of grain size is not available and might be extremely useful in the analysis of siltstones and shales, when application of other techniques, such as direct grain-size analysis or optical evaluation, are difficult.
When sediment provenance is considered, it is important to assess the effect of grain-size sorting in the key parameters used in the reconstruction. In the study area, several chemical and mineralogical indexes of provenance of the sediment (Ni/Al, Cr/Al, Mg/Al; serpentine/silicate) are not obscured or affected by grain-size variations, and, for this reason, they can be used independently of the texture of the sample considered. Only Mg/Al value displays a correlation with the abundance of a coarse silt fraction, suggesting that dolomite is mostly concentrated in this grain size.
We would like to acknowledge the editorial handling of Jose Arribas, and the useful comments of D. Garcia and M.R. Hounslow, which consistently improved the paper.
Figures & Tables
Sedimentary Provenance and Petrogenesis: Perspectives from Petrography and Geochemistry
- chemical composition
- clastic rocks
- correlation coefficient
- grain size
- mineral composition
- Po River
- Po Valley
- regression analysis
- sedimentary rocks
- Southern Europe
- statistical analysis