River infrastructures such as dikes, groynes, and dams are ubiquitous on most large rivers, and although their consequences on the riverbed morphology have often been studied, the effect they might have on the river floodplain and margins remains largely unknown. By investigating the structure and composition of floodplain sediments in three areas along the Rhône River that were extensively engineered during the last 150 years, this paper aims to understand whether river infrastructures might systematically induce a change in sedimentation patterns in the river margins.
Materials and methods
A total of fifteen sediment cores were sampled in three distinct reaches of the Rhône valley downstream of Lyon. They were thoroughly analyzed in terms of grain size, using heatmap representations and end-member analysis. Six metallic elements (Zn, Cr, Pb, Cu, Ni, Cd) were systematically quantified. In six out of the fifteen cores, total organic carbon (TOC), organic contaminants (PCBs) concentrations, and 137Cs activity were also assessed.
Results and discussion
A sharp change in grain size distribution is consistently identified in the sediment cores of all three study reaches. The sediments above this change are fine (D50 < 100 μm), poorly classified and homogenous. They also show a relative increase in contamination when compared with deeper sediments. We interpret this change in sediment characteristics as the consequence of an abrupt decrease in connectivity between the floodplain and the river channel, likely due to the implementation of navigation infrastructures in the channel in the second half of the nineteenth century. In one case, the dating of the sediment cores allows linking the grain size change to the implementation of navigation infrastructures in the channel. In the other study areas, the effect of engineering seems delayed in time due to local variability.
The implementation of river infrastructures resulted in a loss of connectivity between the floodplain and the channel. This was reflected as the homogenization of the floodplain sedimentary environment of the Rhône River all along its course from Lyon to the sea. A similar phenomenon might be present in most engineered rivers across Europe.
Ever since the industrial revolution and the technical developments it allowed, large-scale correction of river systems for purposes of navigation, agriculture, power production, or flood protection has taken place in Europe (Lewin 2013). As a result, infrastructures including dikes, groynes, weirs, and run-of-river dams are almost ubiquitous in European rivers (Tockner et al. 2009; Syvitski and Kettner 2011). The consequences of such river engineering are numerous and have been studied at different scales and through different scopes. A number of site-specific studies investigated, for example, the effects of groynes in terms of hydraulics (Przedwojski 1995; Giglou et al. 2017) or their repercussions on biodiversity (Buczyńska et al. 2018), the sedimentation patterns inside dikes or groyne fields (Lehotský et al. 2010; Savić et al. 2013), or the influence of weirs and run-of river dams on the local channel morphology (Csiki and Rhoads 2014; Poeppl et al. 2015; Howard et al. 2017). Most studies, however, question the effects of river engineering from a geomorphological point of view, at the catchment scale or taking into consideration several river systems. Surian and Rinaldi (2003) compiled the main types of channel adjustments in Italian rivers following human disturbances such as channelization and dam building, and highlighted that the two principal responses were channel incision and channel narrowing. Similar observations were made by Habersack et al. (2016) on the Danube River and by Zawiejska and Wyżga (2010) in the case of the Dunajec River in Poland. The effects of milldams on the morphology of river channels and associated floodplains in the eastern USA were also extensively studied: these dams are considered one of the main factors for excessive sediment accumulation that—along with channel incision—resulted in elevated and poorly connected floodplains (Walter and Merritts 2008; Merritts et al. 2011; Johnson et al. 2018), although this mechanism is still discussed (Donovan et al. 2016). More generally, James (2017) highlights the fact that human disturbances of fluvial systems—including engineering works—are “inhibitors”: factors that govern the channel and floodplain morphological evolution, often leading to the interruption of their geomorphic trajectory.
As a highly engineered system, the Rhône River has also been the subject of a number of studies about the consequences of several types of infrastructures on its morphology and evolution. It has been proven that river channelization through the implementation of numerous navigation infrastructures was the main cause for severe channel incision along most of the Rhône course (Petit et al. 1996; Parrot 2015), as well as in the Rhône delta where it led to lowering of the phreatic water table, soil salinization, and increased flood risk (Arnaud-Fassetta 2003). Provansal et al. (2014) also highlighted that the navigation infrastructure implementation at the end of the nineteenth century resulted in channel narrowing, bed incision, and storage of fine overbank sediments in the downstream part of the Rhône, effects that were further reinforced by spontaneous riparian afforestation and a decrease in sediment inputs from the drainage basins. The study also mentioned that those factors had a greater impact in transforming the river morphology than other human disturbances that affected the Rhône River such as dams and gravel mining.
Most of the aforementioned studies deal with the consequences of infrastructure implementation from a geomorphologic point of view, addressing the effects on the channel (e.g., narrowing, bed incision, armouring). In fewer instances, the consequences on floodplain morphology (e.g., aggradation, fine sediment accumulation, riparian vegetation development) are also addressed. However, as highlighted by Heritage et al. (2016), floodplains tend to be less studied in comparison with river channels, although they are one of the most degraded fluvial morphologic units and are known to provide numerous services (flood regulation, water quality improvement, rare and diverse habitats, etc.) when functioning normally (Entwistle et al. 2019). Besides, in this context, sediments are not commonly the focus point and, when mentioned, are often considered through a budget-type approach, i.e., by examining the inputs into the system versus the storage in the river margins or floodplain. To our knowledge, the effect of river infrastructures on floodplain sediments from a qualitative perspective (e.g., sediment structure, typology, geochemistry, contamination, etc.) has not been addressed yet. Through a comprehensive analysis of sediment cores sampled in floodplains from three engineered areas along the Rhône River, this study therefore aims to understand whether river infrastructures might systematically induce a modification of the floodplain sedimentary environment and, if appropriate, to qualify and understand the underlying mechanisms of such disturbances.
The Rhône River flows from the Furka glacier in Switzerland to the Mediterranean Sea in France (Fig. 1a). It is 812-km long, and its catchment covers 98,500 km2, of which 93% are in France. It is a heavily engineered system along its French reach: there are nowadays nineteen dams (one reservoir and eighteen run-of-river dams) as well as numerous groynes and dikes implemented for navigation or flood protection purposes, making the Rhône a relevant case study to investigate the effects of river infrastructures.
The three studied reaches—Pierre-Bénite (PBN), Péage-de-Roussillon (PDR), and Donzère-Mondragon (DZM)—are located respectively 6, 50, and 160 km downstream of the urban area of Lyon (Fig. 1a). They share a common history of regulation (Fig. 1b–d) that includes two main development phases. First, between roughly 1840 and 1910, the natural river was progressively corrected: flood protection dikes were built in the margins (roughly between 1840 and 1855), in-channel dikes and groynes were implemented to facilitate navigation, secondary channels were cut off (Räpple 2018). When the first navigation engineering works (1855–1884) proved relatively inefficient, a second type of in-channel infrastructure with a case-specific design that takes into account the natural hydraulic conditions was implemented: a combination of longitudinal submersible dikes and associated groynes that forms “squares” along the river banks and are known as “Casiers Girardon” (Räpple 2018). In this paper, for simplification, we will refer to the in-channel infrastructures built between 1855 and 1910 in general as “navigation infrastructures.” The second development stage took place in the second half of the twentieth century: the Compagnie Nationale du Rhône (CNR) built artificial canals equipped with a lock and a hydroelectric power plant parallel to the “old” channel, creating “bypass configurations” (Fig. 1b–d). A run-of-river dam was systematically implemented upstream of the “old” channel in order to derive most of the discharge inside the canal; in PDR for example, only a minimum flow of 50 to 125 m3 s−1 is maintained in the bypassed section, which represents a flow reduction by a factor of ten to twenty compared with its previous discharge.
The consequences of those engineering works from a hydromorphological point of view have been broadly studied at the river corridor and bypassed reach scales. Following the navigation infrastructures implementation, the Rhône went from a multi-channel system to a single, narrow channel with increased transport capacity, bed incision, and armouring, while the bypass implementation lowered peak flows and water level at low flows in the old channel, favoring vegetation encroachment and interrupting bedload transport (Petit et al. 1996; Bravard et al. 1999; Arnaud-Fassetta 2003; Vázquez-Tarrío et al. 2019).
Despite a common configuration and engineering history, the three study areas differ by their extent (PBN and PDR are 12-km long, while the bypassed area of DZM is 25-km long), the density of floodplain infrastructures (there are, for example, a large number of flood protection dikes in DZM compared with the other areas (cf. Fig. 1 b, c, and d)), and the timing of the second engineering phase: the bypass was completed in 1952 in DZM, 1972 in PBN, and 1977 in PDR.
Material and methods
The evolutions of the active channel of the Rhône River between 1860 and 2006–2009 in each of the three study areas are presented in the Electronic Supplementary Material 1. It derives from manual digitalization of historical maps and aerial photographs: the “Ponts et Chaussées” topographical atlas from 1860, the 1905 Branciard map, and photographs from the National Institute of Geographic and Forest Information (IGN-Geoportail). The data was produced in the context of the Rhône Sediment Observatory (http://www.graie.org/osr/), and the digitalization methods were fully described in Tena et al. (2017).
Sediment core sampling and description
Fifteen overbank sediment cores (five in each study area) were sampled with a Cobra TT percussion driller (direct-push action) powered with gasoline and fitted to drill up to a depth of 7 m. A closed gouge equipped with a transparent PVC liner (diameter, 40 mm) was used in order to facilitate visual inspection of the sample in the field. For cores longer than 1 m, several 1-m-long sections (at most) were sampled and sealed to allow transportation to the laboratory. Up to three 1-m-long cylindrical rods were used to retrieve the longest cores. The sediment cores detailed characteristics are summarized in the Electronic Supplementary Material 2. Their positions (Fig. 1a–c) were determined after the chronological analysis of the areas, with the aim of sampling in localizations that were part of the active river margins before the implementation of infrastructures (i.e., around 1840), but also of recording a variety of sedimentary environment (floodplain, former active channels, former islands, etc.).
In the PBN area, core C3c is located on the right bank close to the present channel, on an area that used to be part of the riverbed and recently became terrestrial (post-1954). Cores C15a and C15b are on a former island that became part of the right bank after the first development phase. Cores C18a and C18b are located on the right bank, in a zone that became terrestrial also following the navigation infrastructures implementation.
In the PDR area, cores C10a, C10b, and C10c are located on the same transect perpendicular to the riverbed. Cores C10a and C10c are assumed to represent a typical floodplain deposition environment, while core C10b is located in an infilled channel. Core C12a and C12b are located on a left bank transect, between the main channel and a secondary channel that is semi-aquatic nowadays, in an area of former migrating channels.
In the DZM area, core C2 is located on the left bank, roughly 1 km from the current river channel but close to a former secondary arm of the Rhône River. Core C3 is located on the right bank close to a secondary branch still active nowadays. Core C6 is on the left bank alongside a small secondary arm. Core C8 is located on the right bank, relatively close to the current main channel but in an area protected by dikes since the 1850s and therefore rarely flooded. Finally, core C9 is on a former island that joined the right bank following the navigation infrastructures implementation.
Right after their opening, photographs of the cores were taken and a visual description (lithofacies, grain size global characteristics) was realized.
Grain size analysis
It appeared from the visual description that the sediments in all cores were mostly in the silt/sand grain size ranges (i.e., roughly comprised between 20 μm and 2 mm). We were therefore able to measure the grain size every 2 to 4 cm with a Mastersizer 2000© (Malvern Panalytical) particle size analyzer mounted with a hydro SM small volume dispersion unit. Descriptive grain size statistics (D50, mode, D10, D90, skewness, etc.) were computed with the Gradistat software (Blott and Pye 2001), using the grain size classes configured in the laser particle size analyzer (a hundred classes from 0.011482 to 1000 μm). In order to extensively represent the grain size distributions (GSDs), “heatmaps” of the cores were plotted, in which for every measured sample, the percentage of grain size classes are represented by a color gradient. Finally, unmixing of the sets of GSDs was carried out through end-member modelling analysis (EMMA) with the R package EMMAgeo (Dietze et al. 2012). It is a compositional method based on the assumption that any GSD is a mixture of sediment populations corresponding to different mechanisms of production and/or transport (Weltje and Prins 2003). A linear mixing model is therefore applied to an array of GSDs and allows to identify meaningful end-members (EMs) that correspond to the most representative populations of the GSDs (Weltje and Prins 2007). The EMs can then be interpreted as different sedimentation processes (Dietze et al. 2012) and/or transport mechanisms (Toonen et al. 2015; Dietze et al. 2016). Here, the model was applied separately for each study area, as the significance of EMs can only be judged relatively to an array of GSDs from the same system (Weltje and Prins 2007). For better visualization of the results, the end-member scores (percentages of representation of the different end-members) were plotted as a function of the depth for each core.
A pseudo-total analysis of six elements (Zn, Cr, Pb, Ni, Cu, Cd) was carried out on all cores through an aqua-regia procedure. Samples were collected at a step of 4 to 6 cm, dried at 40 °C during 48 h, and sieved at 2 mm. They were mineralized in a CEM/MARS Xpress microwave with 2 mL of HNO3 and 6 mL of HCl. The measurements were performed on 25 mL of the mineralized sample after its filtration, with a Perkin Elmer PinAAcle 900T atomic absorption spectrometer. Zn was quantified in an air-acetylene flame, while the rest of the metallic elements were dosed in a graphite furnace (temperature varying depending on the element). Detection limits range from 5.5 to 35 μg kg−1 DW.
The seven indicator PCBs (congeners 28, 52, 101, 118, 138, 153, 180) were quantified in the LABERCA laboratory (Nantes, France) on six out of the fifteen cores (PBN-C3c, PBN-C18b, PDR-C10a, PDR-C10b, DZM-C6, DZM-C9). Only the top 70 cm of sediment cores DZM-C6 and DZM-C9 were sampled as we estimated that sediments below this depth would be too old to have recorded any PCBs. The samples were collected at a step of 4 to 6 cm, sieved at < 2 mm, freeze-dried in an overpressure room, packaged in amber glass vials, and sent to the laboratory for further analysis. After extraction and purification procedures, PCBs analyses were carried out by gas chromatography coupled with high-resolution mass spectrometry (GC/HRMS) using a 7890A gas chromatograph (Agilent) coupled with a JMS 800D double-sector high-resolution mass spectrometer (JEOL, Tokyo, Japan). The limit of detection (LOD) ranges from 0.049 (PCB 118) to 0.109 (PCB 28) μg kg−1 DW, whereas the recoveries range from 60 to 120%.
Rock-Eval pyrolysis was used to analyze the total organic carbon (TOC) as this method proved to be suitable for elemental analysis in sediments, even at low values of TOC (Meyers 1997; Steinmann et al. 2003; Carrie et al. 2009; Carrie et al. 2012). The analyses were carried out at the ISTO laboratory (Orléans, France) on the six cores used for PCBs analyses, with a sampling step of 4–6 cm.
Sediment core dating
Cesium-137 (137Cs) activity was assessed on the six cores sampled for PCBs analyses in the LSCE (Laboratoire des Sciences du Climat et de l’Environnement, Gif-sur-Yvette, France) and LMRE (Laboratoire de Métrologie de la Radioactivité dans l’Environnement, Orsay, France) laboratories for PDR and DZM/PBN respectively. Samples were collected with a step of 4 to 8 cm, dried for 3 days at 60 °C, and preserved in polystyrene boxes. Sub-samples from each core were analyzed by counting for at least 24 h, using low-noise gamma spectrometry. Gamma emissions were detected with a germanium detector and used to quantify specific activities of 226Ra, 228Ra, 40K, 210Pb, and 137Cs (Pinglot and Pourchet 1995). 137Cs is an anthropogenic radionuclide that originates mainly from atmospheric nuclear weapons testing (1945–1972) and major nuclear accidents (Chernobyl, Fukushima) (Audry et al. 2004; Abril et al. 2018). It is a good indicator of sediment processes as it binds almost irreversibly to clay and silt material (Francis and Brinkley 1976) and has a half-life of around 30 years. In Europe, 137Cs dating will typically allow identifying two events that appear as activity peaks: the period of nuclear testing that peaks in 1962–1963 (Ritchie and McHenry 1990) and the Chernobyl accident in 1986.
Sediment cores description and grain size results
The sediment cores measure between 53 and 290 cm. Most present a common sedimentary pattern: from top to bottom, a layer of organic topsoil, then an alternating of mix of silt and fine sand, and finally a layer of coarser sand sometimes mixed with gravels (Fig. 2). Some cores however stand out for being mainly composed of sand (e.g., DZM-C2, DZM-C8, PBN-C18a).
The heatmap representations highlight a sudden change in the grain size distributions that can be observed in most cores (represented by the white dotted line). The sediments above this limit are uncommonly homogenous, sharing the following characteristics: composed of silt or a mix of fine sand/silt (D50 < 100 μm), poorly classified and with a rather monotonous D50 and/or mode.
The EMMA algorithm optimally computed four comparable end-members in the three study areas (Fig. 2):
- EM1 is strongly bimodal, with the main mode at 10–35 μm (depending on the area) and the second mode around 200–300 μm.
- EM2 is slightly bimodal in PDR and DZM, with the prevailing mode at 140 μm in PDR and 52 μm in DZM, and the second mode around 500 μm in both areas. In PBN, EM2 is unimodal (~ 70 μm) and skewed to the left with a large standard deviation.
- EM3 is a unimodal distribution strongly skewed to the left and centered around 240, 280, and 160 μm for PBN, PDR, and DZM respectively.
- EM4 is a bimodal distribution. In PDR and DZM, the main mode is at 630 and 360 μm respectively, and the less represented mode around 80–100 μm. In PBN, the main mode is at 1100 μm but a second mode—broadly centered around 100 μm and skewed to the left—represents a significant portion of the distribution.
The grain size change identified on the heatmaps can also be recognized on the EM score graphs as a sharp modification of the EMs distribution, with EM1—and to a lesser extent EM2—becoming predominant: the bimodal and widely spread distribution of EM1 seems to be representative of the homogenous and poorly classified top sediments.
The geochemical profiles of cores PBN-C18b, PDR-C10a, and DZM-C9 only are presented in Fig. 3; the metallic elements (Zn, Cu, Pb), 137Cs, and PCBi profiles of three other cores are available in the Electronic Supplementary Material 3.
In the cores where it was measured, the TOC ranges between 0 and 3% (Fig. 3). The profiles are rather monotonous in the bottom of the cores where the organic content is generally low (< 1%). The TOC starts to increase significantly around 30–40 cm from the surface, showing the progressive transformation of the top sediments into a soil.
The metallic element concentrations are low to moderate in the three studied reaches, significantly below the corresponding probable effect concentrations (PEC) of MacDonald et al. (2000). Boxplots of the concentrations for each of the six elements (Zn, Cr, Pb, Cu, Ni, Cd) in the three study areas are available in the Electronic Supplementary Material 4. The vertical profiles show a relative increase in concentration from the bottom to the top of the cores, indicating enrichment in those metallic elements overtime that is likely due to the increasing industrialization and urbanization of the catchment. Similar trends were observed in other sediment cores from the southern part of the Rhône River basin (Ferrand et al. 2012; Cossa et al. 2018), as well as in cores from other French catchments (Audry et al. 2004; Meybeck et al. 2007; Le Cloarec et al. 2011; Dhivert et al. 2016), with concentrations generally peaking in the 1960s and then decreasing from the 1970s–1980s onward due to efforts to reduce heavy metal releases in the environment. No decreasing trend in metallic element concentrations is observed in those cores, though, except for Pb in DZM-C9. In PDR, the upward trend might also be explained by inputs from the Gier River, a tributary with a long history of metal industry that joins the Rhône River ~ 30 km upstream from the study site. We can notice that in the cores where an abrupt grain size decrease was identified, it almost systematically coincides with an increment in the concentration profiles (Fig. 3). This phenomenon might be partly explained by matrix effects as contaminants—including metallic elements—bind preferentially to fine particles (Karickhoff et al. 1979; Wen et al. 1998)
None of the seven indicator PCBs is detected on the major part of the cores (Fig. 3): the first significant concentrations are at 30, 35, and 50 cm from the surface in cores PBN-C18b, PDR-C10a, and DZM-C9 respectively. We observe then a rapid increase up to ~ 20 μg kg−1 ∑7PCBi. These concentrations are well below the sediment quality guidelines (SQGs) established in MacDonald et al. (2000) (probable effect concentration, 676 μg kg−1 DW of total PCBs; and threshold effect concentration, 59.8 μg kg−1 DW of total PCBs).
The 137Cs activity is recorded starting around 30–50 cm from the surface—which coincides with the emergence of the PCBi concentrations—and increases up to 15–20 Bq kg−1 close to the surface (Fig. 3). In the case of DZM-C9, the activity reaches its maximum (~ 15 Bq kg−1) around 30 cm from the surface and stays constant on the rest of the core. We can notice that in the three presented cores, the patterns of the 137Cs and PCBi profiles are very similar. It has been verified that those trends were not solely explained by a matrix effect related to the grain size or the TOC, although for some cores, the TOC showed a rather good correlation (R2~0.7) with the 137Cs (e.g., PBN-C18b) or the PCBi (e.g., DZM-C9). Redistribution (leaching) of the contaminants is also unlikely in this context.
Sediment core dating
The 137Cs profiles do not present the two activity peaks typically expected in sediment sequences of the last 70 years (Section 3.5). Instead, the emergence of a 137Cs activity followed by a steady increase can be observed in the top parts of the cores (Fig. 3). As no 137Cs activity is detected below, we assume that the recorded activity corresponds to the beginning of the period of nuclear testing (1945–1972). The portions of the cores with recorded 137Cs activity therefore date back from the 1950s or later. Due to the lack of typical 137Cs trend and the relatively few measurement points, it is difficult to go any further in the chronological interpretation of these profiles.
The concentrations in PCBi can also be used as time markers. PCBs are anthropogenic contaminants that were introduced in France in 1929 and were used until 1987 when they were entirely banned from use in any context (Amiard et al. 2016). Due to this restricted period of production, vertical profiles of PCB concentrations typically follow a common pattern: the emergence of the contamination corresponds to the 1940s–1950s and is followed by a steady increase in concentrations, the peak contamination can be expected in the 1970s–1980s, and a decreasing trend is observed afterwards in relation with the PCBs ban (Breivik et al. 2002). Such patterns have been observed in other sediment cores from the Rhône River margins (Desmet et al. 2012; Mourier et al. 2014). In our case, the PCBi concentrations started to be recorded at the same depth as the 137Cs activity, which makes sense as both substances were introduced in the environment roughly at the same time: this confirms that the top parts of the cores date back from the 1950s or later (Fig. 3). It may be observed that the grain size change is systematically below the emergence of the 137Cs activities or PCBi concentrations, meaning that it is older than 1950.
However, no decreasing trend is observed in the PCBs profiles, and the relatively low concentrations (a maximum of 20 μg kg−1 DW) suggest that the peak contamination was not recorded either; indeed, PCBi concentrations up to 417 μg kg−1 have been measured in aquatic sediment cores from the Rhône river in DZM (Mourier et al. 2014). The second expected peak (Chernobyl accident in 1986) in the 137Cs profiles was not recorded either. Moreover, we mentioned previously the absence of decreasing trend in the metallic element profiles, although such a pattern is commonly observed from the 1970s–1980s onwards in sediment cores from the Rhône River (see Section 4.2). Those results suggest that sedimentation might have stopped before the 1980s, failing to capture the decreasing trend in PCBs and metallic elements or the Chernobyl-linked 137Cs peak. This implies that the sediment recording stopped in the 1980s or before.
Link between the sediment cores sequences and their deposition mechanisms and environments
A change in grain size repartition—from sandy well-classified deposits to fine-grained and poorly sorted sediments—was identified in nine out of the fifteen studied sediment cores (see Section 4.1 and Fig. 2). We interpret such a systematic and homogenous pattern in the most recent sediments as a result from a common perturbation in a given system, meaning that the change happened roughly at the same time in the cores of the same study area. To our knowledge, no such sediment typology modification has been previously described in the literature. Fine sediment delivery and accumulation associated with land clearance and agriculture have been observed in several European floodplains (Passmore and Macklin 1994; Brown et al. 2013, Brown et al. 2018). However, it is unlikely that the same process is responsible for the fining grain size in our case: the grain size change appears too recent to correspond to the development of agriculture in the catchments. Magilligan (1992) explained a fining upward sequence in floodplain sediments in Wisconsin (USA) by the progressive aggradation of the floodplain that caused its elevation to exceed the competency of most floods to transport coarse materials from the channel to the banks. A similar mechanism—i.e., a loss of lateral connectivity between the floodplain and the river—could be a realistic explanation for the observed grain size modification. This assumption is supported by the end-member analysis: the bottom of the cores is mainly characterized by EM2, EM3, and EM4, well-sorted distributions of sands and/or coarse silt with a marked main mode. Those characteristics are representative of a rather dynamic deposition process. After the sharp grain size transition, EM1 is predominant: the bimodal and widely spread distribution reflects a calmer deposition environment and suggests suspension-related sedimentation due to the decreased connectivity with the main channel. The grain size parameters then stay constant but the TOC enrichment near the surface show that the connectivity loss quickly resulted in the evolution of the sedimentary environment towards a soil. Finally, it seems that a total loss of connectivity between the floodplain and the riverbed is reached before the 1980s as no sediment deposition is recorded starting around this period (cf. section 5.1).
Six out of the 15 studied cores do not present the grain size change induced by the loss of connectivity, which can be generally explained by their initial deposition environment. In the case of PBN-C3c, for example, it is likely due to the fact that the sediment core is more recent than the others. Indeed, the chronological analysis (Electronic Supplementary Material 2) shows that the core localization was still underwater in 1954, and the PCBs and 137Cs profiles (Electronic Supplementary Material 3) suggest that the bottom of the core dates back from the 1970s, while the perturbation that induced the grain size change is older than the 1950s (cf. Section 5.1). In the case of DZM-C2, it might be explained by the fact that the core is both distant from the current main channel (~ 1 km) and behind non-submersible dikes, some having been implemented in 1860: the core has probably been disconnected prior to the period we are interested in. The same explanation might apply to DZM-C8, which is located closer to the channel but behind flood protection dikes. Finally, the absence of grain size change in cores PBN-C18a and DZM-C3 could be due to their close proximity to the main channel and an active secondary channel respectively, which might still allow deposition of coarser sediments.
Chronology and consequences of the infrastructure-induced connectivity loss
Contrary to what might have been previously described in the literature (Magilligan 1992), the grain size change we observed is far from being progressive, which suggests that the event responsible for the loss of connectivity is potentially of human origin. From the chronological analysis of the areas, we know that two phases of extensive engineering could be the cause of such an abrupt change: the navigation infrastructures implementation (1855–1910) or the dam and bypass construction (second half of the twentieth century). We also know that the grain size change happened earlier than the 1950s (Section 5.1), and therefore cannot be due to the bypass implementation that took place after 1950. This leaves the possibility of the Girardon infrastructures being responsible for the sediment modification. Whether this hypothesis is chronologically consistent or not depends on the study area.
In PDR, the grain size change is approximately 45 cm before the 1950s time markers (137Cs and PCBi). Assuming that it was caused by the navigation infrastructures implementation (between 1840 and 1900 according to Michelot 1983) would imply an average sediment accumulation rate of 0.4 to 0.9 cm year−1, which is low but plausible for a floodplain, especially considering the decrease in hydraulic connectivity. In addition, we know that sedimentation likely stopped before the 1980s as no PCBs decontamination trend or second 137Cs activity peak was recorded in neither of the cores (Section 5.1). The dam and bypass were achieved in 1977 in PDR, resulting in a discharge reduction by a factor of ten to twenty in the old Rhône and consequently a decrease in the flood magnitude and frequency (Dépret et al. 2017; Vázquez-Tarrío et al. 2019). From a chronological and hydromorphological point of view, this second engineering phase could therefore be the cause of the halting of overbank sedimentation.
In PBN and DZM, the chronology is not so straightforward. The grain size limit appears around 20 cm before the 1950s time markers in both PBN-C18b and DZM-C9. However, a majority of the navigation infrastructures were completed by 1896 in PBN; in DZM, they started to be built before 1860 and the latest dates back from 1910 (Gaydou 2013; Räpple 2018). It is therefore unlikely that the grain size appeared at that time, as it would imply inconsistent sedimentation rates: < 0.4 cm year−1 between 1840-1900 and the 1950s, then between 0.8–1.5 cm year−1 between the 1950s and the 1980s, while it would be expected that the loss of connectivity would rather lead to decreasing rates as less and less sediment accumulates. However, given the engineered system the sediment cores are in and the abruptness of the change, it is also unrealistic to think that the loss of lateral connectivity happened naturally. We believe that it was still induced by human engineering but somehow delayed in time.
The difficulty in understanding precisely the mechanisms that led to the loss of lateral connectivity is that the hydraulic and morphological initial conditions are highly variable along the sediment cores considered, due to the numerous human and natural perturbations that shaped the floodplains overtime. Although the three study areas have a similar configuration and river engineering history, they differ in terms of extent, average discharge, anthropic settlement, activities, etc. Even inside a given study area, the response to a common perturbation might vary due to the local context. When considering connectivity, the distance of a site from a secondary channel—even a minor or dried-up one—will have a strong impact. Indeed, in case of flood, it will constitute a preferential water path and the areas close by are therefore likely to be connected frequently by the overflow of the channel. This might be the case in core DZM-C3, for example, which is located a few meters from a semi-active secondary channel and in which the grain size repartition and the thickness of sediments accumulated suggest that the site is still frequently connected. Moreover, in all study areas, the Girardon infrastructures shapes and functions were specifically designed based on the initial riverbed morphology and therefore differ depending on the local context; the detailed chronology of their implementation also remains often unclear. In the riverbed next to the semi-island where core DZM-C9 was sampled, for instance, groynes were added in 1910, i.e., some decades after the implementation of most navigation infrastructures, which could have locally and belatedly triggered the loss of connectivity recorded in the core.
Overall, it is indisputable that a loss of connectivity between the floodplain and the channel was induced following the implementation of the navigation infrastructures (1855–1900) and was recorded as an abrupt change in the grain size typology that can be observed in all three study areas. In PDR, the connectivity loss took place shortly after the engineering phase while it seems to have been delayed in the other areas, likely due to local factors.
This sudden anthropogenic modification of the floodplain environment might have impacted its sustainability in several ways. We demonstrated that the fine sediments above the grain size change tended to be more contaminated than deeper sediments, which may pose a risk in case of resuspension of those particles during floods. Indeed, it has been proven that 89% of the Rhône solid transport load was composed of fines (Vázquez-Tarrío et al. 2019) and that flood events were usually characterized by important fluxes of contaminants (metallic elements, radionuclides) bound to those fine particles (Ollivier et al. 2011; Antonelli et al. 2008; Ferrand et al. 2012). In our case, however, the contamination hazard is greatly reduced due to the fact that the bypass construction induced a quasi-total loss of connectivity between the studied banks and the channel. The loss of connectivity might also impact biodiversity on the riverbanks: Räpple (2018) found that most Girardon navigation infrastructures evolved from aquatic to terrestrial overtime as they aggraded with fine sediments. The forests that developed on those filled infrastructure present compositional characteristics closer to mature than to pioneer systems, with a prevalence of non-native species such as the invasive Box elder (Acer negundo). This shift in biodiversity is mainly attributed to the loss of connectivity between the banks and the main channel that transformed partly aquatic riverbanks into fully terrestrial environments. We can suspect that a comparable shift might have taken place on the studied floodplains similarly affected by loss of connectivity, although further studies would be necessary to ascertain it.
We investigated the floodplain sedimentary environment in three areas along the Rhône River that underwent two stages of river engineering: a progressive channelization through building of groynes and dikes in the second half of the nineteenth century and the implementation of a bypass that induced a drastic flow reduction in the historical channel during the second half of the twentieth century. Through the extensive analysis of multiple sediment cores sampled in the floodplain, we found a common pattern in the grain size sequences. While the bottom of the cores is characterized by coarse and well-sorted deposits, the sediment then abruptly becomes finer, poorly classified, and uncommonly homogenous. This modification of the sediment typology is indicative of a sudden decrease in connectivity between the floodplain and the channel that was triggered by the first engineering phase. Time markers such as 137Cs and PCBs show that the grain size change happened before the 1950s in the three studied areas and allow to chronologically link it to the implementation of the navigation infrastructures in one of the localizations. In the other two, the loss of connectivity seems to have been delayed in time but remains indisputably a consequence of the first river engineering phase. The time markers also indicate that floodplain sedimentation stopped altogether prior to the 1980s in the three areas, which suggests that the floodplain and riverbed became totally disconnected following the second engineering phase. The variable local characteristics amongst the sediment cores—in terms of hydraulics, morphology, infrastructure type, etc.—presumably affected the chronology of both phases of disconnection. Nevertheless, we proved that the effects of river infrastructure implementation overshadowed the local variability to result in a similar sediment pattern in three different areas along the Rhône River, which constitutes an issue as the floodplain heterogeneity and connectivity are essential to sustain its ecological diversity and numerous ecosystem services. This phenomenon of infrastructure-induced loss of connectivity and resulting homogenization of the sedimentary environment might be present all along the channelized Rhône, as well as in most engineered rivers across Europe.
Abril JM, San Miguel EG, Ruiz-Canovas C et al (2018) From floodplain to aquatic sediments: radiogeochronological fingerprints in a sediment core from the mining impacted Sancho Reservoir (SW Spain). Sci Total Environ 631–632:866–878
Amiard J-C, Meunier T, Babut M (2016) PCB, environnement et santé. Lavoisier, Paris
Antonelli C, Eyrolle F, Rolland B, Provansal M, Sabatier F (2008) Suspended sediment and 137Cs fluxes during the exceptional December 2003 flood in the Rhone River, Southeast France. Geomorphology 95:350–360
Arnaud-Fassetta G (2003) River channel changes in the Rhone Delta (France) since the end of the Little Ice Age: geomorphological adjustment to hydroclimatic change and natural resource management. Catena 51:141–172
Audry S, Schäfer J, Blanc G, Jouanneau J-M (2004) Fifty-year sedimentary record of heavy metal pollution (Cd, Zn, Cu, Pb) in the Lot River reservoirs (France). Environ Pollut 132:413–426
Blott SJ, Pye K (2001) GRADISTAT: a grain size distribution and statistics package for the analysis of unconsolidated sediments. Earth Surf Process Landf 26:1237–1248
Bravard J-P, Landon N, Peiry J-L, Piégay H (1999) Principles of engineering geomorphology for managing channel erosion and bedload transport, examples from French rivers. Geomorphology 31:291–311
Breivik K, Sweetman A, Pacyna JM, Jones KC (2002) Towards a global historical emission inventory for selected PCB congeners — a mass balance approach: 1. Global production and consumption. Sci Total Environ 290:181–198
Brown A, Toms P, Carey C, Rhodes E (2013) Geomorphology of the anthropocene: time-transgressive discontinuities of human-induced alluviation. Anthropocene 1:3-13
Brown AG, Lespez L, Sear DA, Macaire JJ, Houben P, Klimek K, Brazier RE, van Oost K, Pears B (2018) Natural vs anthropogenic streams in Europe: history, ecology and implications for restoration, river-rewilding and riverine ecosystem services. Earth-Sci Rev 180:185–205
Buczyńska E, Szlauer-Łukaszewska A, Czachorowski S, Buczyński P (2018) Human impact on large rivers: the influence of groynes of the River Oder on larval assemblages of caddisflies (Trichoptera). Hydrobiologia 819:177–195
Carrie J, Sanei H, Goodarzi F, Stern G, Wang F (2009) Characterization of organic matter in surface sediments of the Mackenzie River Basin, Canada. Int J Coal Geol 77:416–423
Carrie J, Sanei H, Stern G (2012) Standardisation of Rock–Eval pyrolysis for the analysis of recent sediments and soils. Org Geochem 46:38–53
Cossa D, Fanget A-S, Chiffoleau J-F, Bassetti MA, Buscail R, Dennielou B, Briggs K, Arnaud M, Guédron S, Berné S (2018) Chronology and sources of trace elements accumulation in the Rhône pro-delta sediments (Northwestern Mediterranean) during the last 400years. Prog Oceanogr 163:161–171
Csiki SJ, Rhoads BL (2014) Influence of four run-of-river dams on channel morphology and sediment characteristics in Illinois, USA. Geomorphology 206:215–229
Dépret T, Riquier J, Piégay H (2017) Evolution of abandoned channels: insights on controlling factors in a multi-pressure river system. Geomorphology 294:99–118
Desmet M, Mourier B, Mahler BJ, van Metre PC, Roux G, Persat H, Lefèvre I, Peretti A, Chapron E, Simonneau A, Miège C, Babut M (2012) Spatial and temporal trends in PCBs in sediment along the lower Rhône River, France. Sci Total Environ 433:189–197
Dhivert E, Grosbois C, Courtin-Nomade A, Bourrain X, Desmet M (2016) Dynamics of metallic contaminants at a basin scale — spatial and temporal reconstruction from four sediment cores (Loire fluvial system, France). Sci Total Environ 541:1504–1515
Dietze E, Hartmann K, Diekmann B, IJmker J, Lehmkuhl F, Opitz S, Stauch G, Wünnemann B, Borchers A (2012) An end-member algorithm for deciphering modern detrital processes from lake sediments of Lake Donggi Cona, NE Tibetan Plateau, China. Sediment Geol 243–244:169–180
Dietze M, Dietze E, Lomax J, Fuchs M, Kleber A, Wells S (2016) Environmental history recorded in aeolian deposits under stone pavements, Mojave Desert, USA. Quat Res 85:4–16
Donovan M, Miller A, Baker M (2016) Reassessing the role of milldams in Piedmont floodplain development and remobilization. Geomorphology 268:133–145
Entwistle NS, Heritage GL, Schofield LA, Williamson RJ (2019) Recent changes to floodplain character and functionality in England. Catena 174:490–498
Ferrand E, Eyrolle F, Radakovitch O, Provansal M, Dufour S, Vella C, Raccasi G, Gurriaran R (2012) Historical levels of heavy metals and artificial radionuclides reconstructed from overbank sediment records in lower Rhône River (South-East France). Geochim Cosmochim Acta 82:163–182
Francis CW, Brinkley FS (1976) Preferential adsorption of 137Cs to micaceous minerals in contaminated freshwater sediment. Nat 260:511–513
Gaydou P (2013) Schéma directeur de ré-activation de la dynamique fluviale des marges du Rhône. Rapport de synthèse OSR. Lyon.
Giglou AN, Mccorquodale JA, Solari L (2017) Numerical study on the effect of the spur dikes on sedimentation pattern. Ain Shams Engineer J 9:2055–2067
Habersack H, Hein T, Stanica A, Liska I, Mair R, Jäger E, Hauer C, Bradley C (2016) Challenges of river basin management: current status of, and prospects for, the River Danube from a river engineering perspective. Sci Total Environ 543:828–845
Heritage G, Entwistle NS, Bentley S (2016) Floodplains: the forgotten and abused component of the fluvial system. E3S Web Conf 7. https://doi.org/10.1051/e3sconf/20160713007
Howard AJ, Coulthard TJ, Knight D (2017) The potential impact of green agendas on historic river landscapes: numerical modelling of multiple weir removal in the Derwent Valley Mills world heritage site. UK Geomorphology 293:37–52
James LA (2017) Arrested geomorphic trajectories and the long-term hidden potential for change. J Environ Manag 202:412–423
Johnson KM, Snyder NP, Castle S, Hopkins AJ, Waltner M, Merritts DJ, Walter RC (2018) Legacy sediment storage in New England river valleys: anthropogenic processes in a postglacial landscape. Geomorphology 327:417–437
Karickhoff SW, Brown DS, Scott TA (1979) Sorption of hydrophobic pollutants on natural sediments. Water Res 13:241–248
Le Cloarec M-F, Bonte PH, Lestel L et al (2011) Sedimentary record of metal contamination in the Seine River during the last century. Physics Chem Earth, Parts A/B/C 36:515–529
Lehotský M, Novotný J, Szmańda JB, Grešková A (2010) A suburban inter-dike river reach of a large river: modern morphological and sedimentary changes (the Bratislava reach of the Danube River, Slovakia). Geomorphology 117:298–308
Lewin J (2013) Enlightenment and the GM floodplain. Earth Surf Process Landf 38:17–29
MacDonald DD, Ingersoll CG, Berger TA (2000) Development and evaluation of consensus-based sediment quality guidelines for freshwater ecosystems. Arch Environ Contam Toxicol 39:20–31
Magilligan FJ (1992) Sedimentology of a fine-grained aggrading floodplain. Geomorphology 4:393–408
Merritts D, Walter R, Rahnis M, Hartranft J, Cox S, Gellis A, Potter N, Hilgartner W, Langland M, Manion L, Lippincott C, Siddiqui S, Rehman Z, Scheid C, Kratz L, Shilling A, Jenschke M, Datin K, Cranmer E, Reed A, Matuszewski D, Voli M, Ohlson E, Neugebauer A, Ahamed A, Neal C, Winter A, Becker S (2011) Anthropocene streams and base-level controls from historic dams in the unglaciated mid-Atlantic region, USA. Phil Trans Royal Soc London A: Math, Phys Eng Sci 369:976–1009
Meybeck M, Lestel L, Bonté P, Moilleron R, Colin JL, Rousselot O, Hervé D, de Pontevès C, Grosbois C, Thévenot DR (2007) Historical perspective of heavy metals contamination (Cd, Cr, Cu, Hg, Pb, Zn) in the Seine River basin (France) following a DPSIR approach (1950–2005). Sci Total Environ 375:204–231
Meyers PA (1997) Organic geochemical proxies of paleoceanographic, paleolimnologic, and paleoclimatic processes. Org Geochem 27:213–250
Michelot J-L (1983) Evolution des paysages fluviaux de la vallée du Rhône dans le secteur du Péage-de-Roussillon. Géocarrefour 58:307–322
Mourier B, Desmet M, Van Metre PC, Perrodin Y, Roux C, Bedell JC, Lefèvre I, Babut M (2014) Historical records, sources, and spatial trends of PCBs along the Rhône River (France). Sci Total Environ 476:568–576
Ollivier P, Radakovitch O, Hamelin B (2011) Major and trace element partition and fluxes in the Rhône River. Chem Geol 285:15–31
Parrot E (2015) Analyse spatio-temporelle de la morphologie du chenal du Rhône du Léman à la Méditerranée. Dissertation. University of Lyon, France
Passmore DG, Macklin MG (1994) Provenance of fine-grained alluvium and late Holocene land-use change in the Tyne basin, northern England. Geomorphology 9:127–142
Petit F, Poinsart D, Bravard J-P (1996) Channel incision, gravel mining and bedload transport in the Rhône river upstream of Lyon, France (“canal de Miribel”). Catena 26:209–226
Pinglot JF, Pourchet M (1995) Radioactivity measurements applied to glaciers and lake sediments. Sci Total Environ 173–174:211–223
Poeppl RE, Keesstra SD, Hein T (2015) The geomorphic legacy of small dams—an Austrian study. Anthropocene 10:43–55
Provansal M, Dufour S, Sabatier F, Anthony E, Raccasi G, Robresco S (2014) The geomorphic evolution and sediment balance of the lower Rhône River (southern France) over the last 130years: hydropower dams versus other control factors. Geomorphology 219:27–41
Przedwojski B (1995) Bed topography and local scour in rivers with banks protected by groynes. J Hydraul Res 33:257–273
Räpple B (2018) Sedimentation patterns and riparian vegetation characteristics in novel ecosystems on the Rhône River, France: A comparative approach to identify drivers and evaluate ecological potentials. Dissertation. University of Lyon, France
Ritchie JC, McHenry JR (1990) Application of radioactive fallout cesium-137 for measuring soil erosion and sediment accumulation rates and patterns: a review. J Environ Qual 19:215–233
Savić R, Ondrašek G, Bezdan A, Letić L, Nikolić V (2013) Fluvial deposition in groyne fields of the middle course of the Danube River. Tehnički vjesnik 20:979–983
Steinmann P, Adatte T, Lambert P (2003) Recent changes in sedimentary organic matter from Lake Neuchâtel (Switzerland) as traced by Rock-Eval pyrolysis. In: Beres M, Scheidhauer M, Marillier F (eds) Lake systems from the ice age to industrial time. Birkhäuser Basel, Basel, pp 109–116
Surian N, Rinaldi M (2003) Morphological response to river engineering and management in alluvial channels in Italy. Geomorphology 50:307–326
Syvitski JPM, Kettner A (2011) Sediment flux and the Anthropocene. Phil Trans Royal Soc of London A: Math Phys Eng Sci 369:957–975
Tena A, Seignemartin G, Roux G, Winiarski T, Piégay H (2017) Factors controlling groyne-field sedimentation and contamination along the Rhône River. Presented at the 3rd International Conference on the Status and Future of the World’s large rivers, New Delhi, India
Tockner K, Uehlinger U, Robinson CT (2009) Rivers of Europe. Elsevier
Toonen WHJ, Winkels TG, Cohen KM, Prins MA, Middelkoop H (2015) Lower Rhine historical flood magnitudes of the last 450years reproduced from grain-size measurements of flood deposits using end member modelling. Catena 130:69–81
Vázquez-Tarrío D, Tal M, Camenen B, Piégay H (2019) Effects of continuous embankments and successive run-of-the-river dams on bedload transport capacities along the Rhône River, France. Sci Total Environ 658:1375–1389
Walter RC, Merritts DJ (2008) Natural streams and the legacy of water-powered mills. Sci 319:299–304
Weltje GJ, Prins MA (2003) Muddled or mixed? Inferring paleoclimate from size distributions of deep-sea clastics. Sediment Geol 162:39–62
Weltje GJ, Prins MA (2007) Genetically meaningful decomposition of grain-size distributions. Sediment Geol 202:409–424
Wen X, Du Q, Tang H (1998) Surface complexation model for the heavy metal adsorption on natural sediment. Environ Sci Technol 32:870–875
Zawiejska J, Wyżga B (2010) Twentieth-century channel change on the Dunajec River, southern Poland: patterns, causes and controls. Geomorphology 117:234–246
This study was conducted in the combined frameworks of (i) the Rhône Sediment Observatory (OSR), a multi-partner research program funded through the Plan Rhône by the European Regional Development Fund (ERDF), Agence de l'eau RMC, CNR, EDF, and three regional councils (Auvergne-Rhône-Alpes, PACA and Occitanie), and (ii) the EUR H2O’Lyon (ANR-17-EURE-0018) of Université de Lyon (UdL), within the program “Investissements d'Avenir” operated by the French National Research Agency (ANR).
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Responsible editor: Elena Romano
Electronic supplementary material
About this article
Cite this article
Vauclin, S., Mourier, B., Tena, A. et al. Effects of river infrastructures on the floodplain sedimentary environment in the Rhône River. J Soils Sediments 20, 2697–2708 (2020). https://doi.org/10.1007/s11368-019-02449-6
- Rhône River
- River infrastructures
- Sediment cores