The taxonomic structure but not the functioning of riparian herbaceous communities varies with hydrological conditions on a large, highly regulated river: Evidence from a 2-year replicated study
Funding information: INRAE; Agence de l'eau Rhône Méditerranée Corse; OHM Vallée du Rhône; Labex DRIIHM
Abstract
Riparian vegetation encompasses dynamic communities structured by strong environmental gradients. Due to anthropogenic changes, a greater uniformity of environmental conditions is observed along local gradients. To maintain diverse and functional communities in regulated rivers, there is an urgent need to finely characterize the factors controlling the spatial and temporal dynamics of riparian vegetation. In the summer of 2017 and 2019, we repeatedly sampled eight gravel bars along the regulated Rhône River. Using a taxonomic and functional approach, we assessed whether the diversity and composition of species and traits varied between the 2 years at the corridor and site scales. Taxonomic diversity decreased and composition changed between years, while functional diversity and composition did not. Specifically, we showed that variation in taxonomic diversity was largely limited to species sharing the same trait values, suggesting some functional resilience of riparian communities to environmental changes. In addition, we found that changes between the 2 years concerned primarily the communities established on low-lying bars made up of coarse-grained sediments, more frequently and intensely inundated. Beyond being the most responsive to environmental changes, these riparian communities were also the most diverse in species and trait values. However, by inducing a strong decrease in diversity, fine sedimentation seems to be a major threat to the functional and ecological integrity of gravel bars. Overall, these results highlight the conservation priority of low-lying gravel surfaces and the need to promote more morphologically complex riparian habitats by renaturalizing flow variability and sediment regimes.
1 INTRODUCTION
River ecosystems are intrinsically dynamic and exhibit strong environmental gradients that shape the structure of communities over space and time (Allan & Castillo, 2007). In free-flowing, that is, unregulated and unchannelled rivers, complex hydro-geomorphic processes promote water and sediment exchanges in several dimensions of the hydrosystem (Ward, 1989). These exchanges allow the creation and renewal of a mosaic of complex habitats (Junk et al., 1989) that benefit a large number of species (Dudgeon et al., 2006). This is particularly the case in riparian zones, that is, in the transitional area between terrestrial and aquatic habitats, where plant communities face wide variations in disturbance and stress intensities along the flood gradient (Naiman & Decamps, 1997). These rapid changes in environmental conditions induce a strong turnover in plant strategies, that is, from ruderal-anoxia-tolerant species at the bottom of riverbanks to competitive-desiccation-tolerant species at the top of riverbanks (Janssen, Piégay, & Evette, 2020; Kyle & Leishman, 2009; McCoy-Sulentic et al., 2017). River–floodplain connectivity, conceptualized within the flood pulse framework (Junk et al., 1989), is thus a key property of hydrosystems capable of ensuring the functionality and ecological integrity of riparian zones (Naiman & Decamps, 1997).
Around the world, river connectivity has been profoundly affected by human infrastructures such as dams and dikes (Grill et al., 2019; Nilsson et al., 2005). These developments have dramatically altered the flow regime, that is, by decreasing the intensity and magnitude of inundations (Poff et al., 2007), as well as the sediment regime, that is, by trapping sediment load and favouring fine overbank depositions (Wohl et al., 2015), of most rivers. These anthropogenic changes have in turn altered the structure and dynamic of riparian communities, by reducing native species diversity and increasing invasion by exotic species (Catford et al., 2011; Richardson et al., 2007). Specifically, modifications to the flow and sediment regimes due to regulation/channelization works have promoted greater uniformity of environmental conditions along local gradients (Poff et al., 2007) and have led to a simplification of riparian communities (Aguiar et al., 2018; Bejarano et al., 2018; Tonkin et al., 2018). This can compromise the ability of degraded rivers to support functional and diverse communities and, as a result, the wide range of ecosystem services that riparian vegetation delivers to societies (Riis et al., 2020).
Although many studies have documented the impacts of anthropogenic environmental changes on riparian vegetation, many have focused on woody species and forested habitats established on riverbanks (Aguiar et al., 2018; Bejarano et al., 2018; Janssen, Stella, et al., 2020; Johnson et al., 2012). In regulated and channelized rivers, knowledge of the factors that structure herbaceous communities at fine spatial and temporal scales is thus still scarce (Su et al., 2020), particularly for riparian habitats of high conservation value, such as gravel bars (Gilvear et al., 2008). Composed of accumulations of alluvial sediments, these dynamic and morphologically complex habitats are nevertheless a key element of the river landscape and shelter numerous threatened plant and animal species (Kalníková et al., 2018; Rottenborn et al., 2018; Zeng et al., 2015). However, due to anthropogenic changes in flow and sediment regimes, the area of these habitats has been reduced to the extent that unvegetated or sparsely vegetated bars with mobile sediments are classified as vulnerable in the European Red List of Habitats (Janssen et al., 2016). To preserve and restore these species-rich riparian habitats along regulated rivers, it is therefore urgent to better understand the factors that control the structure and dynamics of herbaceous plant communities on gravel bars over space, that is, by contrasting the effects of local and regional gradients at different scales, and time, that is, by resampling vegetation over several years.
In riparian ecology, several recent studies have mobilized a functional trait-based approach to understand the effects of hydraulic alterations on vegetation (Aguiar et al., 2018; Bejarano et al., 2018; Janssen, Stella, et al., 2020; Su et al., 2020; Tabacchi et al., 2019). Combined with a taxonomic approach, this allows us to focus both on conservation issues, by studying variations in species richness and composition, but also to overcome the limits of taxonomic studies, under the control of the local species pool, by studying the underlying ecological mechanisms captured by functional traits (Lavorel et al., 2008). Building on these conceptual foundations, we used a replicated sampling design spaced 2 years apart to study the structure and dynamic of plant communities on gravel bars, at the corridor and site scales, along the highly degraded Rhône River. As shifts in assemblage patterns are likely to result from a combination of different environmental factors related to the flow or sediment regimes (e.g., Gilvear et al., 2008; Škornik et al., 2017), we tested the relative influence of a set of topographic, pedological and hydrological variables on riparian plants, at the community scale. Topographic variables were used to assess the influence of the shape of the bar and thus better guide bank reprofiling operations. Pedological variables were used to assess the influence of fine sediment deposition and thus the need to develop specific management plans. Hydrological variables were used to assess the influence of temporal variations in flows and thus the potential for managing riparian vegetation through environmental flows. Based on this scheme, we addressed the following two questions: (a) In regulated and channelized rivers, how and to what extent do environmental conditions and taxonomic and functional diversity/composition of gravel bar plants vary over a 2-year period at the corridor and gravel bar scales? (b) At the community scale, what is the relative influence of topographic, pedological and hydrological variables in explaining variations in taxonomic and functional diversity/composition over time and space?
2 MATERIALS AND METHODS
2.1 Study area and experimental design
The study was carried out along the Rhône River (total length = 810 km, catchment area = 96,500 km2, mean annual discharge = 1700 m3/s) on a 250-km-long stretch of its middle reach (Figure 1). This area is characterized by a temperate climate with mean annual temperatures and precipitation of 13.6°C and 755 mm in its southern part and 11.6°C and 815 mm in its northern part. Two historical development phases have greatly altered the functioning of this river: (i) During the 19th century, a correction phase was carried out to facilitate navigation and (ii) in the second half of the 20th century, a derivation phase was carried out to produce hydropower and facilitate irrigation and navigation (for details, see Bravard & Gaydou, 2015). These two development phases induced significant impacts in the bypassed reaches (i.e., part of the historical Rhône River channel), including a channel dewatering due to flow by-passing, reduction in the frequency and magnitude of the peak flows (Vázquez-Tarrío et al., 2019), and channel stabilization due to engineering infrastructures. These reaches also underwent a bedload supply interruption due to sediment trapping and extractions at tributary confluences inducing bed armouring within the main channel. Thus, with the exception of the Miribel bypassed reach, which is located in the northernmost part of the study area, present-day bedload transport is limited to fine-grained sediments (Vázquez-Tarrío et al., 2019).
During the summer of 2017, we sampled eight gravel bars (GB1-8) belonging to seven bypassed reaches, along a north–south section of about 250 km (Figure 1). The same eight gravel bars were resampled, at the same location during the summer of 2019. The studied gravel bars are fixed in the bypassed reaches, that is, coarse sediments are no longer remobilized during flood events and are subject to fine-grained overbank sedimentation and colonization by trees. These large fixed bars correspond to a part of the river bottom before derivation and represent the only terrestrial habitats available for species within the Rhône riverbed. They are colonized by different plant communities, dominated by hygrophilic perennial (e.g., Carex acutiformis, Lysimachia vulgaris and Phalaris arundinacea) or annual species (e.g., Persicaria lapathifolia, Persicaria mitis and Rorippa sylvestris) near the water line and xero-nitrophilous species (e.g., Chenopodium album, Bromus sterilis and Elymus pungens) at higher elevations. Many exotic species (e.g., Ambrosia artemisiifolia, Aster × salignus and Cyperus eragrostis) have also taken advantage of these surfaces, forming locally dense patches.
To prevent encroachment of gravel bars by woody species and the associated channel narrowing, river managers periodically rejuvenate the vegetation from the waterline to the edge of the alluvial forest, resulting in a blockage of plant ecological successions. Specifically, in the study area, the four northern gravel bars (i.e., GB1, GB2, GB3 and GB4) are managed by vegetation clearing using a tractor-drawn grinder, while the four southern gravel bars (i.e., GB5, GB6, GB7 and GB8) are managed by ploughing using a bulldozer. These vegetation maintenance works were performed in the fall, annually for brush clearing and every 2 years for ploughing (in 2016 and 2018 for the southern banks studied).
2.2 Vegetation data
On each gravel bar, the vegetation was surveyed along three transects positioned perpendicular to the river in areas with clear elevation gradients relative to the water surface. The gradients started at the water line and ended before the alluvial floodplain. Along each transect, four 5 × 5 m quadrats (25 m2) were positioned in homogenous areas while avoiding the aquatic parts of the river margin (i.e., 12 quadrats per gravel bar). Within each quadrat (n2017 = 96, n2019 = 96), all vascular plants were surveyed following the Braun–Blanquet abundance/dominance methodology (Braun-Blanquet, 1932). Because maintenance works rejuvenated the bars every year or two, tree and shrub layers were absent, and vegetation surveys were conducted by considering the herbaceous layer only. To account for the time lag in vegetation development between high and low level quadrat along the topographic gradient, two complete surveys were conducted per year (end of June and end of July). Then, to characterize the plant community at the scale of each quadrat, the two surveys were combined by listing all species encountered at least once during the two inventories and associating them with the highest Braun–Blanquet abundance/dominance coefficient noted. Finally, to limit observer biases, all surveys were performed by the first author (PJ).
2.3 Functional trait data
Three morphological traits—specific leaf area (SLA; leaf area per dry mass), plant height at maturity and seed mass—were extracted from the TRY database (Kattge et al., 2011). Within Westoby's plant ecology strategy scheme (Westoby, 1998), these three easily measurable traits represent three axes of functional trade-offs to explain plant distribution. SLA is related to resource acquisition and conservation, contrasting ‘conservative’ species with long-lived leaves to ‘acquisitive’ species with rapid turnover of plant leaves. Plant height at maturity is related to competitive ability, contrasting taller dominant-species with a high ability to pre-empt resources but with high construction and maintenance costs to smaller dominated-species but with low construction and maintenance costs. Seed mass is related to dispersal ability, contrasting light-seeded species but with high seedling mortality to large-seeded species but with high long-term seedling survival.
For each trait, we calculated a mean value per species after removing all trait values with an error risk > 3, that is, having a maximum distance of three standard deviations from at least one of the mean trait values, and applied a log10-transformation to comply with normality and to limit extreme values. Then, within each quadrat, we computed (i) the community-weighted means (CWMs) and (ii) the functional dispersion (FDis) for each of the three traits (FD package, Laliberté et al., 2014). CWM is defined as the mean of the trait values weighted by the relative cover of each species bearing each value (Lavorel et al., 2008) and was used as a measure of functional composition. FDis is defined as the mean distance of individual species to the weighted centroid of all species in the assemblage (Laliberté & Legendre, 2010) and was used as a measure of functional diversity. Since the trees/shrubs were present only at the seedling stage and the height values of these species strongly influenced the calculation of CWM for this trait, we arbitrarily set a height value of 0.10 m for all phanerophyte species (n = 18, see Table S1 for details).
In addition, we studied the response of annual species, that is, therophytes based on plant life-forms (Raunkiaer, 1934) extracted from the Baseflor database (Julve, 1998), which constitute communities often with high conservation stakes on gravel bars (e.g., Rottenborn et al., 2018), and exotic species, based on a regional hierarchical list of invasive species (Debay et al., 2020), which represent potential threats to native species (Richardson et al., 2007) (see detailed information for each species in Table S1).
2.4 Environmental variables
At the centre of each quadrat, the elevation difference and distance to the water line were characterized using topographical surveys. Topographic measurements were conducted using a laser rangefinder (TruPulse 200X) positioned on a tripod at a constant height of 1.2 m and a target at a height of 2.0 m. The horizontal and vertical distances between the tripod and the target were measured between the water line at the beginning of each transect and the centre of the four successive quadrats. For soil properties, the proportion of subsoil fine sediment was derived from two soil core samples (30-cm thickness) collected in representative areas within each quadrat and pooled together. Each composite soil sample was then dried, weighed and sieved in the laboratory to estimate the proportion of fine sediments (i.e., <250-μm mesh). Moreover, fine sediment thickness was estimated at the centre of each quadrat using a soil corer inserted into the sediment up to the interface with the gravel and measuring the distance from the surface. All of these measurements were made on each gravel bar in 2017 and 2019 during the survey carried out in June (for illustration of the variations in topographic and pedological variables, within and between gravel bars, see Figure S1).
For hydrological variables, the water elevation (m) of the Rhône was calculated for each transect using a 1D hydrodynamic model based on the 1D shallow water equations (for details, see Dugué et al., 2015; Launay et al., 2019). This numerical model covers the entire length of the French Rhône, including its main tributaries and operation rules of hydropower structures. It has been calibrated and validated on each reach using longitudinal water profiles, provided by the Compagnie Nationale du Rhône, over a wide range of discharges for frequent hydrological events. The model correctly reproduces discharge time-series, especially during floods. For low flow periods, results are not as accurate due to hydropower scheme subdaily operations that are not included in the model. The output of the model provides a detailed elevation of the water, due to the real geometry of the river bed as a series of cross-sections every 200–500 m and the high temporal resolution of the hourly results at any point in the model. Combined with centimetric measures of the quadrat elevation, we were able to calculate two hydrological variables: the inundation duration, that is, the sum of hourly measurements below the water level expressed in days, and the inundation intensity, that is, the highest water level expressed in meters. For each quadrat, these variables were calculated considering a time extension from 1 September to 31 August for the 2016/2017 and 2018/2019 periods (Figure S2).
2.5 Statistical analysis
The continuous independent variables were the elevation difference (denoted ‘Elevation’ in tables and figures) and distance to water line (‘Distance’), the proportion (‘Prop_Sedim’) and thickness (‘Thick_Sedim’) of fine sediments, the duration (‘Dura_Inund’) and intensity (‘Inte_Inund’) of inundation (Figure S3). The independent factors were the year of sampling (‘Year, two-level factor), the gravel bar (‘GB’, eight-level factor) and maintenance measures (‘Manag’, two-level factor). The skewed variables ‘Thick_Sedim’ and ‘Dura_Inund’ were log10-transformed while the ‘Prop_Sedim’ (i.e., proportional data) was logit transformed. Finally, all continuous variables were standardized before performing the analyses.
To determine whether environmental conditions, taxonomic diversity, that is, the richness of total, annual and exotic species, and functional diversity, that is, the functional dispersion (i.e., FDis) of SLA, height and seed trait values, varied between 2017 and 2019, between gravel bars managed by brush clearing or ploughing, as well as with the interaction between both factors, we used two-way analysis of variance (ANOVAs) (stats package). To determine whether environmental conditions, taxonomic diversity and functional diversity varied over time within each bar we used paired t test (stats package). We also used paired t test to determine how inundation duration and intensity varied along the elevation gradient, that is, in each quadrat position, between years. As multiple comparison tests were performed, we adjusted the p values with the Benjamini and Hochberg correction.
To determine how environmental conditions explained differences in taxonomic and functional diversity between years, at the community scale, we used a modelling approach. We considered a set of 12 a priori models, testing additive and interactive effects between the six environmental variables and year, plus a null model (Table S2). Since observations within each bar were not independent, we used mixed models with ‘GB’ as a random effect (lme4 package, Bates et al., 2015). Also, because maintenance measures influenced significantly the taxonomic diversity, we included ‘Manag’ as a covariate in all a priori models. We then fitted mixed models with a Poisson distribution to study variations in species richness and with a normal distribution to study variations in trait dispersion. To identify the most parsimonious model, we used Akaike's information criterion corrected for small sample sizes and performed model averaging, based on all a priori models, to estimate the parameters and associated unconditional standard errors (MuMIn package, Barton, 2015).
To determine whether the taxonomic composition and functional composition varied between 2017 and 2019, between gravel bars managed by brush clearing or ploughing, as well as with the interaction between both factors, we used two-way permutational multivariate ANOVAs (PERMANOVAs) (Anderson & Walsh, 2013) with 999 permutations (vegan package, Oksanen et al., 2013). For the taxonomic composition, we used a Hellinger distance matrix based on the cover of each species in each of the quadrats, while for the functional composition, we used a Euclidean distance matrix based on the mean values of each trait (i.e., CWM) in each of the quadrats. As multiple comparison tests were performed, we adjusted the p values with the Benjamini and Hochberg correction. To determine whether the taxonomic composition and functional composition varied over time within each bar, we used one-way PERMANOVAs and adjusted the p values with the Benjamini and Hochberg correction.
To determine how variations in environmental conditions explained differences in taxonomic and functional composition, at the community scale, we used canonical analysis of principal coordinates (CAP, Anderson & Willis, 2003), with 999 permutations (vegan package, Oksanen et al., 2013). We then calculated the marginal contribution of all independent variables to total constrained inertia and tested for their individual significance (after all other variables were partialled out).
Analyses were performed with R version 3.5.1 (R Core Team, 2020) (see a summary of the statistical analyses performed in Figure S4).
3 RESULTS
3.1 Changes in environmental conditions between years
Two-way ANOVAs showed that, on average, inundation intensity within quadrats was significantly lower in 2019 than in 2017, and that this effect was more pronounced on gravel bars managed by ploughing, that is, a significant interaction effect (Table 1). In addition, compared to quadrats located on gravel bars managed by brush clearing, quadrats on bars managed by ploughing were at a shorter average distance and higher average elevation from the water line. In the same quadrats, that is, managed by ploughing, the average thickness of fine sediments, as well as the average duration of inundation, were significantly lower.
| Variables | 2017: mean (±SD) | 2019: mean (±SD) | Year | Management | Interaction | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| Clearing | Ploughing | Clearing | Ploughing | F | p value | F | p value | F | p value | |
| Environmental conditions | ||||||||||
| Elevation above water line (m) | 1.14 (±0.78) | 1.00 (±0.72) | 1.20 (±0.87) | 0.88 (±0.74) | 0.086 | 0.770 | 4.249 | 0.041 | 0.667 | 0.415 |
| Distance to water line (m) | 28.43 (±21.06) | 34.92 (±25.84) | 28.56 (±20.60) | 35.54 (±25.98) | 0.012 | 0.912 | 3.941 | 0.049 | 0.005 | 0.942 |
| Proportion of fine sediments (%) | 43.20 (±34.73) | 37.16 (±34.79) | 39.06 (±37.61) | 32.95 (±31.24) | 0.014 | 0.906 | 0.607 | 0.437 | 0.185 | 0.667 |
| Thickness of fine sediments (cm) | 22.81 (±27.11) | 18.14 (±27.35) | 28.35 (±32.56) | 18.36 (±26.86) | 0.005 | 0.946 | 7.404 | 0.007 | 0.254 | 0.615 |
| Inundation duration (days) | 15.20 (±12.69) | 11.48 (±9.30) | 19.26 (±22.60) | 10.20 (±11.81) | 1.400 | 0.238 | 9.426 | 0.002 | 1.038 | 0.310 |
| Inundation intensity (m) | 2.11 (±0.89) | 2.97 (±1.81) | 1.93 (±0.94) | 1.74 (±0.92) | 16.546 | 0.000 | 3.664 | 0.057 | 9.106 | 0.003 |
| Taxonomic diversity | ||||||||||
| Total species richness | 18.35 (±9.12) | 31.29 (±11.61) | 11.96 (±4.92) | 25.00 (±10.95) | 21.336 | 0.000 | 89.456 | 0.000 | 0.001 | 0.970 |
| Richness of annual species | 6.19 (±5.83) | 14.92 (±6.09) | 3.69 (±2.49) | 11.71 (±6.04) | 13.758 | 0.000 | 118.459 | 0.000 | 0.212 | 0.646 |
| Richness of exotic species | 4.21 (±3.38) | 8.94 (±3.45) | 2.67 (±1.74) | 7.54 (±3.91) | 9.956 | 0.002 | 106.425 | 0.000 | 0.025 | 0.876 |
| Functional diversity | ||||||||||
| FDis of SLA trait values | 0.64 (±0.32) | 0.61 (±0.25) | 0.59 (±0.29) | 0.59 (±0.21) | 0.925 | 0.337 | 0.242 | 0.623 | 0.110 | 0.741 |
| FDis of plant height trait values | 0.70 (±0.26) | 0.62 (±0.18) | 0.64 (±0.35) | 0.64 (±0.24) | 0.125 | 0.725 | 1.048 | 0.307 | 1.022 | 0.313 |
| FDis of seed mass trait values | 0.56 (±0.23) | 0.70 (±0.29) | 0.50 (±0.25) | 0.72 (±0.22) | 0.355 | 0.552 | 26.414 | 0.000 | 1.371 | 0.243 |
- Abbreviations: FDis, functional dispersion; SD, standard deviation; SLA, specific leaf area.
Paired t test showed that within gravel bars, the elevation increased significantly between 2017 and 2019 on GB3 but decreased on GB7 and GB8, while the distance to water increased on GB5 but decreased on GB8. The proportion of fine sediments increased significantly between years on GB2, GB4 and GB8 but decreased on GB3, while the thickness of fine sediments decreased only on GB7. Finally, inundation intensity decreased significantly between years on all bars, except on GB1 where it increased, while inundation duration increased on GB1 and GB8 but decreased on GB3, GB5, GB6 and GB7 (Figure 2 & detailed results in Table S3). However, while inundations were more intense in 2017 regardless of quadrat position along the elevation gradient, they were longer in 2019 on low-level quadrats composed of coarse-grained sediments (Figure S5).
3.2 Changes in taxonomic and functional diversity between years
A total of 205 species (mean ± SD = 21 ± 12) were recorded in the 192 quadrats along the Rhône River. Among these, 38 were exotic and 87 were annual.
Two-way ANOVAs showed that, on average, the richness of total, annual and exotic species within the quadrats decreased significantly between 2017 and 2019, while for functional diversity, trait dispersion values (i.e., FDis) remained stable (Table 1). In addition, the average richness of total, annual and exotic species, as well as the functional dispersion values of seed mass within the quadrats decreased significantly on gravel bars managed by brush clearing as compared to gravel bars managed by ploughing. However, none of the measures of taxonomic and functional diversity was significantly influenced by the interaction between the years and management factors.
Paired t test showed that within gravel bars, species richness decreased significantly between 2017 and 2019 on GB1, GB2, GB6 and GB8; functional dispersion values for plant height, seed mass and SLA decreased significantly between years on GB2; functional dispersion value for plant height increased significantly between years on GB8 (Figure 3 & detailed results in Table S4).
3.3 Changes in taxonomic and functional diversity with environmental conditions
Mixed models showed that models including pedological variables ranked first for the richness of annual and exotic species and the functional dispersion values (i.e., FDis) of height and seed mass values, while models including hydrological variables ranked first for total richness and the functional dispersion value of SLA (Table 2 & detailed results in Table S5). Model averaging revealed that total, annual and exotic species richness as well as the functional dispersion value of seed mass were higher on bars maintained by ploughing than on bars maintained by brush clearing (Table 3). Also, the proportion and thickness of fine sediments had a significant negative effect on species richness and the functional dispersion values of SLA and height. However, for species richness, the interaction between fine sediments proportion and years was significant, revealing a large variation in richness over time on the surfaces made up of coarse-grained sediments (see interaction plots in Figure S6). Total and exotic species richness, as well as the functional dispersion value of SLA, also decreased with increasing elevation, while all richness measures varied more between years on the surfaces closest to water. Conversely, total species richness and the functional dispersion value of SLA increased on more intensely inundated surfaces, while all richness measures as well as the functional dispersion value of SLA increased with inundation duration in 2017 but decreased in 2019.
| Dependent variable | Model (fixed-effects) | k | AICc | ΔAICc | W | R2marginal |
|---|---|---|---|---|---|---|
| Total richness | Manag + Year * Dura_Inund | 6 | 1601.5 | 0.000 | 0.390 | 0.651 |
| Total richness | Manag + Year + Elevation | 5 | 1602.1 | 0.627 | 0.285 | 0.658 |
| Total richness | Manag + Year * Thick_Sedim | 6 | 1603.9 | 2.435 | 0.115 | 0.648 |
| Annual richness | Manag + Year * Thick_Sedim | 6 | 1262.3 | 0.000 | 0.887 | 0.710 |
| Annual richness | Manag + Year + Thick_Sedim | 5 | 1266.4 | 4.148 | 0.112 | 0.716 |
| Annual richness | Manag + Year * Prop_Sedim | 6 | 1275.4 | 13.160 | 0.001 | 0.691 |
| Exotic richness | Manag + Year + Thick_Sedim | 5 | 983.1 | 0.000 | 0.477 | 0.555 |
| Exotic richness | Manag + Year * Thick_Sedim | 6 | 983.7 | 0.621 | 0.350 | 0.551 |
| Exotic richness | Manag + Year * Prop_Sedim | 6 | 986.6 | 3.511 | 0.082 | 0.539 |
| FDis SLA | Manag + Year * Dura_Inund | 7 | 16.2 | 0.000 | 0.368 | 0.130 |
| FDis SLA | Manag + Year + Thick_Sedim | 6 | 17.8 | 1.626 | 0.163 | 0.144 |
| FDis SLA | Manag + Year + Prop_Sedim | 6 | 18.2 | 2.009 | 0.135 | 0.142 |
| FDis height | Manag + Year + Prop_Sedim | 6 | 30.3 | 0.000 | 0.571 | 0.061 |
| FDis height | Manag + Year * Prop_Sedim | 7 | 32.2 | 1.931 | 0.217 | 0.062 |
| FDis height | Manag + Year + Thick_Sedim | 6 | 33.1 | 2.871 | 0.136 | 0.047 |
| FDis seed | Manag + Year * Thick_Sedim | 7 | −8.9 | 0.000 | 0.415 | 0.184 |
| FDis seed | Manag + Year + Thick_Sedim | 6 | −8.8 | 0.067 | 0.401 | 0.177 |
| FDis seed | Manag + Year + Prop_Sedim | 6 | −4.8 | 4.027 | 0.055 | 0.156 |
- Note: Number of estimated parameters including the intercept and random effect (k), AICc value, difference in AICc (ΔAICc), AICc weight (W) and marginal coefficient of determination for fixed effect (R2marginal) are provided (for the description of model parameters, please refer to Table 1).
| Parameters | Total richness | Annual richness | Exotic richness | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Imp. | Estimate (±SE) | (95% CI) | Imp. | Estimate (±SE) | (95% CI) | Imp. | Estimate (±SE) | (95% CI) | |
| Year | 1.00 | −0.293 (±0.033) | (−0.357, −0.229) | 1.00 | −0.299 (±0.051) | (−0.400, −0.199) | 1.00 | −0.253 (±0.062) | (−0.373, −0.132) |
| Management | 1.00 | 0.599 (±0.105) | (0.393, 0.804) | 1.00 | 0.944 (±0.149) | (0.652, 1.237) | 1.00 | 0.828 (±0.111) | (0.610, 1.046) |
| Elevation | 0.38 | −0.078 (±0.019) | (−0.115, −0.040) | 0.00 | −0.007 (±0.030) | (−0.065, 0.052) | 0.03 | −0.092 (±0.038) | (−0.166, −0.018) |
| Distance | 0.00 | −0.070 (±0.022) | (−0.112, −0.027) | 0.00 | −0.033 (±0.034) | (−0.100, 0.033) | 0.00 | −0.062 (±0.044) | (−0.148, 0.024) |
| Prop_Sedim | 0.03 | −0.106 (±0.028) | (−0.161, −0.050) | 0.00 | −0.280 (±0.045) | (−0.368, −0.193) | 0.09 | −0.163 (±0.055) | (−0.271, −0.055) |
| Thick_Sedim | 0.20 | −0.111 (±0.029) | (−0.168, −0.054) | 1.00 | −0.307 (±0.047) | (−0.400, −0.215) | 0.83 | −0.162 (±0.049) | (−0.259, −0.066) |
| Dura_Inund | 0.39 | 0.132 (±0.029) | (0.076, 0.189) | 0.00 | 0.032 (±0.045) | (−0.056, 0.121) | 0.05 | 0.156 (±0.057) | (0.045, 0.267) |
| Inte_Inund | 0.00 | 0.061 (±0.025) | (0.012, 0.111) | 0.00 | 0.094 (±0.037) | (0.022, 0.166) | 0.00 | 0.056 (±0.044) | (−0.030, 0.142) |
| Year * Elevation | 0.10 | 0.003 (±0.033) | (−0.061, 0.068) | 0.00 | −0.023 (±0.051) | (−0.122, 0.076) | 0.01 | 0.026 (±0.065) | (−0.101, 0.154) |
| Year * Distance | 0.00 | 0.085 (±0.030) | (0.026, 0.144) | 0.00 | 0.111 (±0.045) | (0.022, 0.199) | 0.00 | 0.126 (±0.057) | (0.014, 0.238) |
| Year * Prop_Sedim | 0.03 | 0.121 (±0.034) | (0.053, 0.188) | 0.00 | 0.206 (±0.055) | (0.098, 0.315) | 0.08 | 0.176 (±0.067) | (0.045, 0.307) |
| Year * Thick_Sedim | 0.12 | 0.054 (±0.032) | (−0.009, 0.116) | 0.89 | 0.125 (±0.050) | (0.027, 0.224) | 0.35 | 0.076 (±0.062) | (−0.046, 0.197) |
| Year * Dura_Inund | 0.39 | −0.131 (±0.034) | (−0.198, −0.065) | 0.00 | −0.175 (±0.051) | (−0.276, −0.074) | 0.05 | −0.200 (±0.065) | (−0.328, −0.073) |
| Year * Inte_Inund | 0.00 | −0.070 (±0.040) | (−0.148, 0.008) | 0.00 | −0.236 (±0.060) | (−0.354, −0.118) | 0.00 | −0.132 (±0.075) | (−0.279, 0.014) |
| Parameters | FDis SLA | FDis height | FDis seed | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Imp. | Estimate (±SE) | (95% CI) | Imp. | Estimate (±SE) | (95% CI) | Imp. | Estimate (±SE) | (95% CI) | |
| Year | 1.00 | −0.030 (±0.036) | (−0.100, 0.040) | 1.00 | −0.014 (±0.037) | (−0.086, 0.058) | 1.00 | −0.021 (±0.032) | (−0.084, 0.042) |
| Management | 1.00 | −0.013 (±0.087) | (−0.184, 0.158) | 1.00 | −0.048 (±0.038) | (−0.122, 0.026) | 1.00 | 0.164 (±0.076) | (0.014, 0.314) |
| Elevation | 0.09 | −0.090 (±0.026) | (−0.140, −0.039) | 0.00 | −0.003 (±0.023) | (−0.048, 0.042) | 0.03 | 0.014 (±0.028) | (−0.040, 0.068) |
| Distance | 0.00 | −0.042 (±0.025) | (−0.091, 0.007) | 0.00 | −0.004 (±0.026) | (−0.055, 0.047) | 0.01 | 0.012 (±0.019) | (−0.025, 0.050) |
| Prop_Sedim | 0.20 | −0.104 (±0.025) | (−0.152, −0.056) | 0.79 | −0.064 (±0.021) | (−0.105, −0.022) | 0.08 | −0.039 (±0.025) | (−0.087, 0.009) |
| Thick_Sedim | 0.22 | −0.106 (±0.025) | (−0.155, −0.056) | 0.19 | −0.056 (±0.022) | (−0.099, −0.013) | 0.82 | −0.049 (±0.028) | (−0.105, 0.006) |
| Dura_Inund | 0.49 | 0.128 (±0.035) | (0.059, 0.197) | 0.01 | 0.024 (±0.024) | (−0.023, 0.071) | 0.01 | 0.000 (±0.026) | (−0.051, 0.052) |
| Inte_Inund | 0.00 | 0.069 (±0.025) | (0.020, 0.118) | 0.00 | 0.001 (±0.021) | (−0.040, 0.042) | 0.03 | −0.030 (±0.026) | (−0.080, 0.021) |
| Year * Elevation | 0.04 | 0.046 (±0.035) | (−0.021, 0.114) | 0.00 | 0.024 (±0.038) | (−0.051, 0.098) | 0.02 | −0.064 (±0.033) | (−0.128, 0.000) |
| Year * Distance | 0.00 | 0.046 (±0.036) | (−0.024, 0.117) | 0.00 | 0.049 (±0.038) | (−0.025, 0.123) | 0.00 | 0.002 (±0.033) | (−0.062, 0.066) |
| Year * Prop_Sedim | 0.06 | 0.026 (±0.035) | (−0.043, 0.096) | 0.22 | 0.017 (±0.037) | (−0.055, 0.090) | 0.03 | −0.031 (±0.033) | (−0.096, 0.034) |
| Year * Thick_Sedim | 0.06 | 0.013 (±0.035) | (−0.055, 0.082) | 0.05 | 0.016 (±0.037) | (−0.057, 0.089) | 0.41 | −0.048 (±0.032) | (−0.111, 0.015) |
| Year * Dura_Inund | 0.37 | −0.075 (±0.036) | (−0.145, −0.005) | 0.00 | −0.021 (±0.039) | (−0.098, 0.055) | 0.00 | 0.037 (±0.034) | (−0.030, 0.105) |
| Year * Inte_Inund | 0.00 | 0.027 (±0.043) | (−0.057, 0.111) | 0.00 | 0.012 (±0.045) | (−0.076, 0.099) | 0.02 | 0.071 (±0.039) | (−0.007, 0.148) |
- Note: The 95% confidence interval of coefficients in bold excluded 0. For the maintenance measures (Manag) and year of sampling (Year) factors, positive estimates represent ploughing or 2019, while negative estimates represent brush clearing or 2018 (refer to Table 1 for the description of the other variables).
3.4 Changes in taxonomic and functional composition between years
Two-way PERMANOVAs showed that the taxonomic composition differed significantly between 2017 and 2019 (F1,188 = 2.880, p = 0.002), while the functional composition (i.e., CWM) did not (F1,188 = 1.201, p = 0.297) (Figure 4). In addition, the taxonomic composition differed significantly between gravel bars managed by brush clearing and gravel bars managed by ploughing (F1,188 = 15.395, p = 0.001), while the functional composition did not (F1,188 = 2.573, p = 0.106). However, both the taxonomic (F1,188 = 1.348, p = 0.130) and the functional (F1,188 = 1.016, p = 0.343) composition were not significantly influenced by the interaction between the years and management factors.
Within gravel bars, one-way PERMANOVAS showed that the taxonomic composition changed between 2017 and 2019 on GB7 (F1,22 = 5.032, p adjust = 0.004) and GB8 (F1,22 = 5.027, p adjust = 0.004), while the functional composition changed only on GB7 (F1,22 = 17.789, p adjust = 0.016).
3.5 Changes in taxonomic and functional composition with environmental conditions
CAP showed that 16.8% (F6,185 = 7.415, p = 0.001) of the variation in taxonomic composition and 22.9% (F6,185 = 10.475, p = 0.001) of the variation in functional composition were explained by environmental variables (Figure 5). For the taxonomic composition, the first CAP axis was positively related to elevation level and distance to water, which respectively explained 33.6% (p = 0.001) and 10.1% (p = 0.001) of the constrained inertia, and negatively related to the duration (8.4%, p = 0.001) and intensity (12.2%, p = 0.001) of inundations. The second CAP axis was negatively related to the proportion (20.6%, p = 0.001) and thickness (14.9%, p = 0.001) of fine sediments. For the functional composition, the first CAP axis was positively related to elevation level (26.8%, p = 0.001) and distance to water (2.2%, p = 0.001), but negatively to the duration (6.0%, p = 0.001) and intensity (1.7%, p = 0.008) of inundations. The second CAP axis was positively related to the proportion (22.6%, p = 0.001) and thickness (40.7%, p = 0.001) of fine sediments.
4 DISCUSSION
Understanding the interactions between fluvial landforms and vegetation dynamics is essential for success in restoring biodiversity and functionality to riparian communities on corridors heavily modified by human activities (Palmer & Ruhi, 2019). By studying variations of abiotic and biotic conditions at the corridor and gravel bar scales along the highly modified Rhône River between two seasons in 2017 and 2019, we showed that (i) environmental conditions varied a little from year to year, (ii) taxonomic measures clearly tracked environmental changes and (iii) functional measures were stable over time. Specifically, we showed that variations in richness and composition of herbaceous plants between the 2 years and sites were largely limited to species sharing the same trait values, underscoring the functional resilience of riparian communities to environmental changes. Also, we highlighted that the strongest variations in taxonomic diversity between the years 2017 and 2019 were spatially restricted to low elevation surfaces, made up of coarse-grained sediments and more frequently flooded. Beyond being the most vulnerable, riparian communities established under these conditions were also the most taxonomically and functionally diverse. To improve both the biodiversity and functionality of riparian herbaceous communities along the Rhône River, the conservation and restorations of these gravel bar habitats should be a priority.
4.1 River regulation and bank stabilization lead to relatively stable riparian habitats
At the corridor scale, there was no variation in topographic and pedological conditions between the 2 years. Within the gravel bars, changes of low magnitudes were found between 2017 and 2019, mostly limited to variations in the proportion of fine sediments. Although it cannot be ruled out that these small variations are the results of methodological biases in the field measurements, this tends to confirm that hydro-geomorphic processes along the Rhône River are limited to fine-grained sediments (Tena et al., 2020; Vauclin et al., 2020). Conversely, even if 2017 and 2019 were ‘calm’ hydrological years, that is, none of the peak flows recorded exceeded the 2-year return flood of the Rhône, at the corridor scale, a significant decrease in the intensity of inundation could still be measured between the 2 years. Within the gravel bars and with the exception of GB1, our results confirmed that inundations were more intense in 2017 but that their duration was either longer or shorter depending on the site. However, additional analyses of floods duration showed heterogeneous variations along the elevation gradient and with the extent of fine sediment deposits (Figure S5). Thus, while mid- and high-level quadrats were inundated on average 1 day longer in 2017, low-level quadrats, especially if they were composed of coarse-grained sediments, were inundated up to 20 days longer in 2019. This shows that in this regulated and channelized river system, where the bypassed reaches receive only a minimum baseflow (i.e., the residual flow is 10%–30% of the mean annual flow arriving at the upstream diversion dams) and are subject to a sharp reduction in the frequency and magnitude of peak flows (Vázquez-Tarrío et al., 2019), environmental factors indirectly related to the characteristics of the gravel bars are the main sources of variation that can shape riparian communities. Moreover, this tends to confirm that water and sediment exchanges, as well as scouring, in the absence of important floods, as was the case during this short study period, are very limited in the bypassed reaches (Vázquez-Tarrío et al., 2019) and do not allow, under current hydrological conditions, the creation or renewal of riparian habitats in the Rhône riverbed.
4.2 Taxonomic diversity and composition respond to environmental changes in space and time
Species richness decreased between 2017 and 2019 at the corridor scale and followed a similar trend within each of the bar. Specifically, we showed that maintenance measures, which are confused with the longitudinal and climatic gradient, strongly influenced richness, that is, northern bars maintained by brush clearing were species poor than southern bars maintained by ploughing. By disturbing the soil surface, ploughing artificially increases microtopographic variations on bars (sensu, Pollock et al., 1998), allowing greater local co-occurrence of species with contrasting habitat requirements (Janssen et al., 2019). After controlling for this confounding spatial effect, which was constant between the 2 years studied, we demonstrated that temporal changes in richness were independent of management practices and involved communities established under specific environmental conditions. Indeed, plant richness varied more between years at low elevations, on more frequently flooded surfaces consisting of coarse-grained sediments. Since flooding on the Rhone was shown to be more intense in 2017 regardless of quadrat position along the elevation gradient, but longer in 2019 at the low-level quadrats composed of coarse-grained sediments, we infer that these variations in hydrological conditions were the primary cause of species loss. Indeed, it was shown that flooded areas received a more diverse pool of colonizing species through hydrochory (Gurnell et al., 2008; Jansson et al., 2005) but also that intermediate-intensity floods supported high species diversity (Pollock et al., 1998; Renofalt et al., 2005). However, over the 2 years studied, the magnitude of flooding varied within a narrow range, and the peak flows recorded cannot be characterized as intermediate disturbances (Figure S2). At the same time, it has been shown that frequently flooded sites with low spatial variation in flood frequencies are species-poor (Pollock et al., 1998) but also that in dam-regulated rivers, long unnatural inundations have a negatively effect on species richness (e.g., Su et al., 2020; Zheng et al., 2021). On the Rhone, the spatial structuring of inundations, longer in 2019 at low levels on pebble-dominated bars, and of species richness, lower in 2019 in low-lying quadrats composed of coarse-grained sediments, supports these previous finding. Specifically, our results suggest that a prolonged submergence of about 20 days can induce a strong decrease in the taxonomic diversity of riparian herbaceous communities and highlight, in this highly regulated river system, the greater vulnerability of plant species established on bar surfaces made up of coarse-grained sediments to variations in hydrological conditions. Since along the gravel bars of the Rhone, as in other rivers (Kalníková et al., 2018; Škornik et al., 2017), species richness is highest in these riparian habitats, we believe that their conservation should be a priority (Gilvear et al., 2008; Rottenborn et al., 2018; Zeng et al., 2015).
Species composition also varied between the 2 years studied at the corridor scale. Within bars, changes in composition between years were however restricted to only two sites, which may be due to the small temporal extent of this study but also to management practices that periodically reset succession. Furthermore, community-wide analyses revealed that compositional changes were primarily explained by bar characteristics. Indeed, elevation difference explained most of the variation (Gilvear et al., 2008; Kalníková et al., 2018; Škornik et al., 2017), with a shift from flood-tolerant species at low elevation (e.g., P. arundinacea and Phragmites australis) to drought-tolerant species at high elevation (e.g., Lolium perenne and Elymus repens). Also, the proportion of fine sediments on the bars profoundly influenced species composition (Gilvear et al., 2008; Kalníková et al., 2018). Thus, at low elevation, P. arundinacea cover increased when there was little fine sediment while P. australis cover increased when there was a lot of fine sediment, and, at high elevation, L. perenne cover increased when there was little fine sediment while E. repens cover increased when there was a lot of fine sediment. Beyond bar characteristics, hydrological conditions during the year preceding observations were of little inference value to explain changes in herbaceous plant composition in our study system. This is likely related to the fact that the Rhone floods were of too low intensity in 2017 and 2019 to trigger the scouring process necessary to initiate a new succession (Steiger et al., 2005). Overall, these results show that, even in a regulated river system, the taxonomic composition of riparian habitats remained under the influence of multiple environmental gradients, acting at both the local and regional scales (Kuglerová et al., 2015; Renofalt et al., 2005).
4.3 Traits diversity and functional composition remain stable in space and time
In contrast to species richness, functional dispersion values remained stable over time at the corridor and gravel bar scales, with the exception of GB2. This suggests that the herbaceous species that disappeared between 2017 and 2019 on the Rhone gravel bars had redundant trait values in the community, showing some functional resilience of this regulated system to environmental changes (Su et al., 2020). Also, trait dispersion for SLA was higher on GB1, which is located in the only bypassed reach where bedload transport is still active (Vázquez-Tarrío et al., 2019). Greater co-occurrence of species with different resource acquisition and conservation strategies on ‘active’ bars may confirm that flood disturbance increases functional diversity (Biswas & Mallik, 2010; Lawson et al., 2015) and suggests the use of leaf traits as potential indicators of riparian habitat dynamics. Conversely, the functional dispersion value for seed mass was strictly influenced by maintenance measures. Indeed, ploughing by increasing microtopographic heterogeneity allows a variation in flood frequencies and velocity conditions (Pollock et al., 1998), thus the deposition of light and heavy seeds on restricted surfaces. Controlling for this spatial effect, community-wide analyses confirmed that trait dispersion was nearly unaffected by changes in environmental conditions between the 2 years studied. However, functional diversity peaked under the same environmental conditions as taxonomic diversity. Specifically, we showed that bars covered by a thick layer of fine sediments supported species with a narrow range of trait values (see also Janssen et al., 2019). Thus, even for fixed bars in a highly regulated river system, as is the case of the Rhône, the co-occurrence of herbaceous species with different functional strategies was greater when they consisted of coarse sediments (Janssen, Piégay, & Evette, 2020). This is particularly true at low elevations, on surfaces more intensely flooded (Lawson et al., 2015).
Functional composition did not vary between years at the corridor scale. Within bars, changes in trait composition between 2017 and 2019 were restricted to one site. This confirms the stability of functional assemblages in space and time, probably due to the low amplitude of hydrological conditions over the 2 years studied on the Rhone River, and emphasizes that changes in species identity between gravel bars along the Rhône River do not result in changes in trait values. Regarding the influence of environmental conditions, soil factors explained up to 60% of the variation in the functional composition of riparian herbaceous communities. Specifically, we showed that accumulation of fine sediment on gravel bars resulted in an increase in mean plant height (Kyle & Leishman, 2009). Indeed, silt deposits, by promoting fertile environments (Asaeda & Rashid, 2012), favoured the development of taller dominant-species capable of pre-empting resources, such as P. australis (Minchinton & Bertness, 2003) and P. arundinacea (Chen et al., 2017). Moreover, as these taller species are more typical of lentic aquatic ecosystems, the high cover measured on the bars of the Rhone affected by fine sediment deposits may also reflect their adaptation to the fairly constant minimum flow that is observed most of the year in the bypassed sections. In addition to soil texture, elevation level also explained much of the functional composition, with an increase in mean seed mass from low to high elevation (Kyle & Leishman, 2009; McCoy-Sulentic et al., 2017). This tends to confirm that the colonization potential of light-seeded species (e.g., Artemisia annua, Juncus articulates and Veronica anagallis-aquatica) is higher in more frequently disturbed environments (Gurnell et al., 2008), this even in a highly regulated river system. Beside bar characteristics and certainly due to the low magnitude of inundations in the months preceding the field observations in 2017 and 2019, hydrological conditions did not explain much of the functional composition of communities. Overall, we showed that at the corridor scale, high species turnover along the longitudinal-climatic gradient did not lead to a shift in the functional composition of riparian communities, whereas at the gravel bar scale, trait–environment relationships varied strongly and consistently with local gradients in elevation and soil texture. This underlines, at least for highly regulated river systems and for the range of hydrological conditions studied, the greater relative importance of local processes in controlling the functioning of riparian communities.
5 CONCLUSION
By studying spatial and temporal variations in riparian herbaceous communities along the heavily modified Rhône River, we showed (1) that different gravel bars supported different species pools, which shared almost the same set of traits, and (2) that changes in richness over the 2-year period were restricted to species with redundant trait values in the communities, suggesting some functional resilience to environmental changes. Furthermore, by investigating the relative influence of different environmental factors on assemblage patterns, we showed that the temporal change in taxonomic diversity was spatially restricted to communities established on low-lying bar surfaces composed of coarse-grained sediments, which were more intensely and frequently flooded. In addition to being the most sensitive to environmental changes, the riparian communities established on these surfaces were also the most diverse in species and trait values, highlighting the importance of these habitats for the Rhone river biodiversity and functionality. However, in this regulated river system where water and sediment exchanges are very limited, we showed that fine sediment deposition threatens the functional and ecological integrity of gravel bars by inducing a strong decrease in species and trait diversity. Although consistent and supported by other studies on different systems, these results are still limited to a short observation period and concern an intensively managed river system, where a lagged response of the vegetation to environmental changes cannot be excluded. To better disentangle the relative influence of the many environmental parameters that condition riparian community dynamics, longer, replicated observational studies are needed. Finally, from an applied perspective and in order to improve the biodiversity and functionality of riparian habitats in regulated rivers, our results highlight the urgent importance of conserving still-functional gravel bars and promoting more morphologically complex riparian habitats. This could be achieved by developing management plans, at the corridor and reach scales, aiming at renaturalize the flow regime (e.g., by increasing and irregularizing baseflows, Lawson et al., 2015) and reactivate bedload supply and transport (e.g., through gravel augmentation, Staentzel et al., 2020) to maximize scouring and prevent fine sediment deposits. Applied to restoration operations aimed at improving lateral connectivity through the removals of dykes, our results suggest that it would be more beneficial for biodiversity to profile banks with a gentle slope and to favour materials with a predominance of coarse-grained sediments.
ACKNOWLEDGEMENTS
We thank Gilles Favier, Delphine Jaymond and Sophie Labonne for help in the field as well as Laura Troudet and Jérôme Le Coz from INRAE for extracting data from the hydrodynamic model. This work was cofunded by the Labex DRIIHM, French programme ‘Investissements d'Avenir’ (ANR-11-LABX-0010), which is managed by the French National Research Agency (ANR), the OHM Vallée du Rhône, the Agence de l'eau Rhône Méditerranée Corse and INRAE. It was also conducted within the Rhône Sediment Observatory (OSR), a multi-partner research programme funded through the Plan Rhône by the European Regional Development Fund (contract: CNRS n°144140), Agence de l'eau RMC (France, contract: 2018 1063), CNR, EDF and three regional councils (Auvergne-Rhône-Alpes, PACA and Occitanie). This work was performed within the framework of the EUR H2O'Lyon (ANR-17-EURE-0018) of Université de Lyon, part of the programme ‘Investissements d'Avenir’ (ANR-11-IDEX-0007) operated by the ANR.
Open Research
DATA AVAILABILITY STATEMENT
Data are available from the authors upon reasonable request.
