Long-term perspectives on the evolution of river deltas have provided useful knowledge capable of responding to pending questions related to the ongoing climate and environmental changes. Increasing utilization pressure on delta environments has necessitated increased attention to protect the socio-economic and ecological values. As a result, multiple local initiatives have been designed, aimed at mitigating environmental deterioration and implementing adaptive measures, but many such initiatives have shown limited success. This thesis uses a case study of Save River delta in Mozambique to explore the relation between geomorphological evolution and socio-ecological system dynamics in delta environments. In addition, key environmental variables that concern the society today are highlighted and discussed in a management perspective. The results of the study show the development of Save River delta from the mid-Holocene to the present. The geomorphological settings of the delta suggest a faulted coastline over which subsequent deposition of fluvial sediments has formed a protruding delta front. Between c. 3000 and 1300 years ago, fine-grained sediments accumulated on top of the delta-front in the proximal part of the delta. This type of material was deposited under intertidal conditions and supported the formation of mangrove habitat. The geographical distribution of the mangrove deposit was driven by successive stages of back-barrier swamp formation and sea-level change as the delta evolved. From c. 1300 years ago, the river delta started to receive fluvial sediments from pulses of floods forming an alluvial floodplain. These sediments have accumulated mainly on the fine-grained mangrove wetland deposit. All the geomorphological features have evolved in a shoreward-shifting pattern over time. Centennial to decadal changes observed in the delta have followed a predictable geomorphological pattern, which is also part of the millennial evolution. The mangrove system, the base for the socio-economic system, is consequently strongly affected by the geomorphological development of the area. An increasing sensitivity of socio-ecological systems to environmental stressors, e.g. floods, cyclones and erosion, has motivated multiple initiatives to work towards a sustainable management of delta environments. This thesis highlights the need for interplay between geomorphology and ecology, considering both long- and short-term dynamics of delta environments. Hitherto, management initiatives have been concentrated on fragmented interventions of controlling water flow, which have disrupted the natural dynamics by obstructing the sedimentation-erosion cycle. To change this trend, coastal planners need to consider the significance of natural processes, e.g. cyclones, floods, erosion and accretion, for the long-term ecological and social sustainability of delta environments. / <p>At the time of the doctoral defense, the following papers were unpublished and had a status as follows: Paper 1: Manuscript. Paper 2. Manuscript. Paper 3: Manuscript. Paper 4: Manuscript.</p>

<p>Marion Island is situated in the South Indian Ocean and belongs to the sub-Antarctic island group, Prince Edward Islands. The islands in the sub-Antarctic have over the past few decades been exposed to a warmer and drier climate trend. The aim of this thesis is to achieve better understanding of the small-scale spatial and temporal variability between Azorella selago andthe surrounding microclimate. Due to the consequences of climate change, the interactions between Azorella selago, landforms and soil processes are important for the future of the terrestrial ecosystems in the sub-Antarctic. The theory part in this thesis describes different processes and features that are essential to understand the context of this thesis. The energy balance and the insolation is shown to be an important aspect when looking at the spatialvariability of the microclimate. The summary of the results in the thesis is based on temperature and moisture measurements within two grids. One on the east and one on the west side of the island</p><p>The most important result from the measurements is that different weather conditions create different situations for the microclimate. The weather condition ‘sunny no wind’ created a high spatial variability in temperature on the ground, which was completely absent during overcast days. Temperature variability is highly dependent on cloud cover according to these results. Moisture changes also seem to be less weather dependent than temperaturechanges.</p><p>The data provide a first confirmation that an increase in sunshine hours gives increased spatial variability in temperature (not moisture) and soil frost. An increase in spatial variability of the microclimate within small areas could give rise to an expansion in the patchiness of soil frost processes in the landscape. The representivity of single point measurements of ground surface temperature should be questioned.</p><p>Patterns of areas with low moisture content within the grid correlate with points where measurements were taken on Azorella selago. The Azorella cushion could, according to the results of this thesis, be associated with dry areas within the grid. Azorella selago is thereby suggested to increase the spatial variability of moisture and also contribute to a locally drier microclimate. Moisture variability varies more between the east and west side of the island, than that it is weather dependent.</p><p>Shaded areas show a pattern of lower temperature than for the other variables under sunny conditions. If more shaded areas are created by for example landforms like Azorella selago or solifluction deposits, the temperatures would probably be lower and also create a wider spatial variability.</p><p>This study provides first data on the important interactions between Azorella selago and how it affects through spatial variability in micro-climate, ground frost potential and resulting soil disturbance by frost creep and solifluction.</p>

Marion Island is situated in the South Indian Ocean and belongs to the sub-Antarctic island group, Prince Edward Islands. The islands in the sub-Antarctic have over the past few decades been exposed to a warmer and drier climate trend. The aim of this thesis is to achieve better understanding of the small-scale spatial and temporal variability between Azorella selago andthe surrounding microclimate. Due to the consequences of climate change, the interactions between Azorella selago, landforms and soil processes are important for the future of the terrestrial ecosystems in the sub-Antarctic. The theory part in this thesis describes different processes and features that are essential to understand the context of this thesis. The energy balance and the insolation is shown to be an important aspect when looking at the spatialvariability of the microclimate. The summary of the results in the thesis is based on temperature and moisture measurements within two grids. One on the east and one on the west side of the island The most important result from the measurements is that different weather conditions create different situations for the microclimate. The weather condition ‘sunny no wind’ created a high spatial variability in temperature on the ground, which was completely absent during overcast days. Temperature variability is highly dependent on cloud cover according to these results. Moisture changes also seem to be less weather dependent than temperaturechanges. The data provide a first confirmation that an increase in sunshine hours gives increased spatial variability in temperature (not moisture) and soil frost. An increase in spatial variability of the microclimate within small areas could give rise to an expansion in the patchiness of soil frost processes in the landscape. The representivity of single point measurements of ground surface temperature should be questioned. Patterns of areas with low moisture content within the grid correlate with points where measurements were taken on Azorella selago. The Azorella cushion could, according to the results of this thesis, be associated with dry areas within the grid. Azorella selago is thereby suggested to increase the spatial variability of moisture and also contribute to a locally drier microclimate. Moisture variability varies more between the east and west side of the island, than that it is weather dependent. Shaded areas show a pattern of lower temperature than for the other variables under sunny conditions. If more shaded areas are created by for example landforms like Azorella selago or solifluction deposits, the temperatures would probably be lower and also create a wider spatial variability. This study provides first data on the important interactions between Azorella selago and how it affects through spatial variability in micro-climate, ground frost potential and resulting soil disturbance by frost creep and solifluction.

The eruption of Mount St. Helens in 1980 severely impacted the woody vegetation within the geomorphic floodplain as well as the morphology of the Upper South Fork Toutle River. Historic aerial imagery and LiDAR data were used in combination to create snapshots of the channel and vegetation in 1980, 1983, 1996, 2003, and 2014. This data was mapped and analyzed using GIS, with the primary focus on 2D channel change, vegetation change, and channel-vegetation interactions from 1980 to 2014. No vegetation was discernable in 1980-83 but the vegetation present in 1996 increased in area and in density from 1996 to 2014. The number of channels locations were dependent on vegetation density and presence while vegetation growth occurred predominately in areas previously occupied by the channel.

This research developed a novel workflow for combining different types and scales of data to understand the development of small, steep, and sheltered glacial forefields across space and time using the Avalanche glacier of the Selkirk Range, BC as a case study. As glaciers recede, symbiotic geomorphological and ecological feedback loops determine the ecological succession in recently deglaciated terrain, which can in turn effect landform stability and water quality downstream. In order to describe emergent land cover patterns in the forefield, this research uses Corenbilt’s (2007) fluvial biogeomorphic succession (FBS) framework to interpret a century of land cover changes. To do so, an experimental protocol was developed that combined remotely sensed data – repeat photographs, historic air photographs, satellite imagery, and digital elevation models – and data collected in-situ using a photo transect method. Analysis of more than a century of photographs determined that the Avalanche glacier is receding at a slower rate than has been observed in the region’s larger glaciers, subsequently leading to a slower rate of forefield habitat expansion. Still, all four stages of fluvial biogeomorphological succession were found across the Avalanche glacier’s forefield. It was found that in the Avalanche forefield, terrain age seems to place a limit on which successional stage is possible at any given location within the forefield, but topographic features like slope angle seemed to influence succession patterns within areas that had the same terrain age. Further research is needed to see whether these findings are consistent for similar steep, small, and sheltered glaciers in the region. / Graduate

Dans les écosystèmes, tels que les rivières, les marais salés, les mangroves, les dunes côtières, qui sont exposés à des flux hydrogéomorphologiques fréquents et réguliers (c’est-à-dire à des perturbations physiques), des rétroactions se mettent en place entre la géomorphologie (eau, sédiments et formes fluviales) et les plantes (par exemple Populus nigra L., Salix alba L., Salix purpurea L. dans les rivières). L’établissement de la végétation est contrôlé par des processus hydrogéomorphologiques qui, en retour, sont modulés par la végétation. De telles rétroactions contrôlent la dynamique des écosystèmes riverains. Dans cette thèse, nous avons abordé deux questions principales afin de mieux comprendre les rétroactions entre la végétation riveraine et les processus hydrogéomorphologiques : (i) comment la végétation riveraine répond-elle aux contraintes hydrogéomorphologiques ? (ii) comment et dans quelle mesure les plantes ingénieures, une fois établies, affectent-elles la géomorphologie fluviale ? Nous avons étudié ces questions sur la rivière Allier (France) à travers une approche emboîtée multi-échelles allant de l’échelle du patron paysager au trait de plante. Nous avons testé l’applicabilité de la méthode de photogrammétrie pour quantifier la réponse et l’effet de la végétation riveraine et des rétroactions biogéomorphologiques à différentes échelles spatio-temporelles (corridor, banc alluvial et individu). À l’échelle du corridor, nous avons recherché la signature topographique de la végétation riveraine dans le paysage, en utilisant des données photogrammétriques et LiDAR. À l’échelle intermédiaire du banc alluvial, nous avons étudié l’aptitude des trois espèces pionnières dominantes riveraines de Salicaceae (P. nigra, S. purpurea, S. alba) à s’établir et à agir comme ingénieurs d’écosystème en piégeant les sédiments fins. À l’échelle la plus fine du trait de plante, nous avons quantifié la relation existante entre les attributs de trait de réponse des jeunes plantes de P. nigra et leur exposition à trois niveaux différents de stress mécanique (tête de banc fortement exposée, queue de banc moins exposée, chute alluviale). Nous avons identifié les difficultés et les erreurs à ne pas commettre pour appliquer correctement la photogrammétrie dans les études des rétroactions biogéomorphologiques. En tout état de cause, la photogrammétrie s’est avérée être un outil performant pour quantifier un ensemble de paramètres pertinents pour répondre à des questions de recherche fondamentale aux trois échelles spatiales considérées. À l’échelle la plus large, la signature topographique de la végétation est particulièrement difficile à identifier en raison de la dynamique complexe des formes fluviales de la rivière Allier. Cependant, en concentrant les observations sur des zones de taille réduite et fortement connectées (bancs alluviaux bordant le chenal), la signature de la végétation a pu être identifiée par cette méthode. Elle semble augmenter avec la croissance de la hauteur végétale (progression temporelle de la succession biogéomorphologique), ce qui est en accord avec le modèle de succession biogéomorphologique fluviale (SBF). À l’échelle intermédiaire du banc alluvial, les rétroactions biogéomorphologiques pouvaient être bien identifiées. La capacité des plantes riveraines à s’établir et à agir comme ingénieurs d’écosystème dépend à la fois des espèces et de leur physionomie, de leur âge et de leur position respective sur les bancs alluviaux. À l’échelle la plus fine de l’individu, nous avons capturé la réponse morphologique et biomécanique contrastée de P. nigra à l’exposition aux différents niveaux de contrainte mécanique d’un point de vue de trait. Dans tous les niveaux hiérarchiques, des rétroactions biogéomorphologiques liées aux échelles ont été détectées et synthétisées dans un modèle conceptuel. [...] / N ecosystems, such as rivers, salt marshes, mangroves, coastal dunes which are exposed to frequent and regular hydrogeomorphic fluxes (i.e. physical disturbances), feedbacks between geomorphology (water, sediment and landforms) and plants (e.g. Populus nigra L., Salix alba L., Salix purpurea L. in rivers) can occur. Vegetation esta¬blishment is controlled by hydrogeomorphic processes which in turn are modulated by vegetation. Such feedbacks control riparian ecosystem dynamics. In this thesis, we addressed two main questions in an effort to better understand feedbacks between riparian vegetation and hydrogeomorphic processes: (i) How does riparian vegetation respond to hydrogeomorphic constraints? (ii) How, and to what extent, do established engineer plants affect fluvial geomorphology? We studied these questions through a nested multi-scale approach from landscape pattern to plant trait scales on the dynamic wandering Allier River (France). We tested the applicability of the method of photogrammetry to quantify the response and the effect of riparian vegetation and biogeomorphic feedbacks at different spatio-temporal scales (i.e. corridor, alluvial bar and individual). At the corridor scale, we searched for the topographic signature of riparian vegetation in the landscape, using photogrammetric and LiDAR data. At the intermediate alluvial bar scale, we investigated the aptitude of three dominant pioneer riparian Salicaceae species (P. nigra, S. purpurea and S. alba) to establish and to act as ecosystem engineers by trap¬ping fine sediment. At the finest, plant trait scale, we quantified the relation between response trait attributes of young P. nigra plants and their exposure to three different levels of mechanical stress (a highly exposed bar-head, a less exposed bar-tail, a chute channel). We identified some difficulties or failures to properly apply photogrammetry in biogeomorphic feedback studies. However, photogrammetry appeared as a useful tool to quantify a set of relevant parameters to respond to fundamental research questions concerning biogeomorphic feedbacks at the three nested spatial scales. At the broadest, the topographic signature of vegetation was not easy to capture because of the complex shifting mosaic of landforms of the Allier River. However, by focusing on more connec¬ted, restricted areas (i.e. alluvial bars), the signature of vegetation could be captured. It seems to increase with increasing vegetation height corresponding to the evolutionary phases of the fluvial biogeomorphic succession (FBS) model. At the intermediate, alluvial bar scale, biogeomorphic feedbacks could be well identified. The capacity of riparian plants to establish and act as ecosystem engineers depended both on species and their physiognomy, their age and their location on alluvial bars. At the finest, individual plant scale, we captured the contrasting morphological and biomechanical response of P. nigra to variable mechanical stress exposure from a trait perspective. In all hierarchical levels, scale-related biogeomorphic feedbacks were detected and described in a conceptual model. The three scales were considered as cycles composed of four different phases, which can have a variable temporality. The broadest spatio-temporal scale represents the evolution over several decades of the landscape mosaic resulting from the balance between constructive (vegetation establishment, growth and succession) and destruc¬tive (floods) forces. [...]

The linkage between barrier island morphologies and dune topographies, vegetation, and biogeomorphic feedbacks, has been examined. The two-fold stability domain (i.e., overwash-resisting and overwash-reinforcing stability domains) model from case studies in a couple of islands along the Georgia Bight and Virginia coast has been proposed to examine the resilience properties in the barrier dune systems. Thus, there is a need to examine geographic variations in the dune topography among and within islands. Meanwhile, previous studies just analyzed and compared dune topographies based on transect-based point elevations or dune crest elevations; therefore, it is necessary to further examine dune topography in terms of multiple patterns and processes across scales. In this dissertation, I develop and deploy a cross-scale data model developed from resilience theory to represent and compare dune topographies across twelve islands over approximately 2,050 kilometers of the US southeastern Atlantic coast. Three sets of topographic variables were employed to summarize the cross-scale structure of topography (elevational statistics, patch indices, and the continuous surface properties). These metrics differed in their degree of spatial explicitness, their level of measurement, and association with patch or gradient paradigms. Topographic metrics were derived from digital elevation models (DEMs) of dune topographies constructed from airborne Light Detection and Ranging (LiDAR). These topographic metrics were used to construct dune topographic state space to investigate and visualize the cross-scale structure of dune topography. This study investigated (1) dune topography and landscape similarity among barrier islands in different barrier island morphologic contexts, (2) the differences in barrier island dune topographies and their resilience properties across large geographic extents, and (3) how geomorphic and biogeomorphic processes are related to resilience prosperities. The findings are summarized below. First, dune topography varies according to island morphologies of the Virginia coast; however, local controls (such as human modification of the shore or shoreline accretion and erosion) also play an important role in shaping dune topographies. Compared with tide-dominated islands, wave-dominated islands exhibited more convergence in dune topographies. Second, the dune landscapes of the Virginia Barrier Islands have a poorly consistent spatial structure, along with strong collinearity among elevational variables and landscape indices, which reflects the rapid retreat and erosion along the coast. The dune landscapes of the Georgia Bight have a more consistent spatial structure and a greater dimensionality in state space. Thus, the weaker multicollinearity and higher dimensionality in the dataset reflect their potential for resilience. Last, islands of different elevations may have similar dune topography characteristics due to the difference in resistance and resilience. Notwithstanding the geographic variability in geomorphic and biogeomorphic processes, convergence in dune topography exists, which is evidenced by the response curves of the topographic metrics that are correlated with both axes. This work demonstrates the usefulness of different representations of dune topography by cross-scale data modeling. Also, the two existing models of barrier island dune states were integrated to form a conceptual model that illuminates different, but complementary, resilience properties in the barrier dune system. The differences in dune topographies and resilience properties were detected in state space, and this information offers guidance for future study’s field site selections.

Meandering rivers and associated vegetation communities are highly dynamic systems that interact through various geomorphic and successional processes. However, much is still unknown about these interactions. Studies that focus on system integration rather than examining fluvial-related and vegetation dynamics individually will benefit science and the management of river systems. Tree communities in riparian areas, although consisting mainly of bottomland hardwood species, can be very diverse. Diversity has been linked to environmental influences such as meander migration, and changes in elevation and soil texture. This study focused on a single meander bend of the lower Trinity River in southeast Texas. The purpose of this research was to examine interactions between soil texture variation and the establishment and succession of riparian tree communities, as such interactions contribute to the formation of complex riparian landscapes. A bend-scale approach was utilized to provide a detailed study of vegetation pattern and of soil texture resulting from sedimentation processes, to examine for any relationships between them. Aerial imagery was used to assist in interpreting patterns of vegetation succession. The field portion of the study collected species and size class data on trees and soil samples for textural analysis. These data were analyzed separately to understand variations in tree communities and soils, but also together, to determine any relationships between soil texture and what tree communities are able to establish. Mean annual flow data from gauges upstream and downstream of the site were analyzed for changes in flow following dam construction upstream, as river regulation could potentially alter the vegetation establishment regime. Results showed five distinct communities or zones of vegetation. Soils on the site were strongly skewed toward finer sands and high silt and clay content. Zone locations and community structure were not directly related to soil texture; however, given species had clear relationships of relative density or dominance with specific soil textures. No changes in flow were noted between pre- and post- dam construction periods, indicating that the riparian system at this site may operate under near-natural conditions. Further studies in species-soil texture interactions, and for rare and invasive species in particular, may prove beneficial in improving understanding of the complex functioning of riparian systems and in providing valuable information for their management and restoration.

Previous research in the Ouachita Mountains, Arkansas suggests that, on relatively thin soils overlying bedrock, individual trees locally thicken the regolith by root penetration into bedrock. However, that work was conducted mainly in areas of strongly dipping and contorted rock, where joints and bedding planes susceptible to root penetration are more common and accessible. This project extended this concept to the Cumberland Plateau, Kentucky, with flat, level-bedded sedimentary rocks. Spatial variability of soil thickness was quantified at three nested spatial scales, and statistical relationships with other potential influences of thickness were examined. In addition, soil depth beneath trees was compared to that of non-tree sites by measuring depth to bedrock of stumps and immediately adjacent sites. While soil thickness beneath stumps was greater in the Ouachita Mountains compared to the Kentucky sites, there were no statistically significant differences in the difference between stump and adjacent sites between the two regions. In both regions, however, soils beneath stumps are significantly deeper than adjacent soils. This suggests the local deepening effects of trees occur in flat-bedded as well as steeply dipping lithologies. Regression results at the Cumberland Plateau sites showed no statistically significant relationship between soil depth and geomorphic or stand-level ecological variables, consistent with a major role for individual tree effects. Nested analysis of variance between 10 ha stands, 1.0 ha plots, and 0.1 ha subplots indicates that about 67 percent of total depth variance occurs at, or below, the subplot level of organization. This highly localized variability is consistent with, and most plausibly explained by, individual tree effects. The effects of biomechanical weathering by trees are not limited to areas with strongly dipping and contorted bedrock. Variability of soil depth in the Cumberland Plateau is likely influenced by positive feedbacks from tree root growth, that these interactions occur over multiple generations of growth, and that the effects of trees are the dominant control of local soil thickness. Since lateral lithological variation was minimal, this study also provides evidence that the positive feedback from biomechanical weathering by trees leads to divergent development of soil thickness.

Au cours du XXe siècle, les lits de nombreuses rivières ont été sujets à l'installation de végétation alluviale. Dans le cas des rivières aménagées, cette tendance est souvent associée à des altérations géomorphologiques directes (extractions de granulats, endiguements, etc.) ainsi qu'à des modifications anthropiques de leur régime hydrologique et sédimentaire conduisant à une stabilisation du lit qui permet l'installation de la végétation. Cette végétation augmente le risque d'inondation en diminuant les vitesses d'écoulement et en augmentant les niveaux d'eau en crue. Par ailleurs la biodiversité est dégradée par la diminution des habitats pionniers caractéristiques de ces environnements. Manipuler artificiellement le régime hydrologique d'une manière qui pourrait limiter l'installation de végétation sur les bancs est une option considérée par les gestionnaires. Dans ce contexte, ce projet de thèse a pour objectif de comprendre les impacts des crues d'amplitude variable sur la destruction de végétation, et d'identifier les mécanismes associés. Le site d'étude sur lequel cette thèse se focalise est l'Isère en Combe de Savoie, une rivière à galets très aménagée des Alpes françaises.Dans le cadre de cette étude, la destruction de végétation a été étudiée à l'échelle du tronçon à partir d'une analyse des données hydrologiques, des photos aériennes, et des données topographiques disponibles pour la période 1996-2015. À l'échelle du banc, un suivi de terrain avant et après les événements hydrologiques marquants entre avril 2014 et septembre 2015 nous a permis d'étudier l'action des crues sur la mobilité sédimentaire et sur la végétation. Ces observations ont été complétées par une modélisation numérique bidimensionnelle de l'écoulement en crue.À l'échelle du tronçon (20 km), nous avons trouvé une corrélation très forte entre les volumes d'eau ayant transité dans le chenal sur une période donnée, et la destruction de végétation associée au cours de la période 1996-2015. Les débits associés à des temps de retour infra-annuels semblent permettre la destruction de végétation. Le mécanisme de destruction le plus efficace que l'on observe à cette échelle est l'érosion latérale ; les mécanismes prenant place à la surface des bancs sont très minoritaires. Cependant, les surfaces détruites sont modestes ; 3,4 % de la surface végétalisée est détruite annuellement en moyenne. À l'échelle du banc, la période de suivi de terrain a couvert une série de crues fréquentes (temps de retour < 1 an) et une crue de temps de retour 10 ans. Seule cette crue a partiellement détruit la végétation pionnière sur les bancs suivis. La destruction de végétation ligneuse jeune a eu lieu par le biais de quatre mécanismes : 1) déracinement par érosion de surface supérieure à 20 cm, 2) enfouissement sous une couche de sédiment grossiers supérieure à 30 cm, 3) déracinement par une combinaison d'érosion et de dépôt, et 4) érosion latérale en marge des bancs. La destruction de végétation est toujours associée à une mobilité sédimentaire importante.Ces résultats montrent qu'une crue très importante est nécessaire pour détruire la végétation par la mobilisation de la surface des bancs sur ce site. Par contraste, les débits forts mais non exceptionnels (temps de retour infra-annuel) sont en mesure de détruire la végétation par érosion latérale. Dans le cas de l'Isère en Combe de Savoie, il semble que l'utilisation de crues artificielles ne peut pas seule permettre de maintenir la largeur inter-digues libre de végétation. Pour la suite, on propose de s'intéresser à la destruction de végétation dans le contexte de la dynamique des bancs alternés plus ou moins végétalisés, en prenant en compte les apports et le transport des sédiments en plus de l'hydrologie. / Many rivers worldwide have seen vegetation establish within their beds throughout the 20th century. In the case of managed rivers, this trend is usually linked to direct geomorphological alterations (sediment mining, diking, etc.) as well as anthropic alterations of flow regime and sediment supply. These pressures have stabilized river beds, allowing vegetation to establish permanently. This vegetation increases the risk of flooding by decreasing flow velocities and increasing water levels. In addition, the associated reduction in availability of pioneer habitats characteristic of these stabilized environments typically degrades biodiversity. Managing hydrology in a way that would limit vegetation establishment on bars presents an interesting management option. In this context, our study was aimed at understanding the impacts of floods of varying magnitude on vegetation removal, as well as identifying and quantifying the underlying mechanisms. This work focused on the Isère River, a heavily managed gravel bed river located in the western French Alps.Vegetation removal was studied at the reach scale using hydrological data, aerial photographs, and topographic data available between 1996 and 2015. At the bar scale, field monitoring before and after floods from april 2014 to september 2015 allowed us to document the impact of floods on sediment mobility and vegetation. A 2D numerical model was used to document fine scale hydraulics.At the reach scale, we found a strong correlation between water volume flowing through the river channel and the amount of vegetation removal. Discharges with return intervals of less than one year seem to have an impact on vegetation removal. The main mechanism observed from aerial photographs was lateral erosion; surface processes were negligible in comparison. However, global vegetation removal was modest: since 1996, on average 3,4 % of vegetated area was removed annually. At the bar scale, our study period permitted monitoring of a series of high frequency floods (return interval < 1 year) and a 10-year food event. Only the largest flood partially removed pioneer vegetation from bars. Young vegetation removal occurred through four different mechanisms: 1) uprooting by surface scour > 20 cm, 2) burial under a thick layer of coarse sediments > 30 cm, 3) uprooting by a combination of surface scour and sediment deposition resulting in no net topographic change, and 4) lateral erosion of bars. Vegetation removal was always associated with significant sediment mobility.We conclude that on the Isere River a very important flood is required to remove vegetation by mobilizing bar surfaces. In contrast, high but not exceptional flows (return interval < 1 an) are capable of removing vegetation through lateral erosion. However, artificial floods alone are unlikely to maintain the full width of the channelized bed of the Isere River free of vegetation. In the future, vegetation removal needs to be studied in the context of alternate bar dynamics with or without vegetation. It seems necessary to consider sediment transport as well as hydrology to understand the overall dynamics of the bed.