Putting down roots: Afforestation and bank cohesion of Icelandic Rivers

Riparian vegetation is widely recognized as a critical component of functioning fluvial systems. Human pressures on woody vegetation including riparian areas have had lasting effects, especially at high latitude. In Iceland, prior to human settlement, native downy birch woodlands covered approximately 15%–40% of the land area compared to 1%–2% today. Afforestation efforts include planting seedlings, protecting native forest remnants, and acquiring land areas as national forests. The planted and protected nature of vegetation along rivers within forests provides a unique opportunity to evaluate the various taxa within riparian zones and the channel stabilizing characteristics of the vegetation used in afforestation. We investigated bank properties, sediment textures, and root characteristics within riparian zones along four rivers in forests in Iceland. Bank sediment textures are dominantly sandy loam overlying coarser textures. Undercut banks are common because of erosion of the less cohesive subsurface layer. Quantitative root data indicate that the woody taxa have greater root densities, rooting depths, and more complex root structures than forbs or graminoids. The native downy birch has the highest root densities, with <1 mm roots most abundant. Modeling of added bank cohesion indicates that willow provides up to six times and birch up to four times more added cohesion to the coarse sediment textures comprising stream banks compared to no vegetation. We conclude that planting and protecting the native birch and willow helps to reduce bank erosion, especially where long‐term grazing exclusion can be maintained.


| INTRODUCTION
Riparian vegetation is widely recognized as a critical component of functioning fluvial systems, influencing channel planform (Gurnell, 2014;Satchithanantham et al., 2019), bank stability (Simon & Darby, 1999;Thorne, 1990), and diverse aquatic and terrestrial habitats (Naiman et al., 1993).Throughout history, human pressures on woody plant dynamics including riparian areas, especially in high latitude regions, have had long-lasting effects at the landscape scale (Normand et al., 2017).In Iceland, prior to human settlement in ca.870 AD (Vésteinsson et al., 2002), native birch woodlands covered approximately 15%-40% of the land area (Hallsd ottir, 1995;Ólafsd ottir et al., 2001).Dramatic loss of natural forest and woodlands began after the arrival of Norse settlers who harvested timber for fuel, building materials, and charcoal production, and converted woodlands to grasslands for grazing and haying.Post-settlement deforestation coincided with the onset of massive soil erosion across the country (Arad ottir & Eysteinsson, 2005;Arnalds, 1987;Dugmore & Buckland, 1991;Gíslad ottir et al., 2010).Iceland's landscapes remained largely treeless for centuries.Currently, native and planted forests cover between 1% and 2% of the land area of Iceland (Eysteinsson, 2017;Ólafsd ottir et al., 2001), with remaining birch woodlands considered remnants of the former dominant vegetation (Sigurðsson, 1977).Livestock grazing, natural disturbances from eolian transport, tephra fallout, glacial river floods, ice break up, and pathogen outbreaks, in addition to long-term climate change contribute to ongoing landscape degradation across Iceland (Dugmore et al., 2009;Greipsson, 2012).Efforts to mitigate deforestation and severe soil erosion began in the early 1900s through protection from grazing, erosion control measures, and afforestation.

| Afforestation history in Iceland
Afforestation in Iceland has a 120-year history, beginning with protecting birch woodland remnants from grazing in the first half of the 20th century, with greater emphasis on planting since around 1950 (Eysteinsson, 2017).Currently, planting is at 6 million seedlings per year to achieve a target goal of 5% coverage in the next 50 years, both on private and public land (Fountain, 2017;Halld orsson et al., 2008).Birch woodlands have also expanded naturally in recent years, at an average rate of 360 ha/year, due to changes in land use (Keller et al., 2021).The Icelandic Forest Service (IFS), forestry societies, and individuals are involved in forest protection and cultivation across Iceland, usually with multiple-use management goals (Eysteinsson, 2017;Halld orsson et al., 2008).Forests in Iceland are generally discrete areas, most often a few tens to hundreds of hectares, surrounded by much larger areas of treeless land.
Many forests include rivers where afforestation has occurred directly within riparian areas or where woody vegetation has expanded naturally into riparian zones after protection from grazing.
Previous research assessed the effects of afforestation on ground vegetation, fungi, soil and surface invertebrates, and avifauna (Halld orsson et al., 2008), but no assessment of the effects of afforestation on river systems in Iceland has been completed despite the importance of rivers as a nexus of erosion.The planted and protected nature of vegetation along rivers within national forests in Iceland provides a unique opportunity to evaluate the various taxa within riparian zones and the channel stabilizing characteristics of the vegetation used in afforestation.

| Vegetation and river dynamics
Vegetation exerts a critical control on river dynamics (Corenblit et al., 2007;Coulthard, 2005) both within the channel below water level and adjacent to the channel on floodplains and banks.Within the channel, fluid drag on vegetation reduces flow velocity and decreases flood erosion (Gurnell, 2014).On adjacent floodplains, riparian vegetation increases overbank roughness (Bywater-Reyes et al., 2018) and extensive rooting of vegetation increases bank strength that protects sediment from erosion (Abernethy & Rutherfurd, 2000;Pollen-Bankhead & Simon, 2010).Abundant plants on floodplains also accelerates the rate of chemical weathering (Corenblit & Steiger, 2009), enhancing formation of clays and further increasing bank cohesion.
In addition, riparian vegetation type influences channel width and lateral migration rate (McBride et al., 2008;Satchithanantham et al., 2019).Vegetation increases the stability of stream banks through increased soil shear resistance by providing both hydrological and mechanical stability (Abernethy & Rutherfurd, 2000;Pollen, 2007;Simon & Darby, 1999;Simon and Collison, 2002;Thorne, 1990).The hydrological effect of vegetation on bank stability derives from canopy interception, transpiration, and water absorption by roots, which influence soil moisture content (Simon and Collison, 2002;Pollen, 2007).The mechanical effect of riparian vegetation derives from root reinforcement forming a soil-root matrix that increases soil cohesion (Abernethy & Rutherfurd, 2000;Thorne, 1990).Roots of trees and herbaceous species are fibrous in nature with high net tensional strength that enhances bank strength.This increased soil strength, or added cohesion (c r ), varies with soil texture, vegetation type, and establishment age.In an analysis of streambank cohesion as a function of vegetation type, Polvi et al. (2014) found that greater bank cohesion is imparted by woody vegetation through bankstabilizing root properties, such as deeper roots, relative to nonwoody vegetation.
The presence of riparian vegetation on stream banks also provides numerous ecological benefits as vital centers of biodiversity (Naiman et al., 1993).Riparian areas retain sediments, nutrients, and pollutants, and maintain habitat for the entire aquatic ecosystem, including invertebrates, amphibians, reptiles, birds, and mammals (NRC, 2002).Riparian vegetation that overhangs the channel provides cover, is an allochthonous source of organic material as food for fish and is a primary food source for invertebrates (Naiman & Decamps, 1997).Additionally, the channel banks and vegetation within riparian areas make up the substrate for insects emerging from the water, that in turn provide a food source for breeding and migratory birds (Graf et al., 2002).Researchers in Iceland found a strong link between vegetation cover of river catchments and aquatic ecosystem production, length of food webs, and population density of fish (Gíslason et al., 1998;Medelyte et al., 2010;Stefánsd ottir, 2010).In addition, Gíslason et al. (2002) noted that allochthonous material derived from riparian vegetation contributes to aquatic food resources along rivers in Iceland.

| Channel morphology of rivers in Iceland
Most research on channel morphology of rivers in Iceland focuses on braided proglacial rivers (Maizels, 1995;Marren, 2005), the effects of contemporary glacial outburst floods (jökulhlaups) (Dunning et al., 2013;Snorrason et al., 2002a), or Holocene jökulhlaups as agents of canyon cutting and landscape evolution (Baynes et al., 2015;Carrivick et al., 2004;Wells et al., 2022).Other investigations of river forms and processes and the role of riparian vegetation along rivers in Iceland are limited.One exception is Ielpi (2017) The authors attribute changes in soil thickness to episodic channel movements, river bank erosion, and patterns of vegetation change.
We are not aware of other analyses documenting the types of taxa within riparian areas or the effectiveness of afforested taxa in enhancing stream bank cohesion of Icelandic rivers.
Here we investigate riparian taxa and stream bank properties along alluvial rivers within national forests in Iceland.We address the following questions: (i) What are the dominant taxa within riparian zones in national forests?(ii) What are the stream bank properties, sediment textures, and root characteristics of taxa within these riparian zones?and (iii) Which taxa impart the greatest added cohesion to banks within riparian zones to guide future afforestation efforts?

| STUDY SITES
Iceland is a volcanic island located in the north Atlantic Ocean between 63 and 66.6 north latitude and 13 and 24 west longitude, where rifting along the Mid-Atlantic Ridge and a mantle hotspot coincides (Einarsson, 2008;Saemundsson, 1979) to create the 103,000 km 2 island.Volcanic eruptions occur every 4-5 years (Thordarson & Höskuldsson, 2008).Glaciers cover approximately 11% of the land surface of Iceland (Björnsson & Pálsson, 2008) with many volcanoes located beneath glacial ice.Abundant subglacial volcanic activity generates jökulhlaups that transport and deposit glacial-fluvial and volcano-fluvial sediments (Arnalds et al., 2016).The dominant bedrock is basaltic (J ohannesson & Saemundsson, 1998), with glacial and glaciofluvial deposits covering areas that were formerly glaciated.
Andosols, derived from volcanic materials, are the dominant soil order (Arnalds, 2008).Andosols are highly vulnerable to erosion by water and wind (Arnalds et al., 2001).Tephra layers incorporated into soils are widely used in tephrachronology to understand volcanic eruption history and settlement patterns (e.g., Dugmore & Newton, 2012).
Iceland's lowlands are in the boreal belt with a relatively mild maritime climate.The North Atlantic Current brings warm waters to the southern shores of Iceland while the cold East Greenland Current affects the west and northern portions of the country (Einarsson, 1984;Ólafsson et al., 2007).Low pressure systems and the periodic storms that blow in from all directions create frequent strong winds.The mean annual temperatures are commonly 0 to +4 C in the lowlands and mostly 0 to À4 C in the highlands.Annual precipitation is variable and averages 400 mm in the northeast and approximately 3500 mm in the southeast and mountainous regions (Einarsson, 1984).Up to 50% of the annual precipitation may fall as snow in north Iceland.The vegetation is much influenced by past and present land-use (deforestation, sheep grazing) with extensive grassland, mire plant communities, and heathland vegetation of dwarf shrubs and grass.Glaciation history, isolation, and climate combined result in a species-poor flora.Vegetation is sparse at higher elevations, and areas of active volcanism, frequent flooding or active wind erosion are practically devoid of vegetation.
We investigated bank properties and vegetation along four alluvial rivers in Iceland within three national forests and one forestry society forest (Figure 1).Drainage areas vary from 84 to 2455 km 2 (Table 1).Rivers in Iceland are characterized as dragár, controlled by direct runoff or snowmelt; jökulár, dominated by glacial melt; or lindár, spring fed with constant flow throughout the year (Rist, 1956).Two of the four study rivers are dominated by snowmelt hydrology, one is a mixed river receiving both glacial and snowmelt, and one is spring fed.Peak discharge of dragár occurs in spring during snowmelt, but winter floods may occur when maritime depressions saturated with warm, moist air produce rain on snow events.Rivers in Iceland are subject to ice-break up floods, typically in spring, but also during unseasonably warm periods in winter.None of our study rivers are within active volcanic rift zones where geothermal heat influences hydrologic regime.

| METHODS
We identified forests along alluvial rivers where riparian vegetation was planted and protected from grazing via fencing or where vegetation was establishing naturally and protected.Four rivers met our criteria, the Sandá (south Iceland), Örn olfsdalsá (west), Blanda (north) and Fnj oská (north) (Figure 1, Table 1).We identified vegetation, measured bank characteristics, sediment textures, and root properties of riparian taxa at 15 planted or protected (from livestock grazing) treatment sites and 6 unplanted or unprotected control sites.
In the field, we inventoried the dominant taxa and vegetation functional groups along the banks and adjacent floodplain with a focus on taxa that influence mechanical properties of the bank.We surveyed bank profiles using a laser rangefinder and survey tape, including any bank undercuts (Figure 2).After surveying, a vertical face of the bank was cleared using hand shovels and brushes to expose the roots and bank stratigraphy.A 0.6 Â 0.6 m 2 vegetation grid with 0.1 Â 0.1 m 2 openings was then fixed against the bank (Figure 2) and root diameters were measured with digital calipers.
Root diameters were classified into six size categories: <1, 1-2, 2-3,   Collison , 2002).The stratigraphy of the bank cut was described and photographed in the field.Between two and eight soil samples were collected within heavily rooted layers from the bank profiles (total of 18) and analyzed for sediment texture as percent sand, silt, and clay by the hydrometer method (Gee & Bauder, 1986).
Statistical analyses were performed on the root data using R package software (R Core Team, 2020).We applied Welch's t-test to evaluate differences in means for data sets with unequal population variance.In addition, a one-way analysis of variance (ANOVA) test on ranks was performed to detect significant differences in nonparametric data (Kruskal & Wallis, 1952).All computed p-values were evaluated for significance using a threshold of α < 0.05.
The RipRoot algorithm of the Bank Stability and Toe Erosion (BSTEM) model (Simon et al., 2011) was used to quantify added cohesion using the field-based bank sediment textures and root characteristics.RipRoot is a fiber bundle model that predicts progressive root breakage and subsequent distribution of the applied load to the remaining roots (Pollen & Simon, 2005).Nineteen model runs were conducted for each of three sediment texture scenarios and four woody taxa.We also used RipRoot to calculate added cohesion of roots for a representative herbaceous plant (grass; perennial rye grass selected in the model) on one control reach (Table 3) for a total of 60 model simulations.

| Riparian vegetation inventory
A relatively diverse suite of trees, shrubs, forbs, and graminoids were found along the banks of the study rivers (Table 2) 2).

| Bank profiles and sediment textures
In general, planted and protected banks were higher (0.6-2.2 m) than control banks (0.5-1.2 m), noting the difference in sample size (15 vs.

6; Table 1
).A majority of banks had a distinct undercut zone within the upper 1 m.Sandy loam was the dominant soil texture within the upper 30-50 cm of all banks (Figure 2), with coarser alluvium beneath.
Where undercut, the eroded zone within the bank corresponded to a layer of coarser alluvium.

| Root characteristics
In total, 2572 roots were measured along 15 cutbanks in national forests and 6 adjacent control reaches that were outside the national forest and not protected by fence from grazing (Table 3).Although roots were present within the undercut banks, they were not measured because it was not possible to determine where the freehanging roots emerged from the bank.As a result, root counts are a minimum estimate of root density for undercut banks.Birch and willow had greater root densities (roots per square meter) than the conifers, cottonwood, and forbs (Figure 3).There was a marked decrease in root density for birch and willow below 50 cm.In addition, root branching and rooting depths of the trees and shrubs were more complex and exceed root branching and rooting depths of the forbs and graminoids (Table 2).Banks along the control reaches supported few woody plants and were dominated primarily by forbs and graminoids with shallower roots and less complex root branching.As a result, we focus on the root properties of the woody taxa (birch, conifers, cottonwood, willows) within planted and protected areas throughout the remainder of this analysis.

| Statistical analyses
For each of the woody taxa, we summed the number of roots of all size classes and normalized by bank area (m 2 ) occupied by roots to conduct statistical analyses between taxa.We found that birch had the highest root density of all the four woody taxa (Figure 4).We also T A B L E 2 Dominant plant taxa and root characteristics along study banks within riparian zones of forests in Iceland.found a statistically significant difference in mean root density between birch and willow ( p-value <0.001), which were also the two taxa with comparable sample sizes (n = 7).There was also a significant difference between birch-conifer, conifer-cottonwood, conifer-willow but at unequal and lower samples sizes (conifer n = 3, cottonwood n = 2).Birch root densities ranged from a maximum of 573 roots/m 2 (at bank FRB5 on the Fnj oská) to a minimum of 180 roots/m 2 (at SRB1 on the Sandá).
To investigate how mean root density varied with root diameter, we compared the mean root densities across the six root size classes   for the woody taxa.We found a statistically significant difference in root density for very fine roots (<1 mm) between birch-willow (and all other taxa combinations save birch-cottonwood and cottonwood-willow) but not for larger root class sizes (Figure 5).Significant differences for all taxa pair combinations were relatively minimal for root class sizes larger than <1 mm, mostly because of the small sample size, although significant differences do exist between birch and willow for 1-2 mm and >10 mm roots.The differences in y-axis values in Figure 5 help illustrate the higher number of <1 mm roots for all taxa measured, especially birch.The large number of <1 mm roots were not present at all depths within the bank profiles, however.Figure 6 shows the woody taxa in 10 cm increments within the bank profiles of the study rivers.There was a statistically significant difference in mean root density for <1 mm roots between birch and willow but only between 60 and 90 cm depth.Again, sample sizes varied so the comparison between birch and willow lends the most confidence.

| Added bank cohesion
For all afforested banks where roots were measured (Table 3), three RipRoot model runs were completed for a representative 1 m bank with two layers of sediment.Results for the dominant woody taxa indicated that willow contributed the highest added cohesion for all three grain sizes underlying sandy loam: gravel, sand, and sand and gravel (Figure 7).Added cohesion of the study banks with willow was greatest (6.1 kPa) for the sand and gravel texture underlying sandy loam.Birch contributed the next highest added cohesion (4.2 kPa) for the sand and gravel texture.Cottonwood and conifers contributed lower cohesion for all sediment textures compared to willow and birch.Although, the model results indicated that more cohesion was contributed by vegetation on banks with sand and gravel as the lower substrate unit, overall total cohesion was comparable for sandy textures if the greater effective soil cohesion of sand (gold bar, Figure 7) is considered.The maximum added cohesion of the modeled control bank with only grass roots was lower than at any other site (0.2 kPa; Table 3).

| Riparian vegetation characteristics
Our inventory of riparian vegetation along four rivers within national forests in Iceland is representative of the taxa that can be supported by the climate, soils, and land use practices within protected areas.
Although we found more taxa present within riparian zones than are reported in Table 2, especially forbs, sedges, grasses, and mosses, these species generally were less abundant and had shallower, sparser roots than the woody taxa.Therefore, we deemed additional plant identification of understory taxa unnecessary to answer our research questions about afforestation efforts and added bank cohesion.Banks along the Fnj oská had the greatest diversity of riparian taxa, with 15 of the 17 identified taxa present.The Blanda and Sandá had 10 and 14 of the 17 total riparian taxa, respectively.The one taxon unique to the Fnj oská was Norway spruce, planted as part of No roots were present within the coarse layer of sand, gravel, and cobbles (Figure 2b) in bank exposures of the Örn olfsdalsá, limiting the bank stabilization of the existing vegetation.
In general, root densities, depths, and branching were greater for the woody trees and shrubs than for the forbs and graminoids (Table 2, Figure 3).Although many of the forbs had a relatively high mean number of roots/m 2 , root densities decreased rapidly below 50 cm depths within banks, and root branching was simple.The relatively high density of roots for the birch (Figure 4) compared to other woody species was driven primarily by a greater number of very fine roots <1 mm (Figure 5).Moreover, birch was the only species that extended below 1 m, although our sampling method precluded the discovery of vertical roots directly beneath the trunk for other tree species.Additionally, and of note, is that black cottonwood along the Blanda had a relatively high number of roots/m 2 with roots that branched extensively, but only two banks were sampled.Black cottonwood is an important riparian species (Scott et al., 1996), and in humid environments is a facultative phreatophyte (Rood et al., 2011) with roots that access shallow groundwater.Root branching of black cottonwood is similar to birch and willow (Table 1), and thus likely contributes comparable added bank cohesion.Additional sample banks would allow for a more robust comparison between cottonwood and other woody taxa.It is likely that there are other root wood characteristics we did not measure that influence root tensile strength based on the root architecture (Stokes & Mattheck, 1996), including plate root, taproot, and heart root systems.

| Modeling interpretation and previous research
The modeling results in RipRoot were most sensitive to sediment grain size and root distributions, which in our case were based on field data (Table 3)  The RipRoot modeling indicated that willow then birch, contributed the most added cohesion for all sediment textures underlying sandy loam compared to conifer and cottonwood trees (Figure 7).The generally coarse-grained nature (Figure 2) and low effective cohesion (Figure 7) of alluvial banks in Iceland means that willow and birch may add four to six times more cohesion compared to unvegetated banks.
Control banks lacking more than grass roots were not as strong, with  and 5), which is consistent with other research that found more reinforcement from a large number of small roots with greater tensile strength per unit area (Pollen & Simon, 2005).Interestingly, we observed the highest added cohesion (6.1 kPa) based on modeling along a bank of the Sandá where we measured large willow sprouts from a formerly buried stem (12 roots > 10 mm in diameter; Table 3).
This finding aligns with other results showing that large roots (>5 mm) are also important in bank stabilization (Simon & Collinson, 2002).
Added cohesion from our RipRoot modeling produced values that were generally lower than those of other studies.Simon and Collinson (2002) found that birch contributed 8 kPa and willow contributed 2 kPa of added cohesion to river banks in the southeastern United States.Polvi et al. (2014) found that birch contributed 8-11 kPa, and willow 12-18 kPa depending on grain size of bank sediment within the southern Rocky Mountains of Colorado.Our lower values of added cohesion for birch and willow may be attributed to the relatively smaller size, younger vegetation, less fertile soils, different climatic regimes, and higher frequency of disturbances in Iceland.

1
Location of study rivers within forests in Iceland and images of each river with riparian vegetation.Image of Iceland from Esri.[Color figure can be viewed at wileyonlinelibrary.com]T A B L E 1 Characteristics of study rivers within forests in Iceland.River and forest name (year of planting and protecting vegetation) Lat long (deg dec min); elevation 3-5, 5-10, >10 mm.To ensure roots of individual species were accurately identified, we exposed roots where the ground surface was occupied by a single species.In addition, lateral root extent was measured, and root structure noted.Roots were described in the upper 1 m or until surface water was encountered.Other research indicates that root distributions decrease rapidly with depth with few roots extending beyond 1 m(Abernathy and Rutherfurd, 2000; Simon and The model tests if a given load exceeds the strength of the root system.Values of the root tensile strength and parameters for effective cohesion of the sediment textures, soil friction angle, and saturated unit weight of sediment were selected in the model.A representative undercut bank of 1 m height (average bank height) was simulated in RipRoot with three substrate scenarios as a two-layer model (sandy loam overlying gravel, sandy loam overlying sand, and sandy loam overlying sand and gravel).Additional cohesion due to roots, c r , was calculated by the model based on the type of vegetation, root distributions, and sediment textures determined from laboratory analysis.Similar taxa were selected in RipRoot to represent the taxa measured in Iceland (e.g., river birch (Betula nigra) used to represent downy birch) because the exact taxa are not an option in the model.We selected taxa most suited to high latitude, boreal climates, noting that root data from other regions may introduce some uncertainty.However, any uncertainty introduced is done uniformly for all model runs and the purpose for this study is one of comparison rather than documenting absolute root properties.
, including native downy birch (Betula pubescens), native dwarf, wooly, and tea-leaved or yellow willow (Salix herbacea, S. lanata, S. phylicifolia), and nonnative lodgepole pine (Pinus contorta), Norway spruce (Picea abies), and black cottonwood (Populus trichocarpa).Some of the plants observed were planted during afforestation, others were naturally established.Downy birch was noted along three of the four rivers and occurred as multi-stemmed shrubs averaging less than 2 m in height.Willow was F I G U R E 2 (a) Ternary diagram of sediment textures of bank samples collected from all four study rivers in Iceland.(b) Bank profile with 0.6 Â 0.6 m 2 vegetation grid along the Örn olfsdalsá (ORB2) showing the sandy texture in the upper 50 cm overlying sand and gravel.(c) Bank profile with vegetation grid along the Fnj oská (FRB4) showing sandy loam texture in upper 60 cm overlying sand, gravel, and cobbles.Holes in the bank are where samples were collected for grain size analysis.[Color figure can be viewed at wileyonlinelibrary.com] also found along three of the four study rivers with the three species of willow treated as a single taxon because wooly and tea-leaved willow were not distinguished in the field.Norway spruce, lodge pole pine, and black cottonwood were found along banks of one study river each.Forbs comprised the greatest number of taxa identified with one-scouring rush horsetail (Equisetum hyemale)-present along all four study rivers (Table Mean number of roots versus depth of bank for the dominant riparian taxa found along the four study rivers in Iceland.[Color figure can be viewed at wileyonlinelibrary.com]F I G U R E 4 Boxplots of roots/m 2 for woody taxa measured within banks along study rivers in Iceland.Upper and lower bounds of boxes are 75th and 25th percentiles, horizontal line is median, and vertical line shows the range of data.Letters above boxplots show significantly different species (p-value < 0.001).[Color figure can be viewed at wileyonlinelibrary.com] Abbreviations: B, Blanda; F, Fnj oská; LB, left bank; O, Örn olfsdalsá; RB, right bank; S, Sandá.a Maximum added cohesion from the three sediment texture scenarios modeled in RipRoot.

5
Boxplots of roots/m 2 for all woody taxa by root class size (mm) measured within banks along study rivers in Iceland.Upper and lower bounds of boxes are 75th and 25th percentiles, horizontal line is median, and vertical line shows the range of data.Black dots are outliers.Letters above boxplots show significantly different species (p-value, 0.001).Note that y-axis values differ.[Colorfigure can be viewed at wileyonlinelibrary.com] afforestation of the Vaglask ogur and Þ orðarstaðask ogur forests.Sediment textures within sampled banks of the Fnj oská were the finest grained of all the study sites (loam and sandy loam) (Figure2a,c).Conversely, the Örn olfsdalsá showed the fewest riparian taxa (4 of 17), and bank sediment textures were coarser than any other site (sand and loamy sand) (Figure2a).The low diversity of riparian vegetation along the Örn olfsdalsá may in part be controlled by the coarse sediment texture but also by the height of the banks ($2 m) which appear to exceed the rooting depths needed to access the groundwater table.
which minimizes uncertainty.Model results varied by up to 24% if different taxa were selected in the model (e.g., black willow, Salix nigra [4.4 kPa] vs. Geyer willow, S. geyeriana [5.6 kPa] for the sand texture).Again, we selected taxa in the model most suited for higher latitude, boreal climates because the exact taxa found in Iceland were not available in the model.Despite the moderate level of sensitivity to selected taxa in the model, the calculated added cohesion values are informative for comparisons between taxa.The model is less sensitive to bank geometry, so the 1 m representative bank was deemed useful to compare between the afforested taxa.

F
I G U R E 6 Boxplots of <1 mm roots for woody taxa in 10 cm increments within the bank profiles of the study rivers in Iceland.Upper and lower bounds of boxes are 75th and 25th percentiles, horizontal line is median, and vertical line shows the range of data.Black dots are outliers.Letters above boxplots show significantly different species ( p-value < 0.001).Note that y-axis values differ.[Color figure can be viewed at wileyonlinelibrary.com] modeled added cohesion values of only 0.2 kPa.Birch had the greatest preponderance of roots, especially <1 mm in diameter (Figures 4 photography to delineate channel morphologic changes and channel network expansion are beyond the scope of this research; however, these would address the influence of bank erosion on overall landscape degradation in Iceland.Such an analysis will help land use managers understand the processes and rates of channel widening, and potentially allow for prediction of rates of channel change under a future of variable hydrologic regime.In summer 2022, the highest discharge since 1995 occurred on the Fnj oská causing extensive bank erosion in areas upstream from Þ orðarstaðask ogur.In addition, a large ice-break up flood occurred along the Örn olfsdalsá in January 2022 downstream from our study banks within Norðtungusk ogur.These examples, while not unusual, underscore the dynamic hydrologic processes in Iceland that affect channel morphology.While we are uncertain whether the presence or absence of riparian vegetation influenced bank erosion in these two situations, results of this analysis support ongoing and expanded afforestation within riparian areas to enhance bank stability, increase riparian habitat, and inform ongoing forestry and soil conservation practices.Furthermore, this research may lead to a broader understanding of succession rates of vegetation change that appear to relate directly to grazing intensity(Bjarnason, 1971).Any effort to improve riparian habitat benefits river ecosystems through increased input of organic material, increased habitat complexity through bank shading, and formation of low velocity refugia for aquatic organisms.Large investments of time and money support the afforestation efforts across Iceland.Monitoring the effectiveness of afforestation is T A B L E 3 Number of roots by size class per m 2 for taxa as input for the RipRoot model, and model results for maximum added cohesion (c r ) at the study banks within riparian zones of forests in Iceland.