Valley‐bottom wetlands as temporary buffers for source‐to‐sink dispersal of sediment and associated phosphorus in dryland landscapes

Sediment trapping in wetlands is an essential ecosystem service, with implications for downstream ecosystems and water users. There is however limited empirical evidence of the contemporary rates and magnitude of sediment trapping in valley‐bottom wetlands. Time‐averaged suspended sediment samples from the inlets and outlets of forestry‐ and agriculturally‐impacted valley‐bottom wetlands with contrasting morphometric characteristics were compared in terms of suspended sediment and associated total phosphorous (total P) fluxes over annual scales, a dataset that was limited by Covid travel constraints. Although both wetlands were net depositional, contemporary suspended sediment mass balances for the agriculturally‐impacted wetland revealed a temporal change in the amount of sediment trapped over two water years (2019/2020 and 2020/2021), with trapping efficacies of 91% and 24%, respectively. The proportion of sediment trapped in the water year of 2020/2021 within the adjacent wetland, with a small commercially forested catchment, was up to 4 times higher than the agriculturally‐impacted wetland, which drained a larger catchment. Rates of total P retention showed that the agriculturally‐impacted wetland was a net sink for phosphorus in 2019/2020, but shifted to a source of phosphorus in 2020/2021 as the export of suspended sediment was enhanced. However, this contrasts with the forestry‐impacted wetland, which was a net sink of sediment and associated phosphorus during the one‐year study period of 2020/2021. Overall, despite data constraints, this study suggests that the efficacy of valley‐bottom wetlands in the delivery of sediment trapping and phosphorus removal ecosystem services varies temporally and spatially. This variability is potentially related to the interaction between annual rainfall regimes, catchment size and wetland geomorphic character. The temporary nature of sediment recycling processes could serve to balance wetland dynamics by regulating vertical growth of valley floors and longitudinal slope stability and should be considered in catchment management and wetland restoration planning strategies.


| INTRODUCTION
Globally, elevated fine-grained sediment delivery, transport and deposition in fluvial systems emanating from accelerated soil erosion in response to anthropic pressures (e.g., agriculture, mining and deforestation) and natural disturbance events (e.g., climatic oscillation patterns) represent critical challenges for water resource management and have serious ramifications for socio-economic development (Owens et al., 2005;Walling, 1999).Such changes in fluvial dynamics and fluxes of sediment can irreversibly alter the geomorphic functionality of downstream fluvial systems (Wohl et al., 2015), with adverse effects on lotic ecosystem functioning (Acornley & Sear, 1999;Bilotta & Brazier, 2008) and water quality (Owens et al., 2005).
Fluvially-integrated wetlands, such as floodplains and valleybottoms, support diverse and highly valued regulatory ecosystem services including the attenuation of 'source-to-sink' dispersal of water, sediment, particulate organic matter and dissolved solutes (Johnston, 1991;Kotze et al., 2009;Phillips, 1989).The relative effectiveness of a given wetland type in mediating these material fluxes has been attributed to the complex interplay between catchmentscale hydrometeorological, geomorphological, ecological, biological and land-use conditions (Grenfell et al., 2014;Kotze et al., 2009).
There are also linkages between the rates of sediment trapping and local-scale factors such as hydrogeomorphic setting, sediment supply, wetland morphometry, hydrological regime and surface roughness (Hupp, Woodside, & Yanosky, 1993;Olde Venterink et al., 2006).The relative importance of catchment-and local-scale conditions, and their interactions, are however spatially and temporally variable such that the extent of sediment trapping and transfer within wetlands varies in magnitude (e.g., Brunet & Astin, 2000;Kleiss, 1996;Phillips, 1989).
Fluvial sediment transport and depositional processes are spatially and temporally complex within wetland systems in dry and hydroclimatically variable environments, but are yet to be explored over multiple temporal scales (Grenfell et al., 2019;Tooth & McCarthy, 2007;Wiener et al., 2022).Moreover, most research attention has focused on the sediment-trapping dynamics of floodplain wetlands characterised by distinctly different hydro-geomorphological conditions (e.g., oxbows, backswamps and channel migration) and processes (e.g., lateral and oblique accretion, overbank inundation), rather than valley-bottom wetlands (Wiener et al., 2022).
Valley-bottom wetlands (also referred to as valley fill swamps, ciénegas or valley mire fens) are dynamic fluvial landforms, and are pervasive along discontinuous low-order streams of headwater catchments with relatively steep slopes (e.g., Ellery et al., 2009;Mactaggart et al., 2008).The gentle longitudinal slopes (0.1-3%; Ellery et al., 2009) and the presence of dense vegetation associated with this wetland type contribute to conveyance losses that result in ideal conditions for vertical sediment accretion.As such, these systems have the potential to constitute an important sink for clastic and organic sediments.However, the actual rates and magnitudes of sediment trapping efficacy over short (event and annual) timescales remain unknown while long-term (centennial to millennial) rates are scarce.A detailed understanding of trends in wetland sediment trapping dynamics at different temporal scales not only provides insight into catchment sediment budgets, but also into morphodynamic changes, resilience and evolutionary trajectories of these geomorphic landforms.Previous investigations that explored rates of vertical accretion at centennial and longer timescales have attempted to address the influence of geomorphological processes in the formation of valleybottom wetlands (e.g., Fryirs et al., 2014;Fryirs & Brierley, 1998;Pulley et al., 2018).Reported rates of long-term sediment accretion in valley-bottom wetlands, in a variety of climatic settings, were low and highly variable (0.01 to 0.40 cm year À1 ) (for review, see Wiener et al., 2022).The range of variability of long-term aggradational process rates is influenced by differences in a range of extrinsic (hydroclimatic regime, land-use) and intrinsic (valley morphology, type and cover of vegetation and lithology) factors (Fryirs et al., 2014).
Valley-bottom wetlands maintain a fragile balance between depositional and erosional processes, such that a change in the ratio of water to sediment supply may result in a significant shift in the dominant geomorphic processes (Grenfell et al., 2019).Studies on geomorphic processes have demonstrated that this wetland type alternates between multi-decadal intervals of storage, followed by periods of net erosion (Fryirs & Brierley, 1998;Pulley et al., 2018).Thus, as many wetlands have the potential to intermittently recycle deposited fine sediment and associated phosphorous, they may contribute to downstream water quality deterioration (Fryirs et al., 2007).
Although phosphorous removal pathways in wetlands involve a complex combination of geomorphic sedimentation processes, as well as chemical and biological assimilation (Fisher & Acreman, 2004;Hoffmann et al., 2009;Reddy et al., 1999), several studies have shown that the efficacy of phosphorus retention and fluxes depends on sediment dynamics (accretion and budget) in wetland systems (e.g., Johnston, 1991;Noe & Hupp, 2005).In fluvial systems, phosphorous is primarily delivered and transported in association with fine sediment particles (Ballantine et al., 2008;Bowes, House, & Hodgkinson, 2003;Withers & Jarvie, 2008).As such, the fluxes and storage of phosphorus are typically closely associated with the dynamics of fluvial sediment and organic matter (Kronvang, 1992;Schwarz, Malanson, & Weirich, 1996).Consequently, factors influencing the dynamics of sediment trapping such as hydroperiod, vegetation density and sediment loading rates are also important in the dispersal and attenuation of fluvially-transported P within wetlands (Kotze et al., 2009).
Sediment composition (e.g., type, texture and geochemistry) and local environmental conditions (e.g., pH, temperature, soil water content and redox status) are also critical determinants of the capacity of deposited wetland sediments to retain phosphorus in particulate form (Records et al., 2016).The hydroclimatic regimes of dryland regions generate temporally and spatially variable fluvial process regimes (Jaeger et al., 2017;Tooth, 2013).Such variability can also lead to significant consequences for both sediment transfer and depositional processes as well as the biogeochemical cycling of P in wetlands (Jaeger et al., 2017;von Schiller et al., 2017).
As the extent of environmental pressures in dryland landscapes continues to increase in response to persistent human-induced influences (Tooth, 2018), it has become increasingly important to have a mechanistic understanding of the range of variability in sediment transport and biogeochemical processes in rivers and wetlands.This

| STUDY AREA
The study was performed in two low-order wetland catchments located within the upper Tsitsa River catchment, situated towards the northern Eastern Cape in South Africa (Figure 1).The Tsitsa River rises in the south-eastern hillslopes of the Drakensberg escarpment and forms a critical part of the 'Eastern Cape Drakensberg' strategic water source area (Le Maitre et al., 2018).This river catchment was selected as the Karoo Supergroup geology and grassland-type vegetation communities are widely represented across South Africa.In addition, despite human alterations, fluvial networks are reasonably natural in comparison to systems in other regions of South Africa which have been heavily modified through anthropogenic activities.
Bedrock lithology underlying the two wetland catchments is largely homogeneous and composed of sedimentary rocks of the Karoo Supergroup, which is locally dominated by erodible mudstones and erosion-resistant sandstones with less extensive resistant dolerite sill outcrops present within the upper catchment of the agriculturally-impacted wetland (Table 1; CGS, 2017).Soils and sediments emanating from these parent materials include primarily fine particle size fractions (silt and clay) with minor coarse sand and gravel (Le Roux & van der Waal, 2020;van der Waal, Rowntree, & Pulley, 2015).Therefore, the sediment load is dispersive and fine suspended sediment transport typically exceeds coarse bedload transport.Catchment topography of these wetlands is characterised by moderate to steep sloping hills, with mountainous areas occupying a small proportion.
The catchments' have an aridity index of 0.43 (ratio of MAP: PET; UNEP, 1997), which is classified as semi-arid (following Trabucco and Zomer, 2009).Mean annual rainfall (period: 1897-2020, Maclear weather station) in the catchment is 801 mm a À1 (coefficient of variation = 23%).Rainfall is confined to the austral summer months and is related to high-intensity convective systems associated with the tropical easterlies, in conjunction with the position of the South-Indian high-pressure system and intertropical convergence zone (Tyson & Preston-Whyte, 2000).The winter months are typically dry.The fluvial regimen is typically flashy owing to the small drainage areas and rainfall variability (van der Waal, pers.comm.).Both catchments are ungauged.
Catchment areas are 9.38 km 2 for agriculturally-impacted, to 2.84 km 2 for forestry-impacted.As illustrated in Figure 1b, agricultural activities account for 27% of the agriculturally-impacted catchment with patches of invasive alien trees along streams.Agriculture consists of the cultivation of rainfed crops (soybean, maize and oats), interspersed with natural pastures for cattle and sheep rearing and grazing.
The forestry-impacted catchment hosts extensive Pinus pinaster commercial plantation forests, alternated with Acacia mearnsii on steepsided hillslopes, with endemic 'East Griqualand Grassland' vegetation on valley margins, and small patches of cultivation (mainly oats) (Table 1 and Figure 1c).'East Griqualand Grassland' which is typically associated with the characteristic fine silty-clayey soil matrix of the area covered the remainder of the catchment areas (Mucina & Rutherford, 2006).Wetland vegetation comprises dense, relatively short (<1 m tall) tussock grasses (e.g., Cynodon dactlylon, Sporobolus africanus, Andropogon eucomus and Eragrotis plana) interspersed with sedges and rushes (e.g., Sieben et al., 2017).Both wetland systems often desiccate partly during the dry season.

| Study design
A mass balance approach was used to estimate the relative rates and amount of sediment and associated phosphorus trapped within the wetland over the period of study.Suspended sediment sampling sites were established at the junctions of inflowing tributary and outgoing reforming channels of two hydrologically and geomorphologically 'intact' discontinuous valley-bottom wetlands (Figure 1).These two wetland systems were selected as their 'natural' sediment and flow regimes were largely intact.While comparable in terms of hydroclimatic conditions, edaphic factors and native vegetation, these wetlands had contrasting catchment land-use conditions and differed in terms of slope and size (Table 1).

| Field sampling strategy and data collection
Time-integrated samples of suspended sediment were collected using standard suspended sediment samplers developed by Phillips, Russell, & Walling (2000) and evaluated by Perks, Warburton, and Bracken (2014).Various field and experimental studies showed that these samplers collect representative time-integrated suspended sediment samples to characterise transfer patterns and physicochemical composition in a wide range of fluvial landscapes (e.g., Ankers, Walling, & Smith, 2003;Smith & Owens, 2014).The sampler has several advantages: (1) it is easily constructed and inexpensive and allows multiple samplers to be deployed in remote field sites, (2) it is relatively easily deployable and robust during high-magnitude flow events, and (3) it reduces sampling bias as it collects a time-integrated sample over a range of flow conditions and requires limited maintenance.The samplers are especially useful in dynamic fluvial systems where sediment transport processes are characterised by highly unpredictable patterns.However, Perks et al. (2014) suggest that the suspended sediment sampler generally underestimates the actual suspended sediment mass fluxes by 60% or more and thus may not be representative of the absolute (or total) suspended sediment loads.This underestimation was related to a bias towards coarse-grained particles and the episodic nature of sediment transport processes.
Consequently, despite providing a good estimation of the relative amount, patterns and composition of suspended sediment, the actual magnitude of sediment trapping is likely underestimated.
The samplers consisted of PVC piping (110 mm in diameter, 1 m in length).The inlet was streamlined with a moulded funnel to minimise flow disturbance with a 4 mm inlet tube.The samplers were installed parallel to the channel bed at a depth of 60% of the measured water surface level at the time of deployment and secured with metal fencing posts installed in the channel bed.
The samplers of the agriculturally-impacted wetland were monitored for two consecutive years (2019/2020 and 2020/2021), while the forestry-impacted system was added for 2020/2021 year to account for spatial variability in catchment land-use and local wetland geomorphological conditions.Each sampling campaign lasted for approximately 12 months (hereafter water years referred to by date) and thus collected a time-integrated sample over a range of low-to high-magnitude rainfall events.This low sampling frequency was largely due to the logistical issues associated with the remote location of wetlands and difficulties associated with travelling during an extended Covid lockdown.
Sampling sites were located at channels immediately upstream and downstream of each wetland to establish the suspended sediment mass balances and trapping efficacies.In the agriculturallyimpacted wetland, three samplers were deployed both in the tributary inlet and in the reforming channels.Channels of the forestry-impacted wetland had a small cross-sectional area (<1 m in width) and thus only a single sampling device could be deployed at each channel sampling site.Triplicate sampler sets in the agriculturally-impacted wetland were equally distributed across the channel width to account for Samplers collected suspended sediment for the duration of the water year and were retrieved at the end of the wet season.The sampler contents were emptied into pre-rinsed polyethylene 10 L containers and allowed to settle for 48 h.The suspended sediment samples were retrieved via a combination of decanting, siphoning and centrifuging of the supernatant.
All samples were then oven-dried until a constant weight (at 105 C for ⁓48 h) and weighed (± 0.0001 g accuracy) to determine the total dry mass fluxes of suspended sediment.These masses were converted to kg day À1 to account for the differences in deployment periods and to facilitate comparisons.These daily fluxes were then upscaled to annual suspended sediment mass fluxes and given in kg year À1 .Estimates of sediment trapping efficacy for wetlands were then calculated from the suspended sediment mass fluxes in the tributary and reforming channels (Mekonnen et al., 2017;Verstraeten & Poesen, 2000) with the following equation: where SS in = SS mass flux (kg year À1 ) measured in the tributary channel; SS out = SS mass flux (kg year À1 ) measured in the reforming channel.
Morphological field surveys of channel and wetland reach crosssections and longitudinal profiles were conducted using a Differential GPS (±1 cm accuracy in x, y and z) at each of the wetland systems to quantify variation in bankfull-channel dimensions and wetland morphometry (see Supporting Information S1).Complimentary catchment morphometric parameters were derived from a 5-m resolution digital elevation model (SUDEM, GeoSmart Space, 2019).An estimation of bankfull discharge for each reach was derived from the morphometric data using standard equations.
Rainfall data were obtained between 2015 and 2021 (5-min intervals) from the PG Bison/Mooi River gauge (Figure 1) (Huchzermeyer, Van der Waal, & Schlegel, 2021).This data was used to characterise the seasonal rainfall regimes for the monitoring period and preceding hydrological years.The total daily rainfall data for the study period were classified into the following magnitude categories: low (5 to 10 mm), intermediate (10 to 20 mm) and large (>20 mm).In addition, a long-term rainfall record obtained from the South African Weather Service (SAWS) for the Maclear station was used for the estimation of the recurrence interval of extreme total daily rainfall events recorded during the study period.

| Laboratory analyses
All sediment samples were disaggregated with a mortar and pestle and dry sieved through a 1 mm steel mesh to remove biases associated with large organic fragments, flocs and coarse sand or gravel.
The retained sediment fractions were analysed for particle size distribution, organic content, total phosphorus and associated major elements.For particle size distribution, $1 g of the sample was pretreated with 50% hydrogen peroxide to remove fine organic material.
The sample was subsequently chemically deflocculated with sodium hexametaphosphate.Particle size analysis of the <1-mm fraction was undertaken by laser particle diffraction techniques with a Saturn DigiSizer 5200 (Micromeritics Instrument Corporation).Sediment particle size diameters were characterised and interpreted using the standard categories described in the Udden-Wentworth index (Wentworth, 1922).
Organic matter content (OM) of dried sub-samples (105 C) was measured using Loss-on-ignition (LOI), where the pre-weighed subsample of sediment ($3 g) was combusted for approximately 12 h at 550 C in a muffle furnace, reweighed (±0.0001 g) and used to calculate the remaining mass remaining (expressed as a percentage of initial mass) (Heiri, Lotter, & Lemcke, 2001).Total phosphorous (total P) and associated geochemical elemental concentrations were determined with a Thermo ICAP 6200 ICP-AES following microwave-assisted acid (nitric acid and hydrochloric acid) digestion.Since P in fluvial ecosystems is typically closely linked to metal complexes, measurements of total P were complemented by analyses of total iron (Fe), aluminium (Al), Manganese (Mn), Magnesium (Mg) and calcium (Ca) concentration.The recovery values on reference material WQB-1 of 98 to 103% were obtained for total P and 99% to 110% for selected geochemical elements.All sediment chemistry analyses were performed at the accredited Central Analytical Facility of Stellenbosch University (South Africa).

| Statistical analysis
Mass fluxes of total P were calculated using the product of the average total P concentration in the reach and mass fluxes of suspended sediment and expressed in mg year À1 .The relative proportions of total P trapped in wetlands were calculated from the mass flux of total Comparison of the characteristics of the two wetland catchments.

Forestryimpacted
Catchment area (km P in the tributary and reforming channels, as illustrated in Equation ( 2).
Total P retention efficiency % where TP in = average mass flux of total P (mg year À1 ) measured in the tributary channel; TP out = average mass flux of total P (mg year À1 ) measured in the reforming channel.
General descriptive statistics of suspended sediment fluxes and total P contents in the tributary and reforming channels of the agriculturally-impacted wetland were calculated using TIBCO Statistica 13.The distribution for each of the assembled datasets was examined for normality (and homogeneity) of residuals using the Shapiro-Wilk test and found to conform to non-parametric assumptions ( p < 0.05).A Kruskal-Wallis analysis comparing the tributary and reforming channels of the agriculturally-impacted wetland was used to test whether there were statistical differences in suspended sediment and total P fluxes among the tributary and reforming channels.This statistical analysis could not be done for the forestryimpacted system because of the insufficient sample size (n = 1) and low replication of sampling events, and thus did not allow a robust comparison between the two study systems.Spearman correlation coefficients were performed on all data to explore the magnitude of the relationships between total P, suspended sediment mass fluxes, sediment texture (D50, clay %, silt % and sand %) and selected geochemical elements.Data from the forestry-impacted wetland were also included in this analysis.

| Downstream changes in bankfull-channel and wetland morphometry
The agriculturally-and forestry-impacted wetland systems displayed distinctive downstream variations in morphology (longitudinal gradient, valley width, channel area) (Figures 2 and 3 and Supporting Information S2).The average longitudinal slope through the agriculturally-impacted wetland (0.61%) was distinctly lower than in the forestry-impacted wetland (2.07%) (Figure 2), which was partly related to the difference in the catchment area.
In the agriculturally-impacted wetland, the thalweg slope of the inflowing tributary channel decreased downstream from 0.82% to 0.20%.As the tributary channel terminated, the longitudinal slope rapidly increased to 0.71% in the proximal wetland reach at crosssections Fx 16 to Fx 20, while in the distal reach (Fx 21 to Fx 23), the slope was reduced to 0.36% (Figure 2a).Average bankfull width and Three well-defined reforming channels were located on the lower part of the wetland, and coalesced into a single-thread channel that becomes progressively narrower and shallower downstream.Widths and depths of the main reforming channel at cross-section Fx 24 to 27 were variable and ranged from 6.5 to 2.0 m and 1.02 to 0.36 m, respectively (Figure 2 and Supporting Information S2a).width (20 to 36 m) through the wetland (Figure 3b).The tributary channel and proximal wetland reach had relatively low slopes with values between 1.27 and 1.34% (Figure 2b).In the distal wetland reach, the longitudinal slope steepened to 3.62%, while slope lowering occurred along the thalweg of the reformed channel reach.Small distributary channels were present on the distal wetland reaches and drain into a relatively larger reformed channel (see Figure 3b and Supporting Information).

| Characteristics of seasonal rainfall events
The cumulative rainfall varied from 744 mm for water year 2019/2020 to 870 mm for water year 2020/2021 (Figure 4a,b).Maximum rainfall intensities recorded in 5 min for 2019/2020 (6.2 mm) and 2020/2021 (5.8 mm) were similar between the two water years.
Higher intra-annual variability in the volume of total monthly rainfall was recorded in 2019/2020, where coefficients of variation ranged from 108% relative to 84% in the subsequent water year.Low to moderate daily rainfall events (<5-10 and 10-20 mm day À1 ) predominated the rainfall sequence and accounted for 67.7% of the total cumulative rainfall during 2019/2020 and 56.6% during 2020/2021 (Figure 4c).
In comparison, intense, high-magnitude rainfall events (>20 mm) occurred less frequently (9-14 events) in both years and contributed 32.3% and 43.4% of the cumulative rainfall in 2019/2020 and 2020/2021, respectively (Figure 4c).The highest volume of daily rainfall was recorded in 2020/2021 with a total depth of 65 mm day À1 compared to 33 mm day À1 in 2020/2021.There were also apparent differences in the total amount of accumulated rainfall during the months before these large events (Figure 4b).The total preceding rainfall amounted to 620 mm during 2020/2021 and 390 mm in the preceding water year.A total daily rainfall event of 33 mm and 65 mm have an estimated return period of 0.72 and 5.39 years, respectively (Figure 4d).Similarly, the maximum hourly rainfall intensity recorded was lower for the 33 mm daily event in 2019/2020 (9.6 mm h À1 ) compared to the 65 mm event (19 mm hr À1 ) in 2020/2021.

| Suspended sediment flux dynamics and retention
The variation of suspended sediment mass fluxes for the tributary and reforming channels for the agriculturally-and forestry-impacted wetlands during the 2-year sampling period are illustrated in Figure 5.The largest suspended sediment masses were recorded at the incoming tributary channels.The average amount of suspended sediment transported into the two wetlands ranged between 0.10 and 0.14 kg year À1 (SD = 0.07 kg year À1 ), while the average amount of suspended sediment exiting through the reforming channels was between 0.01 and 0.08 kg year À1 .
In the agriculturally-impacted wetland, the tributary suspended sediment influx was relatively consistent for the 2-year study period with a mean of 0.11 kg year À1 during 2019/2020 and 0.10 kg year À1 during 2020/2021 (Figure 5a).In contrast, the magnitude of suspended sediment export increased markedly from 0.01 kg year À1 The total amount of suspended sediment input to the forestryimpacted wetland (0.14 kg year À1 ) was within the range of values obtained in the tributary channel of the agriculturally-impacted wetland (0.02 to 0.24 kg year À1 ; Figure 5b).The total amount of suspended sediment exported from the forestry-impacted wetland was 0.03 kg year À1 (24%) and lower than the range for agriculturallyimpacted wetland (0.05 to 0.10 kg year À1 ) during the same year (2020/2021).
Both wetland systems were a net sink for suspended sediment of 0.02 to 0.10 kg year À1 , which reflects an average trap efficacy of 67.71% (range = 24 to 91%; Table 2).The average net suspended sediment release of the study wetlands as calculated by a mass balance was $0.01 to 0.08 kg year À1 (or 9% to 76%) during the entire study period.The highest proportion of suspended sediment trapping was observed in the agriculturally-impacted wetland during 2019/2020, with a trapping efficacy of 91%.However, the trapping efficacy declined to 24% in the subsequent year.A comparison between the two study wetlands (Table 2) during the same period (2020/2021) showed that the efficacy and magnitude of suspended sediment trapping for the agriculturally-impacted wetland was lower than that of the forestry-impacted wetland (Table 2).

| Particle size composition and variability
The average median particle diameter of suspended sediment in the reforming and tributary channels of the agriculturally-impacted wetland ranged from fine clay (2.7 μm) to medium (23.7 μm) silt (Figure 6a,b and Table 2).Clay (22% to 24%) and silt (52 to 70%) proportions were relatively consistent between samples, while in the reforming channel, suspended sediment had consistently higher proportions of sand (Figure 6a,b).In particular, there was an apparent increase in the d50 values and proportion of sand-sized particles (23.3%) in the reforming channel during 2020/2021 (Table 2 and   Figure 6b), which coincided with the higher magnitude of suspended sediment export.
In the forestry-impacted wetland, median particle size values were in the same range as in the agriculturally-impacted wetland with fine silt-sized fractions dominating the suspended sediment load (Figure 6c and Table 2).There was also a minimal difference in the d50 values in the tributary (7.7 μm) and reforming (10.5 μm) reaches.
Suspended sediment from the tributary reach contained similar proportions of clay-sized particles relative to the reforming channel of the forestry-impacted wetland at 24.7% and 22.6%, respectively (Figure 6c).In contrast, the reforming channel had the highest sand contents at 24.4%, while silt contents were highest in the tributary reach (70.03%) (Figure 6c).

| Total P concentration dynamics in SS and percentage reduction
Total P concentrations of suspended sediment in the tributary and reforming channels of the two wetland systems varied from 124.03 to 676.85 mg kg À1 , with an average value of 334.53 mg kg À1 (Figure 7a,b).The highest averaged concentrations of total P (542.75 ± 124.95 mg kg À1 ) were measured in the reforming channel of the agriculturally-impacted wetland during 2019/2020 (Figure 7a).Mean concentrations of total P in the tributary channel (286 ± 11.74 mg kg À1 ) were generally lower than in the reforming channel during this period.Organic contents and Fe concentrations were also consistently higher in the reforming channel as compared to the tributary channel during both years (Table 3).
In the agriculturally-impacted wetland, there was a substantial change in total P concentrations between 2019/2020 and 2020/2021 for both the tributary and reforming channel reaches, though not statistically significant ( p > 0.05) (Figure 7a).Specifically, mean concentrations of total P during 2020/2021 decreased from the previous year by 30% and 51% for the tributary and reforming channels, respectively.Although variable, mean total P concentrations (262.02 ± 120.02 mg kg À1 ) were also greatest in the reforming channel during this period.Organic contents and Fe concentrations exhibited the same temporal pattern as total P concentrations (Table 3).
In the forestry-impacted wetland, the tributary channel (543.41 mg kg À1 ) exhibited a higher total P concentration relative to the reforming channel (266.21mg kg À1 ) (Figure 7b).The concentration of total P in the tributary channel of forestry-impacted measured in 2020/2021 was similar to the mean total P concentrations in the reforming channel of the agriculturally-impacted system during 2019/2020 but higher than in 2020/2021.OC values and Fe concentrations of the reforming channel were generally higher than in the tributary channel (Table 3).
The concentration of total P in suspended sediment was found to be higher downstream of the agriculturally-impacted wetland in both 2019/2020 and 2020/2021, increasing by 91% and 30%, respectively (Table 3).However, when expressed per unit mass of suspended sediment flux, the average removal efficacy of total P at the agriculturallyimpacted wetland equated to 82.77% during 2019/2020 compared to À4.71% in the 2020/2021 period.The mean net flux balance of total P in the agriculturally-impacted wetland was 26.07 mg year À1 during 2019/2020, while the net release amounted to À0.94 mg year À1 for 2020/2021.In contrast, the net flux balance of total P was 60.09 mg year À1 in the forestry-impacted wetland during the 2020/2021 period, with a removal efficacy of 89.50% (Table 3).
Concentrations of total P were significantly positively correlated with OC (R 2 = 0.89, n = 14, p < 0.001) and Fe (R 2 = 0.75, p < 0.001) (Figure 8a,b).Total P concentrations were also significantly but weakly correlated to Ca (R 2 = 0.41, p < 0.05) and Al (R 2 = 0.32, p < 0.05) (Figure 8c,d).There was however no significant relationship between total P and the percentage of silt and clay in suspended sediment ( p > 0.05).to the broad range of between 6% and 77% in large river floodplains (Kleiss, 1996;Walling and Quine, 1993;Walling, Owens, & Leeks, 1998, 1999;Tockner et al., 1999;Brunet & Astin, 2000;Walling & Owens, 2003;Kronvang et al., 2007;Gellis et al., 2017;Mekonnen et al., 2017).In comparison, the proportional range of 30% and 98% recorded for best management practices, including vegetated filter strips (Abu-zreig et al., 2004;Lee et al., 2000;Sheridan, Lowrance, & Bosch, 1999), grassed waterways (Fiener & Auerswald, 2003, 2005;Mekonnen et al., 2017) and constructed wetlands (Braskerud, 2001;Kadlec et al., 2010), are also similar to those  Despite the overall net accumulation of sediment in these systems, the magnitudes and proportional rates of sediment trapping and downstream export in these systems varied both temporally and spatially.For instance, the relative proportion of suspended sediment trapping decreased substantially in the agriculturally-impacted wetland during 2020/2021 (24%) compared to the preceding (2019/2020) year (91%).The year 2020/2021 presented relatively higher amounts of total rainfall (870 mm), maximum 24-h rainfall (65 mm) and antecedent rainfall (620 mm) than in the previous year with values of 744, 33 and 390 mm, respectively.Hence, the temporal change in sediment trapping efficacy and transfer in this wetland system appears to occur, at least in part, as a result of the apparent difference in annual rainfall patterns and magnitude.Similarly, a study performed on a floodplain wetland in Arkansas (USA) determined that higher amounts of annual rainfall generally resulted in lower rates of sediment retention (Kleiss, 1996).However, no previous studies exist that have explicitly provided comparable estimates for temporal variability of sediment trapping efficacies in valley-bottom wetlands.
Generally, increased volumes of rainfall enhance longitudinal flux connectivity of water in fluvial systems (i.e., greater efficiency of water transfer through catchments) through an increase in catchment surface runoff volumes, which can lead to higher magnitudes of sediment recycling and transport in fluvial systems (e.g., Seeger et al., 2004;Rodríguez-Blanco et al., 2010;Rovira, Ibáñez, & Martín-Vide, 2015).For example, a study on suspended sediment transport conducted in a small agriculture catchment of the semi-arid northeast Mallorca (Spain) by Estrany, Garcia, & Batalla (2009) revealed that increased volume and intensity of rainfall can lead to higher surface runoff, sediment production and suspended sediment loads.
Similarly, antecedent rainfall conditions essentially govern the rate of sediment availability and movement (e.g., Gray et al., 2015;Rodríguez-Blanco, Taboada-Castro, & Taboada-Castro, 2020).This temporal shift in sediment transport processes is substantiated by the apparent changes in particle size composition in the reforming reaches of the agriculturally-impacted wetland between the two years.In particular, the relative proportion of sand-sized particles and median grain size in 2020/2021 was higher than in the preceding water year while the fine-sediment fractions remained essentially consistent.
Coarser sediment of fine-to coarse-grained sand emanating from proximal sources (channel banks and bed) is preferentially transported in suspension during higher energy conditions (Fryirs & Brierley, 2012;Stone & Walling, 1997).As such, the higher total rainfall amounts (870 mm) and proportion of rainfall events greater than A key distinction between the agriculturally-and forestry-impacted systems is the relative catchment size and local wetland morphology.
The agriculturally-impacted wetland has a larger catchment area than the forestry-impacted system.This suggests that the catchment surface runoff and resulting flows in the larger agriculturally-impacted wetland system exhibit a greater transport capacity.The difference in flows is also evident in the contrasting channel dimensions, where the larger width and depth of the tributary and reforming channel reaches associated with agriculturally-impacted wetland is indicative of a higher discharge (e.g., Grenfell & Ellery, 2009).
In addition, the spatial variability and arrangement of reach-scale slope conditions may influence the degree of longitudinal coupling between geomorphic units within the fluvial system (Harvey, 2002;Wohl et al., 2017Wohl et al., , 2019)).Longitudinal fluvial coupling between individual wetland and channel reaches, which refer to the degree of upstream-downstream linkages between geomorphic units, typically determines whether sediment is trapped, or whether flows can systematically recycle and transfer deposited wetland sediments downstream (Fryirs, 2013;Fryirs et al., 2007).In the forestry-impacted wetland, progressively lower slope values of the tributary and reforming channel reach, and the proximal wetland reach relative to the small catchment area are likely to provide low potential fluvial energy conditions and promote longitudinal flux (dis)connectivity in terms of both flow and sediment dispersal processes.This is further enhanced by the longer ($100 cm), more robust wetland vegetation (grasses and sedges), relative to the shorter grasses ($50 cm) of agriculturally-impacted wetland (field observation during sampling campaigns).These conditions are likely to constrain the rate of hydraulically-induced sediment entrainment and conveyance; and consequently promote sediment accrual.As a result, sediment progressively accumulates in these reaches over several years until an inherent critical slope-and/or runoff-related threshold is reached, and valley deposits are evacuated (Ellery et al., 2016;Patton & Schumm, 1975).Contrarily, the local steepening in the gradient of the proximal wetland and reforming channel reaches of the agriculturally-impacted wetland provides conditions that are favourable for sediment transport and/or remobilisation, especially during high-magnitude rainfall events.
Cross-valley profiles of the agriculturally-impacted system illustrate a downstream reduction in wetland width and three initial reforming channels that coalesce into a single-thread stream channel in the lower section of the wetland.In this lower wetland reach, the lateral spread of floodwaters is limited while gullying processes such as gradually propagating head cuts and surface erosion are likely to be key sediment generation processes (Figure 9).Therefore, despite the relative importance of local rainfall conditions in controlling the temporal variability in sediment transport and trapping dynamics, spatial variability in these processes across the two wetland systems is essentially governed by the longitudinal arrangement of reach-scale slope gradients.
The context-specific response of contemporary sediment trapping and transport dynamics in these valley-bottom wetlands could likely be considered in terms of a hydro-geomorphic threshold.This threshold concept is similar to the critical slope threshold of long-term incision described by Patton & Schumm (1975) and Ellery et al. (2009Ellery et al. ( , 2016)), in which the likelihood of valley incision is determined by slope relative to catchment size, with the result that incision may be initiated in larger valley-bottom wetlands at slopes that are relatively low when compared to smaller wetlands with small catchments.As an example, for the agriculturally-impacted wetland, once a critical threshold of cumulative seasonal rainfall (>744 mm) is exceeded, and a specific proportion of the rainfall occurs during daily events >20 mm (>32%), longitudinal flux connectivity of surface water across the wetland units

| Contemporary total P retention efficacy and associated controls
The values of total P in suspended sediment observed were highly variable, yet were similar for the stream channels in the two wetland obtained for other humid river systems in agriculture-dominated catchments (e.g., Walling, Russell, & Webb, 2001).There may, however, be an influence on concentration levels of P and the quality of organic matter due to the annually aggregated sampling intervals.The extensive period of deployment might have led to altered pH, redox conditions and dissolved oxygen levels and as such potentially triggered changes in the bioavailability and solubility of P in accumulated sediments (Smith & Owens, 2014).
Based on the total P flux balance approach, the removal efficacy of total P was generally higher in the small wetland in a forested catchment compared to a larger wetland in a catchment dominated by agricultural activities.In 2019/2020, a large proportion ($83%) of the total P fluxes were effectively captured at the agriculturally-impacted wetland.Similarly, the forestry-impacted wetland attested to a net retention of total P (positive P balance) with an $90% reduction in fluxes (68.1 mg year À1 ) and was shown to represent an effective sink for total P.However, the percentage reduction in total P fluxes for the agriculturally-impacted wetland was negative (À4.7%) during the same period of 2020/2021, indicating net total P losses to downstream river catchments, despite the net retention of fine-grained suspended sediment (i.e., particle size fraction <1 mm).This net loss of total P coincided with a larger proportion of suspended sediment exported from the wetland and a higher magnitude of rainfall observed during 2020/2021.This mixed sink/source behaviour is in contrast with the findings of previous research in other alluvial wetland types (such as floodplains) in humid climates that predominantly reported a net retention of sediment-associated phosphorus with a range between 4 and 54% (Brunet & Astin, 2000;Kronvang et al., 2007;Noe & Hupp, 2009;van der Lee et al., 2004;Walling & Owens, 2003).
The removal dynamics of phosphorus in streams and wetlands are largely controlled by the textural and geochemical composition of fine-grained sediments and the interaction between phosphorus and cationic metal oxides, such as iron and aluminium (Johnston, 1991;Reddy et al., 1999).However, in this case, no collinearity was observed between phosphorus and decreasing particle size of suspended sediment.Instead, the total P concentrations were significantly positively correlated with both the organic fraction and Fe concentrations in the suspended sediment.This indicates that total P in these wetland systems is primarily stored in the organic fraction of suspended sediment-bound to Fe and not clay particles or Al (e.g., Kirk et al., 1998).
The phosphorus and organic matter associations in fluvial ecosystems have largely been attributed to the large specific surface areas of organic particles (Evans, Johnes, & Lawrence, 2004).This suggests that phosphorus removal and export dynamics are associated with the balance and dynamics of the organic matter component of suspended sediment.In addition, the high presence of Fe contained on organic matter particles as surface coatings through ligand exchange tends to increase the specific surface area and affinity of organic matter particles for phosphorus adsorption and storage in aquatic environments (Gerke & Hermann, 1992;Horowitz, 1991).Thus, the general linkages between organic contents, total P and Fe of suspended sediments highlight the importance of Fe in the adsorption of phosphorus on fine organic particles in controlling the transfer and storage dynamics of total P, while reactive Fe surfaces contribute to reducing OM bioavailability and degradation in anoxic environments (Lalonde et al., 2012;Repasch et al., 2021Repasch et al., , 2022)).This process could be especially important in dryland regions where oxidising conditions associated with the prevalence and prolonged periods of desiccation of most wetlands often promote OM mineralisation (Tooth & McCarthy, 2007).
The linkage of organic matter particles and phosphorus also implies that phosphorus exists primarily in organic forms, which likely emanated from a variety of biogenic organic sources, including wetland detrital material and dung deposits from livestock activities, and/or diffuse losses of agricultural fertiliser in the catchment (Dunne & Reddy, 2005).Since there are no localised point source phosphorus inputs, we suggest that erosion and contemporary recycling of fine-grained surficial deposits at the distal portion of the wetland (Figure 9), typically enriched in organic matter, are likely a key source of organic matter and associated phosphorus to the lower catchment.Fluvially-deposited organic matter associated with reactive Fe surfaces in wetlands are readily remobilised and transported during a range of flow events (Madej, 2005;Repasch et al., 2021) and thus are generally more dynamic than discrete organic matter particles (i.e., plant fragments) and flocculated fine-and coarse-grained sediments.
On the other hand, the general lack of collinearity associated with the relative proportions of fine-grained suspended sediment fractions (clay and silt) and total P is most likely due to the influence of particulate organic matter (Evans et al., 2004;Foster & Lees, 1999) and minor differences in particle size composition of the suspended sediments in the tributary and reforming channels (e.g., Crocker et al., 2021).Most of the suspended sediment in the tributary and reforming channels comprises silt-sized particles, with minor differences in clay contents.
As such, the discrepancy in total P removal dynamics and potential between wetlands is certainly associated with the transport and storage dynamics of phosphorus bound to OM and variation in land-use conditions.This association is a common occurrence and has previously been observed in numerous humid river systems.For example, Walling et al. (2001) demonstrated that phosphorus in humid UK rivers was primarily transported in association with the fine organic fraction of suspended sediment, with a limited control exerted by the particle size distribution of suspended sediment.
A comparison of the organic fraction of suspended sediment in the agriculturally-impacted wetland shows greatly elevated masses in the reforming channel than in the tributary stream, while the percentage of organic material in the forestry-impacted wetland was slightly lower in the reforming channel compared to the tributary stream.These contrasting organic matter balances between the two wetland systems likely indicate marked differences in the local storage, recycling and transport dynamics of fine organic particles in wetlands.Downstream reductions in slope in the tributary channel and proximal wetland reach result in a gradual decline in the capacity of the forestry-impacted wetland to transport fine-grained particles.In contrast, steeper slopes in the reforming channel of the agriculturallyimpacted wetland contributed to increased longitudinal flux connectivity of low-density organic material (e.g., Sutfin, 2015;Wohl et al., 2012).The high organic content in the reforming channel of the agriculturally-impacted wetland and significant overall correlation between organic content and total P is thus indicative of the influence of localised contemporary recycling processes on the downstream transport of phosphorus-rich organic matter particles in association with suspended sediment that was previously stored within the wetland and channel.This result also suggests that a large part of fine particulate organic materials that have accumulated on the wetland surface are consistently recycled and transported through preferential hydrological pathways by fluvial processes, limiting the potential for long-term retention.
In the agriculturally-impacted wetland, a large reduction in the concentrations of total P was observed in 2020/2021 in both the tributary and reforming channels.The longitudinal flux connectivity of surface water was also likely enhanced due to a higher magnitude of rainfall during this period, resulting in a higher stream transport capacity and the rapid remobilisation of a large range of sedimentsize fractions.The particle size distributions of suspended sediment reflect this increased competence, with larger median particle sizes and elevated proportions of sand-sized particles in suspension.A dilution of nutrient-rich organic and fine sediment particles by increased entrainment of mineral-rich coarse sand particles with low organic matter contents emanating from in-channel sources often occurs during high suspended sediment load and flow conditions (Gao et al., 2007;Madej, 2005;Zhang et al., 2009).Therefore, the reduced concentrations of total P in suspended sediment were mainly because of the selective routing of organic matter and the dilution behaviour of mineral-bound organic matter particles in fluvial systems.The extent of these hydrodynamic mixing and dilution processes is mediated by the magnitude of rainfall events and longitudinal flux connectivity of surface water as well as particle availability (Ballantine et al., 2008;Owens & Walling, 2002) (Figure 9).The general issue of spatial and temporal complexity in sediment trapping and total P removal efficiencies also poses a challenge in terms of generalisation and predictability of ecosystem services, especially when using rapid assessment tools (e.g., WET-Ecoservices tool; Kotze et al., 2009Kotze et al., , 2020)).However, there are opportunities to improve the integration of indicators related to process geomorphology within South African wetland protocols for appraisal of ecosystem service provision.For instance, a more accurate evaluation of ecosystem service provision requires greater emphasis on qualitative descriptions of wetland-specific process-form associations and system-specific geomorphic units that are indicative of temporal and spatial variability of process dynamics.Qualitative assessments of the research aimed to advance understanding of the potential suspended sediment and associated phosphorous trapping capacity of valley-bottom wetlands with differing morphometric characteristics, catchment sizes and land uses.The specific objectives were (1) to determine the magnitude and variability in contemporary mass fluxes and textural character of suspended sediment above and below two selected wetlands, (2) to examine the magnitude and patterns of the phosphorus content in suspended sediment and potential in-situ controls of sediment-associated phosphorus fluxes, (3) to examine the effectiveness of valley-bottom wetlands in regulating sediment and associated phosphorus fluxes and (4) to explore how rainfall, wetland geomorphology and catchment size affected the magnitude and efficacy of suspended sediment and attached phosphorus removal.The field sampling campaign of this investigation was temporally limited due to site constraints and the institution of travel restrictions during an extended national lockdown associated with the Covid pandemic.

F
I G U R E 1 Location of the (a) Tsitsa River catchment (Eastern Cape, South Africa) showing the (b) agriculturally-impacted (Fairbairn) and (c) forestry-impacted (Killarney) wetland catchments with terrestrial vegetation (after Mucina & Rutherford, 2006) and the main land-use units indicated.Insets indicate longitudinal profiles of the agriculturally-and forestry-impacted fluvial systems (constructed from SUDEM 5 m resolution).[Color figure can be viewed at wileyonlinelibrary.com] cross-sectional variation in suspended sediment fluxes.The samplers of the forestry-impacted wetland were placed mid-channel.
Downstream variation in longitudinal profiles and associated slope gradients for the thalweg of identified channel and mainstem wetland reaches along the (a) agriculturally-and (b) forestry-impacted systems.Reach divisions were based on step changes in slope values of sub-reaches.Note that the scales are different to improve visualisation due to the marked differences in wetland and channel dimensions.Refer to Figure1for the extent and locations of surveyed longitudinal profiles.[Color figure can be viewed at wileyonlinelibrary.com] depth of the tributary channel reaches of the agriculturally-impacted wetland declined from 15.9 m (±4 m) and 3.2 m (±0.5 m) at crosssections Fx 1 to 10 to less than 4 m (±2.6 m) and 1 m at crosssections Fx 11 to 15, respectively (Figure2aand Supporting Information S2a).Similarly, the valley floor of the wetland decreased from a maximum width of 353 m in the proximal wetland reach (Reach IV) to 178 m towards the distal reach (Reach V) (Figure3a).
The channel and wetland dimensions of the forestrywetland.Here, bankfull channel depths between cross-sections Kx 1 to Kx 4 in the forestry-impacted tributary reach gradually decreased downstream from 0.8 m to less than 0.5 m (Figure2b), and ranged from 3.25 to 0.64 m in bankfull width (see Supporting Information S2b).There was, however, a progressive downstream increase in bankfull-channel width and depth of the reforming reach (Kx 12 to Kx 14) and the channel dimensions were much larger than those of the tributary channel in the forestryimpacted system (Supporting Information S2b).Compared to the agriculturally-impacted wetland, the forestry-impacted wetland had a substantially narrower valley floor (<40 m), with a near-uniform valley F I G U R E 3 Cross-valley profiles of the (a) agriculturally-and (b) forestry-impacted wetlands, showing the down-valley changes in width.Note that the scales are different to improve visualisation due to the marked differences in valley dimensions.Refer to Supporting Information S1 for the extent and relative positions of cross-valley profiles.
in 2019/2020 to 0.08 kg year À1 in 2020/2021 (Figure 5a).Pairwise F I G U R E 4 Total daily rainfall record and cumulative rainfall during the sampling period in (a) 2019/2020 and (b) 2020/2021 measured at the PG Bison/Mooi rainfall station (Data from The Tsitsa Project).(c) Relative proportions of the cumulative rainfall.(d) Return interval of annual total daily rainfall derived from the data for Maclear obtained from the South African Weather Bureau.The boxes and vertical dashed lines represent the timing of suspended sediment sampling events in 2019/2020 (S1) and 2020/2021 (S2).[Color figure can be viewed at wileyonlinelibrary.com]comparisons showed that the differences in mean suspended sediment mass fluxes between the tributary and reforming channels were not statistically significant ( p > 0.05) during the study period.This was likely due to the high temporal and cross-channel variability in suspended sediment as well as the low sampling size and frequency.

F
I G U R E 5 Mass fluxes of suspended sediment (SS) in the tributary and reforming channel reaches of the (a) agriculturally-impacted during the 2019/2020 and 2020/2021 sampling periods and (b) forestry-impacted wetlands during the 2020/2021 sampling period.Bars in (a) indicate mean values and whiskers depict ranges (n = 3), while bars in (b) indicate absolute values (n = 1).[Color figure can be viewed at wileyonlinelibrary.com]

|
Variability in suspended sediment mass fluxes and extent of trapping efficacy The findings of this study provide a baseline understanding of the contemporary rates and magnitude of suspended sediment and associated total P fluxes and trapping efficacy in discontinuous, clastic-dominated valley-bottom wetlands in small fluvial systems of a semi-arid environment.Overall, a mean relative suspended sediment mass flux rate of 0.01 to 0.11 kg year À1 was observed in the tributary and reforming channels of the wetlands studied for the monitoring period (2019/2020 to 2020/2021).Over the two years, between 24% and 91% of the fluvially transported suspended sediment was sequestered by the wetlands.While there is a lack of comparative quantitative estimates of sediment trapping efficiency in valley-bottom wetlands, the proportional range of the present study is comparable obtained for the valley-bottom wetlands herein.These large variations in sediment trapping efficiencies, especially in naturally functioning systems, have often been attributed to differences in wetland geomorphology, flood duration, sediment characteristics, inundation frequency and flow patterns across the wetland surface (e.g.,Noe & Hupp, 2009;Van der Lee, Olde Venterink, & Asselman, 2004;Walling & He, 1998).T A B L E 2 Differences in mass fluxes and median particle size of suspended sediment (SS) in tributary and reforming channels and estimates of trapping efficacy of SS in the agriculturally-and forestry-impacted valley-bottom wetland during the sampling periods (2020 to 2021).
20 mm (43%) recorded in this water year might have contributed to the greater localised fluvial activity and downstream sediment transport.In comparison, the lower volume of cumulative rainfall (744 mm) and proportion of events greater than 20 mm (32.2%) during the preceding year may have limited the longitudinal flux connectivity of water across the wetland unit and downstream sediment redistribution.The average suspended sediment mass flux measured at the outlet of the agriculturally-impacted system during 2020/2021 was about 2 times higher than that recorded at the outlet of forestry-impacted wetland during the same period.Given the relative proximity of the two sub-catchments and similarity in terms of geological provenance and rainfall conditions (magnitude and intensity), large-scale differences in the local catchment and wetland geomorphological characteristics potentially contribute to the contrasting sediment trapping and dispersal dynamics within these valley-bottom wetland systems.

F
I G U R E 6 Particle size composition of suspended sediment in the tributary and reforming channel reaches of the agriculturallyimpacted wetland during (a) 2019/2020 and (b) 2020/2021 sampling periods and (c) the forestry-impacted wetland during the 2020/2021 sampling period.Lines in (a) and (b) indicate mean values (n = 3), while lines in (c) indicate absolute values (n = 1).Green lines depict the tributary channel while red lines depict reforming channels.[Color figure can be viewed at wileyonlinelibrary.com]

F
I G U R E 7 Comparisons of total P concentrations in suspended sediment in the tributary and reforming channel reaches of (a) agriculturallyimpacted during 2019/2020 and 2020/2021 sampling periods and (b) forestry-impacted wetlands during 2020/2021 sampling period.Bars in (a) represent mean values and whiskers depict ranges (n = 3), while bars in (b) indicate absolute values (n = 1).[Color figure can be viewed at wileyonlinelibrary.com]T A B L E 3 Difference in selected geochemical attributes of suspended sediment (SS) in tributary and reforming channels and estimates of total P removal efficacy in the agriculturally-and forestry-impacted valley-bottom wetlands during the sampling periods (2019 to 2021).
is enhanced.As a result, during the 2019/2020 season, the cumulative seasonal rainfall amount of 744 mm with fewer events exceeding 20 mm ($32.2%), has likely contributed to enhanced sediment trapping.In contrast, during the 2020/2021 season, sediment transport was enhanced, and sediment trapping effectiveness was reduced, and reach-scale slope conditions favoured longitudinal coupling and flux connectivity of both water and suspended sediment across wetland units.This longitudinal flux connectivity was potentially not established within the forestry-impacted wetland, which likely suggests that smaller wetlands with steep longitudinal slopes in relatively smaller catchments and as such lower runoff have a higher hydro-geomorphic threshold for sediment recycling and dispersal.However, given the temporal constraints of the dataset, the results merely provide preliminary evidence for the alluded hydrogeomorphic threshold behaviour of these systems.Additional studies at a higher temporal resolution that take into account different peaking rainfall intensities are therefore required to elucidate broader temporal and spatial patterns and substantiate the potential threshold phenomenon.The question also remains as to how variability in other biophysical factors such as vegetation type and cover, hydroperiod, wetland morphometry and sediment composition, influence the sediment effectiveness of these systems.

F
I G U R E 8 Relationships between Total P and (a) organic content, (b) iron (Fe), (c) calcium (Ca), and (d) aluminium (Al) concentrations for suspended sediments along the tributary and reforming channels of the two valley-bottom wetlands.systems, despite the contrasting catchment land uses.These sediment total P concentration values (124 to 546 mg kg À1 ) are in the same order of magnitude as reported for suspended sediment in humid rivers within agricultural (330 to 540 mg kg À1 ) and forested (810 mg kg À1 ) catchments (e.g., Pavanelli & Selli, 2013; Stenfert Kroese et al., 2021) but below ranges (1490 to 4300 mg kg À1 )

F
I G U R E 9 Photographs depicting ongoing (a) head cut propagation (upstream view) and (b and c) localised channel bank erosion (downstream view) upstream of the suspended sediment sampling location in the main reforming channel of the agriculturally-impacted wetland.[Color figure can be viewed at wileyonlinelibrary.com]

5. 3 |
Implications for rapid assessment and monitoring strategies in valley-bottom wetlandsValley-bottom wetlands in dryland regions such as southern Africa are prominent features on drainage lines and form a critical part of resilient fluvial landscapes.The widespread deterioration and loss of these systems have prompted large-scale investment in their conservation and restoration(Dini & Bahadur, 2018).However, current rehabilitation or recovery protocols for valley-bottom wetlands in South Africa are mainly concerned with surface hydrological and ecological principles and do not explicitly engender process-based geomorphological perspectives (e.g., vertical and longitudinal fluxes of sediment)(Grenfell et al., 2020;McCarthy et al., 2010;Ralph, Hesse, & Kobayashi, 2016).The spatial and temporal variability of contemporary suspended sediment mass fluxes, availability and composition obtained in these study systems can also have profound implications for future management and restoration activities of wetlands as well as water resource management planning.Although sediment input generally exceeds sediment export, the magnitude and dynamics of sediment trapping and transfer within these systems can change over spatial and temporal scales.These differential rates and system-specific dynamics of sediment trapping or export are likely underpinned by a potential hydrogeomorphic threshold condition, where abrupt changes are likely locally induced by the interaction of rainfall-and reach-scale slope-defined thresholds.The rainfall magnitude that activated the reforming channel and enhanced the export of sediment from the agriculturally-impacted wetland was 65.2 mm in 24 h with a maximum intensity of 19.2 mm hr À1 .The low return period (2.5 years) of rainfall events of this magnitude suggests that sediment transfer and exchange occur in relatively frequent pulses.As such, it appears that valley-bottom wetland systems contain a 'self-limiting' criticality with recurring sequences of sediment trapping and recycling over short timescales.
geomorphic conditions and features specific to valley-bottom wetlands that could provide insight into spatial and temporal variability of sediment transfer processes could include (1) characterising longitudinal and lateral flux connectivity patterns of both water and sediment along wetland reaches based on longitudinal slope and valley morphology and (2) identifying and analysing diagnostic types, assemblage and pattern of geomorphic units in contextual terms of process-form associations (i.e., processes that create a particular unit) using both desktop and field-based investigations.For instance, the relative extent and character of key depositional (floodouts) and erosional (head cuts and reforming channel) features associated with valleybottom wetlands could provide process-based insights into the wetland-scale patterns of sediment dispersal as well as the extent to which this process have adjusted over space and time (e.g.,Fryirs & Brierley, 2021).6 | CONCLUSIONThe current study generated baseline knowledge and provides insights into the efficacy of two hydrologically variable valley-bottom wetlands in regulating downstream fluxes of suspended sediment and associated total P concentrations.A contemporary trapping efficacy for these valley-bottom wetlands of between 24% and 92% was estimated between 2019 and 2021.This indicates that these systems are important features for attenuating sediment fluxes in fluvial networks and temporally alleviating the siltation of downstream water resources.However, the estimated suspended sediment trapping efficacy and downstream suspended sediment export in these systems vary temporally and spatially in response to changes in seasonal rainfall regimes and wetland geomorphology.The magnitude and proportion of total P retention varied across wetlands and temporal scales, underscoring the inherent complexity of phosphorus removal dynamics.This complexity and contrasting patterns of sediment-associated total P removal fundamentally reflect the variability in organic contents associated with suspended sediment transport and storage regimes in these wetlands, which may result from the dynamic interaction between localised hydrometeorological conditions and wetland geomorphic character.However, the extent to which these results can be generalised to other wetland types and valley-bottom wetlands in a different hydrological and geomorphic context remains uncertain.There is a scope to further develop detailed estimates of both the relative rates and quantity of sediment and phosphorus storage over different temporal and spatial scales to confirm and/or refine the proposed hydrogeomorphic threshold governing sediment trapping efficiency in valley-bottom wetlands.Moreover, given the relative importance of local hydroclimatic regimes in sediment trapping and phosphorus flux dynamics, changes in rainfall and flow magnitudes and patterns associated with climate change and anthropogenic activities can have implications for sediment trapping and dispersal processes.The consequent changes in the fluvial hydrological regimes have the potential to result in profound temporal shifts in sediment cascades with significant variation in several ecosystem functions within wetlands in dryland fluvial systems.This is an especially great concern for water-stressed dryland systems due to general predictions of current climate models.As such, the ongoing management plans for these valley-bottom systems need to consider dynamism in flow and sediment fluxes in light of changes in hydroclimatic conditions and land-use pressures, for ensuing future resilience of ecosystem services provision.