Evaluating composition and efficacy of vegetation filtration treatments to ameliorate adverse impacts from agricultural runoff, conserve soil and protect stream water quality.

Dr. Orah Moshe: Ministry of Agriculture, Department of Soil Conservation, Soil Erosion Research Station
Ms. Tal Ratner: Kishon Basin Drainage Authority
Dr. Nativ Dudai: Agricultural Research Organization, Volcanii Institute, Newe Ya’ar

Submitted as report to Nekudat Hen

The goals of this project target advancement in a best management practice that conserves agricultural soils and protects water quality in Israeli streams from the deposition of agricultural sediment and chemicals.

Abstract

We compared the role of species composition in planted vegetation buffer strips (VBS) to filter agricultural runoff, capture agricultural sediments and nutrients, conserve soils and protect stream water quality. We established VBS in an agricultural field bordering Nachal stream, which drains into the Kishon River in northern Israel. Twenty four plots (10 m  7 m) were created near the base of a sloped agricultural field. Each plot was assigned to one of six treatments in a randomized complete block design: 1) barley (Hordeum vulgare); 2) wheat (Triticum Aestivum ); 3) oats (Avena sativa) combined with vetch (Vicia Atropurpurae); 4) yablit cross one (Cynodon dactylon); 5) vetiver (Chrysopogon zizanioides); and 6) a bare soil control.

We measured aboveground plant height, cover and biomass. Vetiver had highest plant height and biomass and Yablit had the lowest, as compared to other seeded species. Cereal treatments did not significantly differ between barley, wheat and oat, however, oat had the highest growth rate. Yablit and oat/vetch treatments maintained the highest cover of the seeded species, was least invaded with weeds, and had the lowest amount of exposed ground. Wheat the lowest cover with the highest weed invasion.

We measured soil loss from the primary gully along the farm road, estimating net soil loss at 84-107 m³.

We designed, constructed, installed and tested passive runoff collector units to collect sheet runoff samples during rain storm events. We successfully captured surface runoff samples during four storm events and measured total suspended sediment and NPK nutrients in three slope locations, corresponding to buffer widths: plot influent (0m), mid plot (5m) and plot effluent (10m). We demonstrate a reduction in TSS concentration of between 11-17% after 5 m of VBS, showing that even a small VBS (5m) may result in environmental benefit on these marginally productive lands.

The oat/ vetch treatment performed best in terms of providing cover, protecting the soils and efficiently capturing sediment and is well suited for VBS application on farms. Optimizing VBS placement is needed on a site specific basis to maximize filtering capacity.

Additional studies are needed to investigate how sediment capture efficiency is affected by different buffer widths in a variety of settings, and testing native perennial plants that are appropriate for riparian VBS restoration. The movement of the stream bed channel supports the need to plan riparian zones to enable this natural process, which produces additional flood reduction benefits. We suggest that each application will need to be evaluated for site specific conditions to maximize ecosystem benefits.

Background

The susceptibility of soil under intensive tillage, especially on sloped agricultural lands, renders the land prone to erosion under natural rainfall events (1). Globally, high rates of soil erosion from agricultural fields results in arable land degradation and risk of agricultural lands coming out of production (2,3). Agricultural non- point source chemicals (NPS) include fertilizers (nitrogen (N), phosphorus (P) and potassium (K)), and pesticides (pesticides, herbicides and fungicides), which are transported during rain events, causing off-site environment pollution in agricultural watersheds (4–6). Nitrate-N is highly soluble and is rapidly transported with water whereas, ammonium-N, organic N and P are largely insoluble, adhering to soil particles, transported with sediments, becoming particulate-bound pollutants, resulting in decreased soil productivity in addition to degradation of water quality in receiving waters (7).

Agricultural pollution is the leading source of water quality contamination in surveyed rivers, streams and ground water (8). Today, leaching and runoff from agricultural lands contribute pollution to natural waters to a much greater degree than it did a few decades ago (9). Approximately 90% of agricultural land worldwide suffers moderate to severe erosion impact (2,3) with sediment loss exacerbated by high concentrations of fertilizer, herbicide, fungicide and pesticide-laced runoff entering streams negatively impacting ecological habitats, environmental indices (10) and public health (11).

Intensive agricultural activities combined with a lack of site-specific best management practices explains accelerated soil erosion and increasing water degradation (12,13). Wetlands in cultivated watersheds may contain 8.5 times more sediment than wetlands surrounded by grazed grasslands (14). High concentrations of suspended sediments are visibly evident during storm events in most streams in Israel. Soil erosion has become a priority global concern, contributing to the decline of agricultural productivity and decreased food supply, impairing water quality, (15) and degrading overall ecosystem services (16–18). Increasingly, decision makers are recognizing the goods and services that are provided by relatively intact ecosystems, such as flood control, maintenance of soil fertility, minimization of soil erosion, and pest control (19).

Bare soils are vulnerable to soil erosion, therefore, in a Mediterranean climate, the rate of establishment is a critical factor in species selection. Optimal species must succeed in providing sufficient ground cover to protect soils from rain-driven erosion early in the rainy season. Other important factors in the species composition selection, includes rooting depth, plant morphology, life cycle, and habitat value, both for wildlife and for agricultural predators and pests.

VBS studies conducted in many countries have demonstrated that VBS at the land-water interface have potential to mitigate soil erosion, reduce agricultural NPS stream deposition and thus protect water quality (20–24). In addition, they have the potential to provide other important ecological services, such as native plant restoration, wildlife habitat, and recreational values. Buffer effectiveness is dependent on several factors, including vegetation structure, buffer width, attributes of the surrounding watershed (i.e., area, vegetative cover, slope and topography, soil type and structure, soil moisture, amount of herbicides and pesticides applied), and intensity and duration of rain events (14). The effectiveness of vegetative filter strips (VBS) for reducing herbicide transport has not been well documented for runoff prone soils (25).

Commonly, agricultural activities extends to the streambank, resulting in a lack of existing riparian vegetation in many Israeli stream ecosystems further exacerbates adverse water quality impacts. Recent recognition of this problem in Israel has resulted in the development of collaborative interagency efforts to identify and target some streams for strategic expansion of the corridor, focused on riparian vegetation buffer restoration (26). Specific scientific data is needed to inform policy and provide a basis for riparian buffer implementation, as well as to facilitate adoption of on-farm best management practices that will conserve soil resources and protect water quality in agricultural watersheds. Due to the seasonal climatic regime, the Mediterranean climate in Israel may have important implications for VBS function (27).

Vegetation has been used as a natural bioengineering tool to control erosion and stabilize slopes for centuries. Its popularity has increased in recent decades due partly to more information available about vegetation and due to the cost-effectiveness and environment-friendliness of this “soft” engineering approach (28,29). The role of species composition in fulfilling the potential range of ecosystem services is an important area to investigate in order to optimize benefits. Species composition is an important factor in VBS efficacy, with different species contributing different ecosystem services, such as sediment trapping, nutrient uptake, soil stabilization, groundwater remediation, and ecological habitat.

The width of the VBS is important in determining the reduction efficiency of both soil and NPS pollutants. Castelle et al. (30) conclude that Riparian buffers from 3-200 m in length are effective in protecting aquatic resources from adjacent agricultural land use practices. Increasing the width of an herbaceous filter from 4.3 to 8.6 m reduced the amount of runoff and sediments passing through a VBS, but a wider buffer was less effective during very intense rainfall, probably due to increased runoff speed (31). The composition of a buffer and its width is site-specific and should be based on land use intensity, aquatic resource sensitivity, and integration with other best management measures (32).

We investigated the role of species composition in filtering agricultural runoff by comparing the sediment capture efficacy of VBS treatments consisting of different plant species. We also evaluated plant growth and micro-topographic changes across the lower part of the slope, to determine the role of species composition in both stabilizing soils and providing cover early in the rainy season.
This study was conducted in a natural catchment under natural rainfall events, as compared to most studies worldwide where simulated rainfall is used. We hope to advance sustainable agricultural methods, protect soil and water resources and increase participation of farmers willing to adopt this method. Results from this study will help to inform policy makers and improve the scientific basis of conservation agriculture with agro-ecological benefits.

Knowledge Gaps

While VBS have demonstrated success in capturing eroded agricultural sediments and filtering agricultural NPS chemicals, several important questions remain, including role of species composition and the effective width of buffers necessary in a range of natural field conditions (24). Using native grasses in VBS potentially provides an opportunity to simultaneously reduce NPS pollution while restoring native biodiversity(33) . Individual plant species differ in their ability to ameliorate NPS pollutants and control erosion (34,35). In Israel, the efficacy of specific native grasses in controlling erosion has not been tested as well as a lack of species- specific information in Israel regarding ecosystem services that plants may provide. Data are needed with detailed planting density for specific conditions to define target plants species that may be efficient at both erosion control, filtering agricultural runoff and native grass restoration. The lack of these data currently limits the ability to make recommendations and disseminate this information for implementation as a best management practice.

We tested whether the efficiency of VBS receiving agricultural runoff varied depending on the species of plants comprising the VBS. Direct measures in “real-world” systems and field validations of buffer-effectiveness models are crucial next steps in evaluating how grass buffers will impact the abiotic and biotic variables attributes that characterize small, isolated wetlands (14). We evaluated the ability of different VBS treatments to trap sediments and reduce sheet erosion on steep slopes in an agricultural catchment in a Mediterranean climate under natural rainfall conditions, comparing perennial grasses and annual grasses. This is the first study in Israel evaluating VBS.

Methods:

Study Site
This study was conducted in the Newe Yaar Agricultural Research Center, near Ramat Yishai, Israel (Fig 1).

 

 

 

 

 

 

 

 

 

 

Figure 1: Site location map showing boundaries of block treatments

We surveyed the land directly above plots, with slopes ranging from 1 to 9 % (Fig 2). Soils within the plots are classified as accumulative hydromorphic dark brown Grammasole and alluvial brown clay , according to a new soil survey conducted in 2018. We wanted to test buffer areas implementation on marginal lands adjacent to streams. We established buffer plots as close to the stream as we thought practical. A total plot area of 1,680 sqm were taken out of agricultural production at the base of the slope and used for this study (Fig 2).

Figure 2: Study location map showing slope contour lines and plot locations

We conducted this study between May 2018 and June, 2019. The local Mediterranean climate is characterized by an extended dry season (May-September) and episodic rain storm events primarily between November and April. We obtained rainfall data from the meteorological station onsite, which measures rainfall frequency, intensity and accumulated rainfall depth. The study area is adjacent to Nahalal, a small perennial stream draining into the Kishon River, surrounded by agricultural activities and Neve Yaar Agricultural Research Facilities. The total area of the watershed is 41,706,640 sqm with an estimated 23,184,218 sqm in agricultural production, approximately 56% of the watershed. Figure 4 shows land use activities in the basin.

      Figure 4: Land Use in Nahalal Watershed

Plot Establishment

Soils were disked to prepare the land for planting in early May, 2018. We surveyed the sloped land using RTK (±0.6cm) soil elevation measurements on May 17, 2018 in order to select exact plot locations to maximize receiving sheet flow runoff and minimize cross plot flows. Twenty four plots, 7m by 10m, were created using a randomized split plot block design to test six different manipulations of plant species, with four replicates of each treatment. The plots were located down-slope from a field planted in wheat, receiving surface and subsurface water flow from the agricultural area. As of November, 2018, the field above the plot was bare soil until wheat established early December. This crop cycle is different than last year, when corn was the target crop, planted in rows, and the slopes were maintained bare for the beginning winter months. The amount of bare soil up gradient from the VBS is an important factors affecting the amount of soil erosion occurring on the farm field.

During a site visit on November 4, 2018, we observed construction of a drainage channel along the farm road, created in the same location of last year’s gully. The experimental plots are located outside of the influence of this drainage channel. However, we observed substantial soil loss occurring in this location and we investigated the gully volume over time. There is a small drainage swale between blocks 2 and 3 that concentrates runoff that we cannot replicate, therefore this area was excluded from the study. We seeded oat/vetch combination to reduce soil erosion.

Experimental Design

1. Establishment of vegetation filter strips (VBS) and detailed species description

Six vegetation treatments were established to compare the role of species composition in the capture efficiency of sediments and agricultural nutrients. Twenty four plots (10 m  7 m) were created along the topographic line near the base of the hillslope (Figure 5). Each plot was assigned to one of six treatments in a randomized complete block design: 1) barley (Hordeum vulgare); 2) wheat (Triticum Aestivum ); 3) oats (Avena sativa) combined with vetch (Vicia Atropurpurae); 4) yablit cross one (Cynodon dactylon); 5) vetiver (Chrysopogon zizanioides); and 6) a bare soil control that received herbicide two times during the season. However, this frequency was insufficient to keep the soil bare and some non-target species established. These annual plants are commonly used by farmers in Israel. We compare annual and perennial, nonaggressive (sterile) species (Vetiver and Yablit cross one) that potentially offer multiple ecosystem services.

Figure 5: Vegetation treatments in randomized block design

Plant Species

Vetiver (Chrysopogon zizanioides) is a perennial bunchgrass of the Poaceae family, native to India. Vetiver’s root system is finely structured and very strong. Unlike most grasses, which form horizontally, spreading mat-like root systems, Vetiver’s roots grow downward, potentially growing 2 to 4 meters in depth within the first year, surviving in both flooding and drought conditions (36). Vetiver has been used effectively to protect soil against sheet erosion (37), and is currently used in Israeli fields with success (ie. Harduf). In areas with high sediment deposition, new roots can grow out of buried nodes, propagating itself by small offsets instead of underground stolons. This genotype is sterile, noninvasive and easily controlled. We investigated this species because of the range of potential benefits, including sediment capture due to its dense basal biomass, soil stabilization due to its root structure and groundwater remediation due to its rooting depth, as well as the plant having economic value (35,36,38).

Figure 6: Vetiver seedlings separated at roots, hand planted in plots

Vetiver is dormant during the summer and winter. In preliminary trials in 2017-2018, all the hand planted seedlings suffered mortality, either due to insufficient water or rooting depth, or lack of sufficient rooting biomass. Therefore, we installed irrigation lines to facilitate establishment and enable plant development to occur prior to the rainy season. Dr. Nativ Dudai, who introduced the plant to Israel, supervised the hand planting effort in each of four replicates plots on June 11, 2018 to ensure that there was sufficient root biomass and planting occurred at the correct planting depth and plant spacing. The plots were surveyed on July 22, 2018 for survival and dead seedlings were replaced in early August 2018. By November 2018, these plants had successfully established within the plots (Fig 7).

Figure 7: Vetiver established, October 2018

Yablit Cross one, Bermuda Grass (ברמודה X1 יבלית מכלוא)
Joseph Guggenheim (Shaham, Ministry of Agriculture), introduced this species to Israel in 1978 from Georgia, where it underwent agricultural experiments by Dr. Tal Kipnis, Volcani Institute. Originally it was cultivated as a grazing plant. They quickly discovered it had high durability properties, both in the area of “soil stabilization” and in the field of water conservation. The cross one variety is sterile, preventing the species from becoming invasive in the field, like the native form. Kibbutz Gaash grows the plant material and delivered it to Neve Yaar on November 5, 2018 in a large roll like a carpet. Yablit was planted in plots on November 6, 2018, and irrigated weekly to ensure establishment (Figure 8).

Figure 8: Yablit treatment on November 18, 2018, with Vetiver plot in background

Annual cereals

Annual grasses commonly used for erosion control by farmers include barley, wheat, and oats. Their growing strategy does not require investing in substantial root development, as compared to perennial grasses, and therefore can provide above ground cover early in the rainy season, however, their rooting structure is shallow, with thin roots, less effective is maintain soil cohesion. In addition farmers are familiar with these plants, removing any potential obstacle from introducing them on their farms in VBS. We compared three common annual cereal treatments: Barley, wheat, and a combination of oats and vetch (Fig 9). The nitrogen fixing capacity of the vetch may provide additional ecosystem services within VBS and famers are familiar with seeding this as a cover crop. In addition, it complements the oat, providing dense ground cover, while the oat grows more erect with less ground cover.

Figure 9: Cereal treatments (a) wheat; (b) oat/ vetch; (c) barley

Plots were planted using an air seed drill on November 8, 2018, with 3 cm seed spacing (Fig 10). Seeding densities were as follows: barley (8 kg /dunam), wheat (14 kg /dunam), oat (8 kg /dunam), vetch (5 kg/dunam). We tested soil NPK prior to seeding, only adding nitrogen pellets (46%), triggered to be released at 30 mm of rain. Fertilizer pellets were added with the drill seeder to increase plant establishment success. A light rain had occurred prior to seeding, wetting the soils and facilitating establishment.

Figure 10: Air seed driller planting in surveyed plots

We measured species composition, percent cover, and plant height in 25 x 25 cm quadrats randomly located in the top 5 m and lower 5 m of each plot three times during the growing season: January 2, February 25, and April 7, 2019. Plants were separated into two groups: species seeded in a plot (target) and non-seeded species (other) (including species that were seeded in another treatment and spread). Cover was estimated based on presence within a 16 grid square and calculated to the nearest percent (Dethier et al. 1993). In the April sampling event, we conducted destructive sampling and measured above-ground biomass. Above-ground biomass was clipped, dried in an oven at 60°C, and weighed to the nearest 0.01 g.

2. Sheet Erosion (Micro-topographic Baseline)

Soil sheet erosion and deposition were measured using two complementary methods: erosion pins and profile analysis. Erosion pins (39,40) were used initially to measure changes in surface micro-topography. Erosion pins consisted of 6 mm-diameter black plastic piping, spaced 1.5 m apart in horizontal transects, with 2 pins per transect, and 4 transects per plot (Fig 11). Pins were inserted into the soil 10-20 cm deep and locations were surveyed at time of installation using two methods. Point data collection involved hand measuring existing soil elevation by defining the baseline height above ground (from top of pin to surface of soil). A metal disc (0.25mm) was placed over each pin, sitting on the surface of the soil to enable analysis of both depositional and erosional processes, in addition to determining net change. When deposition occurred, we measured the amount of material above the metal disc. We measured pin height using a handheld meter tape on November 18, 2018 and resurveyed after the rainy season on April 7 and 14, 2019 (Fig 11). Erosion pin transects were placed at distances of 0 m, 1 m, 5 m, and 10 m from the top of the slope (Fig 12).

Figure 11: Erosion pin transects and point monitoring method

Figure 12: Erosion Pin and Runoff Collector Unit (RCU) locations in plots

Although a commonly used method, erosion pins can cause a direct effect, sometimes resulting in preferential deposition or erosion around the pin (21). Therefore, we used a second method to calculate the soil profile in the area between the pins, where no interference to surface flow occurs. Using RTK, we continuously measured soil elevation along the 1.5 m line between each set of pins comprising transects. We improved the resolution of the measurement (± 1-3 mm) by holding each position for 10 seconds. Typical RTK data for streambank work is three seconds to triangulate between satellites and base stations. Cross sections were generated using two vectors: horizontal distance from the starting point (pin A-pin B) and soil elevations representing the vertical height of the cross section. We analyzed RTK data using Matlab to calculate the area for each pin transect, representing a cross section of the surfaces micro topography. RTK pin surveys were conducted November 18- 19, 2018 and April 19-20, 2019. We compared the change in area between the two measurements (before and after rainy season) at each slope location. The number of measurements and length of each plot was not always consistent for comparison. Therefore, we standardized the length of each pin transect by aligning the measurements and deleting points that exceeded the shorter segment by a maximum distance of 9 cm. Statistical analysis was conducted on the soil elevation differential between the two measurements and analyzed for both slope location and treatment effects.

Gully Erosion

We measured the morphology of the constructed drainage channel parallel to the farm road in November 2018 by collecting 20 cross-section analyses using RTK measurements. Transect locations are shown in Figure 13.

Figure 13: Cross section transect locations (green line) collected on drainage gully

We conducted RTK cross section elevation surveys before the onset of the rainy season in October 2018 and again at the end of the rainy season in March, 2019 (Fig 14).

Figure 14: Photo of drainage channel gully before and after rainy season

3. Sheet Runoff Sample Collection

We designed innovative surface runoff collector units (RCU) to collect sheet flow runoff water samples during storm events (Fig 15). We constructed the units from galvanized steel, produced by Offer Yefet, Holon (Fig 16). RCU consist of a 450 ml plastic sample bottle placed inside the RCU steel sleeve, buried at ground level. RCU have a retractable arm with a 15 cm metal flume that lays on the up-gradient soil surface to maximize contributing area for sheet runoff water collection. Lifting the retractable arm opens the RCU to enable removal and replacement of the full sample bottle without disturbing the soil (Fig 16). A roof protects the unit from rainfall entry and dilution. Water flows into the RCU flume channeled into a hole in the metal cover, filling the sample bottle with runoff. Inside the bottle, a ping-pong ball floats as the sample bottle fills, sealing the entry hole when the sample bottle is full, preventing dilution with less concentrated runoff after continuous rain.

Figure 15: Surface Runoff Collector Units (RCU) Design

We installed RCUs in the field between November 26 and December11, 2019 in three locations within each plot, representing different lengths of buffer: 0m (plot influent), 5m (mid plot) and 10m (plot discharge towards stream), in order to investigate capture efficiency at 5 m compared to 10 m (Figure 12).

Figure 16: Constructed Surface Runoff Collector Units (RCU)

The RCU enable high integrity sample collection during episodic rain events, which is a difficult task due to the short term nature of storms and size of study area. Full sample bottles were collected after storms and replaced with empty bottles.

4. Water Quality Analysis of Sheet Flow Runoff : Quantification of Suspended Sediment Concentration and NPK (Three slope positions)

We collected RCU sample bottles as soon as possible after storm events and delivered them to the Neve Yaar laboratory for analysis. To measure total suspended sediment concentration, samples underwent vacuum filtration, separating sediments from the water, enabling water chemical analysis on the remaining sample. Vacuum filter paper was dried in an oven and weighed to the nearest 0.1g to measure total suspended sediment concentration in the 450 ml runoff sample. The laboratory returned the filter papers and we are currently developing methodology to analyze the particle size distribution of these dried sediments. Water samples were analyzed for N, P, and K in defined states using certified laboratory protocols. If samples were unable to be collected the following day, they were not analyzed for NPK, due to limited holding time for analysis because microbial transformation affects nutrient concentration results. Suspended sediment concentration is unaffected by holding time.

5. Stream Geomorphic Baseline

One goal of this study was to investigate if Drainage Authority annual stream maintenance costs would be reduced with large-scale implementation of vegetation buffers, based on effective sediment trapping, which would result in decreased sediment deposition into the stream. We proposed analyzing historical data of dredging schedule, volumes and frequency. However these data do not exist. Therefore we are unable at this point to evaluate the impact on dredging. We propose an indirect measure for future analyses, comparing stream morphology data using detailed cross section analyses within Nahalal stream before and after implementation of VBS. Measurable differences in stream sediment deposition is limited due to the small scale of this pilot plot experiment (four replicate blocks along the stream), 180 linear m only on the right bank of the stream. In addition, the complexity of sediment transport in streams makes it challenging to correlate small land use changes to specific sections of stream, as high volumes of sediment are transported downstream during storm events. However, we succeeded in establishing a meaningful baseline that will enable us to evaluate stream geomorphology changes over the short and long term. Our survey included repeating baseline measurements collected in 2016, in addition to establishing new permanent transect locations every 50 m, beginning down-gradient from our plots and extending upstream 1 km to capture the area beneath the model farm, enabling future assessment of intervention activities.

Statistical analytic methods

We defined this study overall as a split plot design in a randomized block. The main effect plots are the treatments, which are set out in four blocks. The subplot effect is the slope location. Erosion Pin and RCU data were collected over time in a repeated measures design. Plant sampling locations were randomized. Our model included interactions between treatment and location, treatment and time, and location and time, and a three-way interaction. Six different plant compositions comprised each plot, representing treatment effects, with four replicates. RCU were installed in three slope locations, (influent 0m), middle (5m) and effluent (10M), for each plot. Erosion pins were in the same slope locations, with an extra transect at 1m (from top of buffer plot). We used JMP (14) statistical software (SAS) for all statistical analyses.

Results

Figure 17: Monthly Rainfall at Neve Yaar Meteorological Station

Monthly rainfall (Fig 17) and individual 10 minute storm intensity (Fig 18) for 2018-2019 are presented below. Total annual rainfall for this year was 864 mm, which is above the annual average of 600 mm, based on past 10 years data. Maximum ten- minute storm intensity was 60 mm/hr, which occurred early in the rainy season, before the plants had fully established. Maximum storm intensity on January 8 was 34.8 mm/10 minutes, which generated runoff and enabled the first sample collection event. Maximum hourly storm intensity for the winter was 40 mm/hr.

Figure 18: 2018-2019 Rainfall Intensity (mm/hour)

1. Establishment of vegetation filter strips (VBS)

Plant Height

Plant height significantly differed between treatments (p<.0001), and sampling date (p<.0001). Vetiver had higher plant height as compared to other seeded species and Yablit had the lowest plant height (Fig 19). Annual cereal grass treatments did not significantly differ between barley, wheat and oat, which were clustered in the middle growth rates. Oats however, had the highest growth rate, differentiating from the other annual grasses, resulting in higher plant height than the other annual cereals by the end of the season. There was an expected significant interaction between treatment and date (p<.0001), showing that different species grew at different rates during the season.

Figure 19: Plant height (cm) by Treatment and Sampling Date

Plant Percent Cover (Target, Weeds and Bare Ground)

Plant cover data measured cover of the target species, weed species and percent bare ground. We analyzed each data set for treatment and date effects. Plant percent cover data was averaged for each target species (seeded or planted species) within treatment plots and arcsin transformed for analysis. We detected significant treatment differences (p<.0001) and significant date effects (p<.0001), although there was no significant interaction (p=0.1635). Based on a Tukey analysis, Yablit maintained the highest percent cover of the seeded target species and wheat had lowest percent cover of seeded species. Oat/vetch combined treatment had the second highest percent cover, being significantly higher than either barley or wheat.

Figure 20: Plant Percent Cover (cm) by Treatment and Sampling Date
Percent Cover by Treatment was categorized by: Target (seeded species), % bare ground and % other species (weeds). Barley=B; Oat/ Vetch=OV; Vetiver=V; Wheat=W; Yablit=Y; and Control=C. Letters above treatment show statistically significant differences in Target species

There was a significant seasonal effect (p<.0001) with February having the highest percent cover of target species and significantly lower weed cover compared to December or April sampling events. Based on Tukey analysis, the oat/vetch treatment maintained the lowest weed invasion throughout the study.  The wheat treatment suffered extensive weed invasion.

Figure 21: Wheat treatment showing extensive weed invasion

The percent cover of weeds significantly differed by treatment (p<.0001) and by sampling date (p=.0001), with the control as expected having the highest cover of weeds, and the oat/vetch treatment having significantly lower cover of weeds compared to other treatments. Based on Tukey analysis, weed cover in wheat, barley, and vetiver did not differ significantly from each other.

Bare soil is vulnerable to rain driven erosion. We detected significant differences in bare soil cover by treatment (p<.0001) and by sampling date (p<.0001), with a significant interaction effect (p<0.0002). Based on Tukey analysis, Yablit had the lowest percent bare ground, followed by oat/vetch, which provided a dense cover with little to no bare ground after vegetation established. All treatments show a decrease in bare ground during the growing season, except barley at the last sample date, which began to senesce earlier in the spring and therefore had more exposed ground.

Figure 22: Percent cover of bare soil in each treatment, by sampling date

Plant Biomass

Biomass measurements significantly differed by treatment (p<.0001), with Vetiver having significantly higher biomass and the control having significantly lower biomass as compared to the annual cereal treatments and yablit, which did not significantly differ from each other (Fig 23). Although not significantly different, the oat/vetch treatment had the second-highest biomass.

Figure 23: Biomass (g) per treatment

2. Sheet Erosion

Erosion Pins

Figure 24 shows the mean soil elevation differential between baseline measurement and final post rain survey, based on point data collected from the hand erosion pin measurements.  Although treatment differences were not statistically significant, vetiver and oat/vetch treatments had slight deposition at 5 m, and yablit at 10m.  Although all other locations had slight soil loss, <0.02 cm, with wheat treatment having highest overall soil loss, and vetiver having high soil loss only at 1m.  There were no significant differences detected in slope position.

Figure 24: Mean soil elevation change (cm) by treatment and slope position

Erosion Transect Profile

Figure 25 shows an example of surface micro-topographic elevation profiles produced using RTK along the pin transect line. Changes in micro-topography result from both sediment trapping through deposition and sediment loss through erosion. High variance made it difficult to detect treatment effects.

Figure 25: Surface micro-topographic representative profile: pin transect soil elevation (cm)

Analyzing the erosion pin profile data across the study area, no significant differences were detected by treatment or slope position. Analyzing the data by slope position without considering treatments showed significant deposition at both 1m and 5m slope positions (p=0.05), compared to the influent (0m). The treatment effect became significant (p=0.0006), when we averaged all slope positions and analyzed the differential in soil elevation. Wheat differed from other treatments at the 10 m (C) location. As we removed wheat and the C position from TSS analyses, we reanalyzed the erosion profile data without the C location and wheat treatment, with the result that treatment effects were no longer significant.

3. Gully Erosion

Gully volume data were analyzed using two different methods to define gully morphology and estimate amount of soil lost by comparing before and after the rainy season measurements.
1- GIS analysis: We created a surface raster using elevation data collected with the RTK and a GIS tool for interpolation. We used the cut and fill tool to quantify the volume of soil lost during the winter, taking into consideration depositional areas observed at the bottom of the gully. Using this method we calculated a volume of an estimated 123.6 m³ lost and an estimated 39.3 m³ deposited for an estimated total soil loss from this one gully alone of approximately 84.3 m³. We generated contours of the surface to represent the gully morphologic change over time.

Figure 26: Topographic analysis of gully volume, showing contours representing areas of soil loss and soil deposition, as measured in changed gully dimension during winter 2018-2019.

2- Autocad surface analysis: this tool creates a surface raster with the data by triangulating the points and calculating a surface based only on geometry. These two rasters, created from the October and April surveys, are then overlaid, enabling calculation of net amount lost (cut) and net amount gained (fill). Results from this method were comparable, yielding an estimated volume loss of 135.68 m³ and an estimated deposition of 28.6 m³ net soil loss of 107.08 m³

Figure 27: Gully morphologic change before and after rainy season 2018-2019, showing areas of erosion and deposition on sides and bottom of slope, based on soil elevation data collected at 20 cross sections in gully.

Both methods yielded similar results with an average net soil loss of 84-107 m³. We used Excel to generate a line, based on field measurements, and calculated accumulated distance from starting point of each cross section. This enabled us to create a visual representation to distinguish spatial locations where significant changes occurred. Figure 28 presents representative cross sections. Appendix A presents graphs of each of the 20 cross section data.

Figure 28: Representative cross section profiles

4. Sheet Runoff Sample Collection

Runoff Collector (RCU)
We designed, constructed and installed runoff collector units to test their ability to collect sheet runoff samples during rain storm events, simultaneously across the landscape, collecting only the most concentrated runoff. Most studies investigating VBS have been conducted in a controlled setting using rainfall simulation, due to the challenge of collecting statistically robust surface runoff samples during short lived episodic storm events. We analyzed RCU water samples, measuring concentrations of total suspended sediments (TSS) and NPK nutrients, in order to compare the capture efficiency of different species composition and different lengths of VBS, as measured by comparing different slope positions (0m plot influent, 5m middle, and 10m plot effluent).

Total Suspended Sediments (TSS)
The ability of VBS to reduce TSS concentration was evaluated by calculating the difference in TSS as measured from filtering RCU runoff samples at three slope positions. We collected 63 of 72 full sample bottles (87.5 %) on January 9, 2019. We relocated the empty bottles, however we collected similar amount of full bottles again on January 29, February 17, March 4 and April 4, 2019. If we did not find more than 80% of sample bottles filled, we did not analyze the samples, due to lack of statistical power. Based on the four times during the winter we collected bottles, we detected reduced suspended sediment concentration of runoff VBS by an average of approximately 12-17% within the first 5 m across all treatments, compared to runoff entering the VBS (Fig 29). Samples from the 10m slope position showed higher TSS concentrations than the influent (see discussion).

Figure 29: Mean total suspended sediment concentration (mg/l) by slope position

The RCU TSS data were log transformed, in order to normalize the data and stabilize the variances. The results of the analysis of variance showed a significant date effect (p<.0001), and a significant interaction between treatment and location (p=0.0087) and between date and treatment (p=.0372). The treatment effect on TSS differed over time and by slope position (Fig 30). We reanalyzed average TSS concentration without C slope position and wheat treatment, finding significant reduction (p=0.027) from the influent (A) to midplot (B). Analyzed individually by treatment, two treatments showed a significant slope location effect: barley (p=.024) and the combined oat/vetch (p=.0015) treatments, resulting in significant TSS reduction occurring within 5 m. Yablit shows TSS reduction after 5 m, but differences were not statistically significantly. The control and wheat did not show any reduction in TSS.

Figure 30: Mean total suspended sediment concentration (mg/l) by treatment

We analyzed the effect of the maximum storm intensity of rain events prior to the RCU sample collection on TSS concentrations. The greater the storm intensity, the greater the TSS concentration (p<0.0001). Location was significant (p=.047), with influent (A) having higher concentrations than mid plot (B).

Nutrient analysis

RCU samples were analyzed to quantify concentrations of nitrogen (N), phosphorus (P), and potassium (K) pools in agricultural runoff leaving the farm fields. Only one RCU sample date was analyzed for nutrients (January 9, 2019). No significant differences were detected for N, P or K. Nitrate did not significantly differ by treatment or slope location. We checked residual variances for potassium (K) and phosphorus (P). No significant differences were detected in K or P by treatment or slope location, and no interaction effect was identified.

5. Stream Morphology

Using RTK measurements, we conducted a total of 16 cross section analyses on Nahalal stream. We began by resurveying cross sections initially measured in 2016 (Fig 31). Results clearly show how the streambed has shifted (Fig 32).

Figure 31: Locations of 2016 and 2019 Cross Section Measurements on Nahalal

We established a meaningful baseline for future evaluation of intervention activities. We also measured cross sections in locations that will be down gradient from the model farm, which is planned to be established this year.

Figure 32: Stream bed shift, based on 2016 and 2019 Nahalal cross section measurements

Discussion

We investigated five different species compositions as vegetated buffers to test their ability to reduce impacts from agricultural runoff. Species were selected based on plants that are commonly used on farms or that are familiar to farmers, to increase the likelihood of adopting VBS as a best management practice. Another application for VBS is directly along the stream as riparian vegetation. There is an interest in expanding stream corridors for improved flood management, increased ecological values, and protecting stream water quality from agricultural runoff. Species selected for this study were not targeting riparian vegetation as the primary goal, but was initially targeting application on farms, using annual species that is compatible with the farmers’ annual land management cycle.
In targeting riparian vegetation, we are interested in restoring native perennial species that will require minimal long-term maintenance. However native perennial species often invest initial energy in root structure rather than aboveground biomass, leaving the soil exposed early in the rainy season. Previous research testing VBS efficacy at filtering agricultural runoff considered using annual species as a nurse crop to provide short-term erosion control and enable native perennial species to successfully establish, which will then provide long-term erosion control and other ecological services (34). None of the species that we tested are native to Israel. Future work is planned to investigate suitable native perennial species that can both provide erosion control and filter agricultural runoff effectively while restoring riparian vegetation. One challenge to perennial species is resisting weed invasion in this environment, which receives high nutrient agricultural runoff.
Morphology of annual species differs from perennial species, for example perennial grasses often have greater density of biomass at the base of the plant compared to annuals species. We considered two noninvasive perennial species and compare them to four annual species.
Plant Establishment and Efficiency

We investigated the ability of each treatment to establish and provide plant canopy cover successfully, early in the rainy season. This is an important factor in protecting the soil from erosion at the onset of the rainy season, and for filtering high concentrations of accumulated agricultural chemicals that are transported in the first rains, known as first flush (23,41). We measured runoff water quality, comparing suspended sediment and nutrient reduction and capture efficiency, based on species composition as the primary treatment effect.

Vetiver has dense biomass at its base, which may serve to provide a physical obstacle, increasing sediment capture of TSS from runoff. However, the data do not support high sediment capture in vetiver treatments. This may be due to the planting density of the vetiver seedlings. Seedlings averaging in height 25 cm were planted 12 cm apart along irrigation lines, in rows one meter apart. While the seedlings grew substantially, having the highest plant height and biomass, in rows they did not spread outside their planted area. Therefore, space remained between plants and between rows all winter, while all the other treatments established cover throughout the plots.
It is likely that the planting density was insufficient for the purposes of creating a physical barrier to intercept surface runoff and trap suspended sediments. Possibly, sheet runoff went around the plants in open areas as the path of least resistance, therefore, this treatment did not result in high capture efficiency. Increasing seedling planting density may improve sediment capture efficacy. Due to its efficacy in soil stabilization, this plant warrants additional investigation. In addition, this species has deep roots which both facilitate infiltration and potentially uptake of excess fertilizers from subsurface and groundwater flows. Although we did not measure infiltration rates, Vetiver plots appeared to be less muddy than other plots throughout the season, suggesting higher infiltration. There was insufficient nutrient data to evaluate the efficacy of this species ability to uptake excess nutrients from sheet runoff and we did not look at shallow groundwater. Additional research is needed regarding optimal planting density and the potential use of this species for ameliorating excess fertilizers and pesticides in water flows.
The other perennial species tested in this study was Yablit. As the native form of this species is very aggressive, we chose the sterile “cross one”, which was genetically modified from a non-native variety. We established the plot using a “carpet” of the grass, which resulted in having highest plant cover and lowest percentage of bare soil at the onset of the rainy season. While this treatment had the lowest erosion, as shown by the erosion pin data, it did not result in high sediment capture, based on the RCU TSS results. It is possible that this “carpet” placement on top of the soil may have created a slight micro-topographic obstacle, which may have resulted in sheet flow bypassing these plots treatment. Small differences in micro topography are sufficient to result in developing preferential hydrologic flow paths (21). We detected sediment capture, but not at an efficient level.
The three cereals tested, barley, wheat and oats, provide a comparison between similar growth cycle species. This is clear based on the clustering of these species in plant height, cover and biomass. However, combining the oat with the vetch served to provide dense pant cover, as shown in plant cover data, with the oat/vetch combined treatment having the second highest percent cover, being significantly higher than either barley or wheat. In addition to the oat/vetch combination providing dense cover, vetch is a nitrogen fixer, providing an additional ecologic benefit in this high nutrient environment.

We measured percent cover of weeds because a farmer is reluctant to introduce opportunities for weeds to establish on his farm. The oat/vetch treatment had significantly lower cover of weeds compared to other treatments, showing that it successfully resisted weed invasion. Weeds established in vetiver treatments in the open spaces between plants and rows. Wheat plots were invaded by weeds, although not statistically different from barley and vetiver treatments. Dominant weeds included mustard (Sinapis alba L), Mediterranean milk thistle (Scolymus maculatus L.) and/or (Silybum marianum) and mallow (Malva nicaeensis). There was an observable gradient in wheat plant cover, with the top half of the plots having denser cover of wheat and the bottom half of the plots (closer to stream) being shorter and more yellow, as well as dominated by weeds later in the season. This gradient by slope location was observed to a lesser extent in other treatments, likely caused by differences in soil and soil nutrients, as well as soil texture, as the plots were closer to the stream.

Bare soil is vulnerable to rain driven erosion. Yablit had the lowest percent bare ground, due to its growth form. The oat/vetch had the second lowest bare soil cover. These species were seeded at the same time as the other annual cereals, yet the combined treatment provided a dense cover early in the growing season, with little to no bare ground exposed after vegetation established. Barley began to senesce earlier in the spring and therefore had more exposed ground at the last sample date. The control plot was meant to be maintained as bare soil. Although it was sprayed with herbicide, it was not done at a sufficient frequency to prevent the establishment of weedy opportunistic plant species. Due to plant cover establishing during the season, it did not provide a strong representation for control, which may have masked treatment effects, making it more difficult to distinguish.
Rainfall, Runoff and Agricultural Activities
Early rains wet the soil, beginning to generate surface runoff when soil pores are saturated. Runoff volumes are influenced by rainfall amount and intensity, soil type and slope, and plant cover (39). Plant species differ in their rooting structure and therefore influence the infiltration capacity of the surrounding soil (21). While a rain storm event on Dec 19-20 resulted in 20 mm of accumulated rain onsite, it was insufficient to generate surface runoff and no water was collected in the RCU. This can be explained partly by rainfall intensity and partly by agricultural activities. Rainfall intensity of the storm averaged 20 mm over 10 hours, without periods of high rainfall intensity.
In 2017-2018, we conducted preliminary plant establishment rate trials in the same area as the present plots. The sloped agricultural lands up gradient from the treatment plots were left bare and exposed throughout the early winter, until corn was planted in rows. A farm road was present between the plots and the agricultural field, compacting soils and accelerating runoff generation. This year, 2018-2019, the sloped agricultural lands up gradient from the treatment plots were planted in wheat in November, 2018. Field preparation did not include bed construction or rows, rather the whole field was covered uniformly. This reduced runoff volumes because there is higher infiltration, less compaction and minimal concentrating of surface runoff in the farm field. In this sense, this agricultural years’ activity is comparable to cover crops. Further there was no road installed above the plots this year, also reducing compaction and runoff generation. The combination of these two factors resulted in higher infiltration rates and less surface runoff.

Runoff Collector Units (RCU)

We succeeded in demonstrating the efficacy of the RCU design in the field. RCU bottles were checked after each early storm event. January 9, 2017 was the first time that the majority (87%) of bottles were filled. The four sets of samples collected during the winter included similar and slightly higher sampling success. Ping pong balls effectively floated to seal the inflow and prevent diluting the sample. We tested adding additional water onto the flume in the field to confirm that the entrance was closed. Water flowed over the ping pong ball and out the back of the RCU arm. We relocated RCUs that were found empty, moving them within one meter to locations that appeared to be slightly lower elevation to increase chances of filling bottles. Some bottles remained empty in all events despite relocation efforts. This seemed especially true in the vetiver treatments, which may be due to higher infiltration resulting from its rooting depth and structure.

Stream Bed and Flooding

While vegetation filter strips have been demonstrated to be effective in providing several ecologically beneficial services, optimizing their efficacy is site specific. Many previous VBS studies are conducted on sloped farm lands that receive agricultural effluent or adjacent to water bodies. In this study, plots were intentionally established near the bottom of the slope, close to the stream, in support of planning efforts to expand stream corridors. The soils are heavy clays. During the winter, these soils become saturated and fluid.
During winter storm flows this year, the steam left its established channel bed and instead established a new path (Fig 33) on the upgradient edge of the existing riparian vegetation, primarily the Common reed (Phragmites australis). The new channel was approximately 5-6 meters from the previous bed.

Figure 33: Stream flowed in new path upgradient of existing riparian vegetation

RCU TSS

The explanation for the effluent location having higher TSS concentrations than influent TSS concentration suggests that the sample was contaminated and we are not measuring surface runoff as it flows through the plots. We observed two different field issues. Storm event flooding extended into the downgradient end of the plots in Block 3 and 4 (Fig 34). Results from the RCU TSS data showed a decrease in TSS concentration between the influent as compared to 5m into the plots. However, there was a significant increase in TSS concentration after 10m, at the effluent of the plots. This was unexpected and the source of the increased sediments is unclear. It is not likely that the VBS itself between 5 and 10 m contributed high sediment. We propose that what we measured here may not be runoff water quality as it flows through 10 m of VBS, but may instead be stream water contaminated RCU samples. As a result, effluent plot (10m) RCU samples had significantly higher TSS concentrations compared to the 0m and 5m RCU samples. While these results do not demonstrate improved VBS sediment capture efficiency with additional buffer width, it does demonstrate the natural process and function of the floodplain, in terms of depositing stream sediments, nourishing the floodplain, conserving soils and improving water quality. For the purposes of VBS planning, it is important that we target native species adapted to floodplains and tolerant of flooding.

Figure 34: Visible evidence of stream flood extending into bottom of plots

Heavy clay soils in the plot area remained saturated during and after storm events. In some locations, when RCU sample bottles were removed for collection, the RCU sleeve was immediately filled with saturated soil water (Fig 35). When we tried to place a new empty sample bottle, soil water flowed into the empty bottle (Fig 35). As a result of this observation, we tried to reduce contamination by delaying replacement of new empty bottles at the time of sample collection. Instead, we waited until before the next storm event before placing the bottles to prepare for storm runoff collection. It is still possible that what we measured here may not be runoff water quality as it flows through 10 m of VBS, but may instead be saturated soil water that did not drain and instead flowed into our collecting bottles, contaminating RCU samples. This could occur if the rate of soil saturation around the RCU was higher than the runoff rates, providing another possible explanation for the effluent plot (10m) RCU samples having significantly higher TSS concentrations compared to the 0m and 5m RCU samples. This study established plots as close to the stream as possible, in order to investigate the processes occurring in the riparian zone adjacent to the stream. Results demonstrate that the specific soil type adjacent to the stream must be considered in defining appropriate and effective plant species, again stressing the importance of site specificity.

Figure 35: Saturated soils with standing water in RCU days after storm events

In addition to testing buffer width, we compared plant species composition to begin defining specific guidelines for plant restoration along streams or in farm field. Our objective is that these plants provide a range of ecosystem services, including: soil stabilization and runoff filtering, increase infiltration and groundwater recharge and improve ecological habitat. We also test the role of plant morphology in sediment capture by comparing perennial and annual species, discussed earlier.
Treatment effects on TSS differed over time and by slope position, with treatment differences significant overall only if wheat treatment was removed from analysis. This may be because the wheat treatment was heavily invaded with weeds and it performed similar to the control or because the entire field was planted in wheat above the plots, therefore this treatment did not differ from the rest of the field. Roots of cultivated wheat are shallow and thin. The above ground canopy of wheat is also thin, basal leaves are also less dense compared to the other cereal species tested in this study. This may have lowered rainfall interception resulting in increased erosion. Data support this theory, in that wheat had high bare ground exposure.

RCU Nutrient Analyses

Nutrient analyses were only conducted on one sampling date, due to field conditions. Nutrient flux is driven by microbiological activity. Samples must be collected from the field and submitted to a laboratory within 48 hours. Throughout the 2018-19 rainy season, storms often occurred Thursday night into the weekend, delaying sample collection necessary to enable these analyses. This demonstrates the challenge of collecting storm runoff samples in a natural catchment.
Data from the one sampling event January 9, 2019, show no statistical differences between influent and effluent N, P, K concentrations. However, concentrations were low overall, for ex Nitrate samples were <5 mg/l in all samples, including entering the plots. Detecting nutrients in runoff is challenging, in coordinating sample collection with fertilizer application timing, as compared to slow release fertilizers, each affecting nutrient pools differently. Previous studies show a range of nutrient reduction efficacy. Site specific conditions, including slope and runoff rate, rainfall intensity and residence time of the runoff, as well as species composition are variables that may affect nutrient uptake performance(21). Additional studies are needed to correlate agricultural input schedules with sampling to increase the likelihood of meaningful data. Other variables should be investigated in order to evaluate opportunities for potential benefits.

Erosion Pins

Subtle landscape features cause variability in surface sediment deposition and soil erosion, resulting in high surface micro-topographic heterogeneity across the field. This variability made it difficult to detect treatment and slope position effects. In addition, the method of collecting continuous measurements between the two erosion pins, which defined erosion transects, lacked precision in repeating the exact point when re-surveying. This variability may have led to compounding small errors in calculating the area, again likely decreasing the ability to detect treatment effects.

Stream Morphology

We measured elevation of representative points along cross sections perpendicular to the stream. This work was done with extreme care, but some error in soil elevation is likely due to the earth work needed to provide access, which may have changed micro-topography. The resulting cross sections demonstrate how the stream bed has shifted, within several meters of Blocks 3 and 4, flooding the lower half of the plots.
In recent years, there has been a shift in river management practice from an engineering-based ‘command and control’ paradigm that emphasized concerns for channel stability, hydraulic efficiency, predictability, and hazard mitigation, towards a more holistic focus that included concerns for river health and sustainable management, promoting the idea of allowing the stream space to move , as part of “freedom space” initiatives (42). The shift of the stream bed observed in this year demonstrates the importance of erodible corridors and giving the stream room for movement.

Conclusions

One primary goal of this study included investigating specific buffer widths to determine optimal buffer size. For both political and economic viability, we seek to maximize sediment capture and nutrient uptake, while minimizing the amount of land needed to come out of production. These data will assist in developing guidelines urgently needed to specify the amount of land needed to implement buffer areas for stream water quality protection.
We demonstrate a reduction in TSS concentration of between 11-17% after 5 m of VBS. The combination of the placement of the buffers in heavy clay soils and the above average rainfall resulting in stream flooding into the plots made it difficult to determine treatment effects. Opportunities for increased sediment capture rates is high, based on previous studies. Optimizing placement of buffer strips to maximize filtering capacity is likely to be site specific. Planting VBS on the sloped lands before the clay riparian zone may increase VBS capture efficacy and additional investigation into VBS placement is needed. We conclude that even a small area for VBS (5m) will result in environmental benefits without extensive economic impacts.

For the application of VBS as part of stream expansion, plants appropriate for implementation on riparian corridors must be tested to evaluate their efficacy in a range of locations. Additional studies are needed to investigate how sediment capture is affected by various buffer widths in a variety of settings with a variety of plants, to begin to develop appropriate, general guidelines that will need to be evaluated for each site specific application.
We measured a significant reduction in TSS concentration in two of the treatments within 5 m, oat/vetch and barley. Overall, oat/vetch treatment performed the best in this study, yielding TSS reduction and maintaining the highest cover with the least weed invasion and bare ground. These plants have both economic and environmental value, therefore, we recommend that these species should be selected as cover crops and VBS, as compared to barley and wheat for on farm VBS services. They may also serve as short term nurse crops on riparian VBS, when planted with native perennial species, although this needs further investigation to ensure that the oat/ vetch will not compete with native perennial long term establishment.

While vetiver did not perform well in TSS reduction, potential ecological benefits warrants additional investigation into this plant. Increasing the plant density may results in higher sediment capture and should be investigated in future studies.
We designed, constructed and installed RCU to capture runoff samples in order to quantify nonpoint source pollution. We succeeded in demonstrating that the RCU effectively captured robust samples and recommend this as a cost effective method for future studies.
Stream restoration now considers providing the stream with some freedom to move, as compared to constructing trapezoidal channels that flood only in 10 year storm events or greater. The movement of the stream bed that we demonstrate here supports the need to plan riparian zones to enable this natural movement, which is likely to have flood reduction benefits if planned properly. Additional consideration should be given to this issue as a management goal.

References
1. Zhang Z, Fukushima T, Onda Y, Mizugaki S, Gomi T, Kosugi K, et al. Characterisation of diffuse pollutions from forested watersheds in Japan during storm events — Its association with rainfall and watershed features. Sci Total Environ [Internet]. 2008 Feb 1 [cited 2019 Apr 14];390(1):215–26. Available from: https://www.sciencedirect.com/science/article/pii/S0048969707010510?via%3Dihub
2. Pimentel D, Resosudarmo P, McNair M, Crist S, Sinclair K, Blair R, et al. Environmental and Economic Costs of Soil Erosion and Conservation Benefits. Science (80- ) [Internet]. 2006;267(5201):1117–23. Available from: http://science.sciencemag.org/content/267/5201/1117.short
3. Zhang Q, Liu D, Cheng S, Huang X. Combined effects of runoff and soil erodibility on available nitrogen losses from sloping farmland affected by agricultural practices. Agric Water Manag [Internet]. 2016;176:1–8. Available from: http://dx.doi.org/10.1016/j.agwat.2016.05.018
4. Zhang Q, Chen S, Dong Y, Liu D, Yang X, Yang Z. Controllability of Phosphorus Losses in Surface Runoff from Sloping Farmland Treated by Agricultural Practices. L Degrad Dev. 2017;28(5):1704–16.
5. Jain MK, Das D. Estimation of Sediment Yield and Areas of Soil Erosion and Deposition for Watershed Prioritization using GIS and Remote Sensing. Water Resour Manag [Internet]. 2010 Aug 2 [cited 2019 May 14];24(10):2091–112. Available from: http://link.springer.com/10.1007/s11269-009-9540-0
6. Carkovic AB, Pastén PA, Bonilla CA. Sediment composition for the assessment of water erosion and nonpoint source pollution in natural and fire-affected landscapes. 2019;6–8.
7. Scorpio V, Crema S, Marra F, Righini M, Ciccarese G, Borga M, et al. Basin-scale analysis of the geomorphic effectiveness of flash floods: A study in the northern Apennines (Italy). Sci Total Environ. 2018;640–641(June):337–51.
8. EPA. https://www.epa.gov/nps/nonpoint-source-agriculture. 2016;
9. Zhu QD, Sun JH, Hua GF, Wang JH, Wang H. Runoff characteristics and non-point source pollution analysis in the Taihu Lake Basin : a case study of the town of Xueyan , China. 2015;15029–36.
10. Carpenter DO. Human Health Effects of Environmental Pollutants: New Insights. In: Trends in Levels and Effects of Persistent Toxic Substances in the Great Lakes [Internet]. Dordrecht: Springer Netherlands; 1998 [cited 2019 May 14]. p. 245–58. Available from: http://www.springerlink.com/index/10.1007/978-94-011-5290-7_11
11. Schwarzenbach RP, Egli T, Hofstetter TB, von Gunten U, Wehrli B. Global Water Pollution and Human Health. Annu Rev Environ Resour. 2010;35:109–36.
12. Eshel G, Egozi R, Goldwasser Y, Kashti Y, Fine P, Hayut E, et al. Benefits of growing potatoes under cover crops in a Mediterranean climate. Agric Ecosyst Environ [Internet]. 2015 Dec 15 [cited 2019 Jul 3];211:1–9. Available from: https://www.sciencedirect.com/science/article/pii/S016788091500184X
13. Mtibaa S, Hotta N, Irie M. Analysis of the efficacy and cost-effectiveness of best management practices for controlling sediment yield: A case study of the Joumine watershed, Tunisia. Sci Total Environ [Internet]. 2018 Mar 1 [cited 2019 Jul 3];616–617:1–16. Available from: https://www.sciencedirect.com/science/article/pii/S0048969717330085
14. Skagen SK, Melcher CP, Haukos DA. Reducing sedimentation of depressional wetlands in agricultural landscapes. Wetlands. 2008;28(3):594–604.
15. Defersha MB, Melesse AM. Effect of rainfall intensity, slope and antecedent moisture content on sediment concentration and sediment enrichment ratio. CATENA [Internet]. 2012 Mar 1 [cited 2019 May 14];90:47–52. Available from: https://www.sciencedirect.com/science/article/pii/S0341816211001949
16. Rein FA. An economic analysis of vegetative buffer strip implementation case study: Elkhorn slough, monterey bay, california. Coast Manag. 1999;27(4):377–90.
17. Fu B-J, Su C-H, Wei Y-P, Willett IR, Yi-He •, Lu¨• L, et al. Double counting in ecosystem services valuation: causes and countermeasures; Double counting in ecosystem services valuation: causes and countermeasures. [cited 2019 May 14]; Available from: https://esj-journals.onlinelibrary.wiley.com/doi/pdf/10.1007/s11284-010-0766-3
18. Lal R. Soil conservation and ecosystem services. Int Soil Water Conserv Res [Internet]. 2014 Sep 1 [cited 2019 May 14];2(3):36–47. Available from: https://www.sciencedirect.com/science/article/pii/S2095633915300216
19. Howarth RB, Holl K. Paying for restoration. Restor Ecol. 2000;8(3):260–7.
20. Correll DL. Principles of planning and establishment of buffer zones. In: Ecological Engineering. 2005.
21. Rein (Moshe) FO. Vegetative buffer strips in a Mediterranean climate: Potential for protecting soil and water resources. [Internet]. [cited 2019 May 14]. Available from: https://elibrary.ru/item.asp?id=5305336
22. Dukes MD, Evans RO, Gilliam JW, Kunickis SH. EFFECT OF RIPARIAN BUFFER WIDTH AND VEGETATION TYPE ON SHALLOW GROUNDWATER QUALITY IN THE MIDDLE COASTAL PLAIN OF NORTH CAROLINA. Am Soc Agric Eng. 2002;45(2):327–36.
23. Zhao J, Zhao Y, Zhao X, Jiang C. Agricultural runoff pollution control by a grassed swales coupled with wetland detention ponds system: a case study in Taihu Basin, China. Environ Sci Pollut Res [Internet]. 2016;23(9):9093–104. Available from: http://dx.doi.org/10.1007/s11356-016-6150-2
24. Tingle, C., Shaw, D., Boyette, M., & Murphy G. Metolachlor and metribuzin losses in runoff as affected by width of vegetative filter strips. Weed Sci. 1998;46(4):475–9.
25. Lerch RN, Lin CH, Goyne KW, Kremer RJ, Anderson SH. VEGETATIVE BUFFER STRIPS FOR REDUCING HERBICIDE TRANSPORT IN RUNOFF : EFFECTS OF BUFFER WIDTH , VEGETATION , AND SEASON 1. J Am WATER Resour Assoc. 2017;53(3).
26. Ministry of Agriculture SCD. השיגה תונורקע תעמטהל םיכרד תפמ ו תוינידמ ריינ לארשיב : םיניינע ןכות םדקל ומיכסהו וליבוה ש , םיבר םיפוגל תפתושמה יוגיה תדעוו השיגה תמאתה תועצמאב עבט תרימשו עקרקה , םימה יבאשמ. 2018;1–35.
27. De Jalón DG, Bussettini M, Rinaldi M, Grant G, Friberg N, Cowx IG, et al. Linking environmental flows to sediment dynamics. Water Policy. 2017;19(2):358–75.
28. Aguiar TR, Rasera K, Parron LM, Brito AG, Ferreira MT. Nutrient removal effectiveness by riparian buffer zones in rural temperate watersheds: The impact of no-till crops practices. Agric Water Manag [Internet]. 2015;149:74–80. Available from: http://dx.doi.org/10.1016/j.agwat.2014.10.031
29. Carretta L, Cardinali A, Zanin G, Masin R. Effect of Vegetative Buffer Strips on Herbicide Runoff from a Nontilled Soil. Soil Sci. 2017;182(8):285–91.
30. Castelle AJ, Johnson AW, Conolly C. Wetland and Stream Buffer Size Requirements — A Review. J Environ Qual. 1994;23(5):878–82.
31. Parson JE, Gilliam J, Munoz-Carpena R, Daniels RB, Dillaha, Theo A. Nutrient and sediment removal by grass and riparian buffers. Environ sound Agric Proc Second Conf. 1994;
32. Tate KW, Pereira MDGC, Atwill ER. Efficacy of Vegetated Buffer Strips for Retaining. J Environ Qual. 2010;33(6):2243.
33. Gumiero B, Rinaldi M, Belletti B, Lenzi D, Puppi G. Riparian vegetation as indicator of channel adjustments and environmental conditions: the case of the Panaro River (Northern Italy). Aquat Sci. 2015;77(4):563–82.
34. Rein FA, Los Huertos MW, Holl K, Langenheim JH. Restoring native grasses as vegetative buffers in a coastal California agricultural landscape. Madroño. 2007;54:249–57.
35. Dudai N, Tzion I, Nitzan N, Chaimovitsh D, Shachter A, Zemak Shamir S, et al. Agronomic and economic evaluation of vetiver grass (Vetiveria zizanioides L.) as means for phytormediation of diesel polluted soils in Israel. J Envron Manag. 2018;
36. Truong P. The global impact of vetiver grass technology on the environment. Proc Second Int Conf Vetiver Off R Dev Proj Board. 2000;48–61.
37. Truong P, Vetiver T, Pinners E. Vetiver System for Erosion and Sediment Control , and Stabilisation of Steep Slopes ( With special references to the Ho Chi Minh Highway , Vietnam ) Vetiver System for Erosion and Sediment Control , and Stabilisation of Steep Slopes ( With special referen. 2015;(May 2015).
38. Hughes BB, Fountain MC, Carlisle AB, Levey MD, Chavez FP, Gleason MG. Climate mediates hypoxic stress on fish diversity and nursery function at the land–sea interface. Proc Natl Acad Sci. 2015;112(26):8025–30.
39. Dunne T, Leopold L. Water in Environmental Planning. New York, N.Y: W.H. Freeman and Company; 1978.
40. Griggs G, Tait J. ction structures on beaches along Northern Monterey Bay, California. The effects of coastal protection. J Coast Res. 1988;Special Is(4).
41. Li S, Wang X, Qiao B, Li J, Tu J. First flush characteristics of rainfall runoff from a paddy field in the Taihu Lake watershed , China. 2017;8336–51.
42. Brierley G, Hooke J. Emerging geomorphic approaches to guide river management practices. Geomorphology. 2015;251:1–5.

Appendix A: Gully cross section profile comparisons are presented below: