Slope length effect on sediment and organic litter transport on a steep forested hillslope : upscaling from plot to hillslope scale

Upscaling and the effect of slope length on the sediment production are important issues in hillslope hydrology. This study investigated the effect of slope length at a steep forested site between 2006 and 2010 using erosion plots and neighboring sediment traps. The hillslope had high litter cover but sparse understory due to overgrazing by wild deer. Sediment transport increased for short distances (5 m to 10 m), then decreased over the hillslope (> 10 m). In plots this effect was due to organic litter accumulation. Organic litter production declined progressively along the hillslope continuum. When accumulated organic litter altered the microtopography, sediment production declined more with increase in slope length. Heavy precipitation events did not change the pattern of these scale effects. These results demonstrate that the slope length effect had inconsistent controls on sediment transport due to variations in soil surface conditions of the landscape. Such variations in the slope length effect cause uncertainty in sediment flux estimations in steep forested landscapes.


INTRODUCTION
Sediment movement by overland flow (sheet flow) over hillslopes is a non-continuous, intermittent transport process in which particles move a short distance with a gamma distribution before deposition (Kirkby, 1991).Travel distance has been linked to stream power and soil particle size in transit (Hassan et al., 1992), with stream power being mainly related to the soil surface slope, flow depth, and flow velocity (Bagnold, 1977).Of these factors, flow depth and velocity are dependent on infiltration processes, in that infiltration of excess rainfall and re-infiltration of overland runoff can be enhanced by the presence of organic litter on the soil surface.Previous studies have reported that the travel distance of sediment is variable during storm events and depends on the surface soil conditions, rainfall properties (Moreno-de las Hears et al., 2010), and overland runoff characteristics as well as spatial connectivity (Gomi et al., 2008).
Sediment yield is governed by the redistribution of water and sediment (Zehe and Sivapalan, 2009).However, the scale effect on sediment redistribution is still a significant area of uncertainty in hillslope hydrology that needs to be addressed (Cammeraat, 2002).In small catchments in which hillslopes make up a large proportion of the area, it is important to understand the connectivity of fine-scale processes over the hillslope and their effect on sediment production.Examination of scale issues in plots of different lengths can help improve our understanding of the internal erosion processes of hydrological systems because system evolution is largely related to smaller scale processes (Cammeraat, 2002).
Variations in the slope length scale effect are a response to differences in land use and ecosystem organization (Wilcox and Newman, 2005), which themselves vary greatly and require clarification in regard to their effects (Cammeraat, 2002).The nature and intensity of scale effects vary significantly in different landscape contexts (Cerdan et al., 2004), being influenced by ecosystem characteristics and the dominating effect of biological processes (Cammeraat, 2002;Willcox et al., 2003).The scale effect is associated with the spatial distribution of soil surface characteristics (e.g., Zehe and Sivapalan, 2009), which produces spatial heterogeneity in the landscape.Spatial heterogeneity causes uncertainty in process-based modeling of hydrology and erosion (Beven, 1996); this problem can be examined by looking at the effect of scale.If the scale effect can be better understood, runoff and sediment delivery in a hillslope and catchment can be modeled more accurately.
Organic litter can affect the development of microtopography (Ellis et al., 2006) by controlling raindrop erosion, and consequently the hydraulic properties of overland flow by influencing roughness and the travel distance of sediment (Moreno-de las Hears et al., 2010).The nature of the litter layer affects the infiltration rate and consequently overland runoff generation, connection, transport distance, and distribution over the slope.
Few studies have examined the slope length effect on sediment production of steep forested hillslopes with sparse vegetation, rich ground cover, and accumulated organic litter, also referred to as litter dams (Leguèdois et al., 2008).The present study investigated the effect of slope length on sediment and organic litter transportation over a steep forested slope that was mainly covered by organic litter and which experienced heavy rainstorms.We were particularly interested in determining how the presence of highly altered microtopography with litter dams alters sediment and organic litter production.In addition, we investigated the function of surface conditions under heavy typhoon storm events.In a forested landscape, organic litter production not only interacts with erosion processes but may also play a significant role in nutrient cycling.Organic litter is a source of nitrogen and phosphorus for forest floor (e.g.Bridgham et al., 1995).The study area was a steep hillslope of a headwater catchment, where soil erosion due to reduced vegetation cover by overgrazing of wild deer is a major concern of water resource and forest management (Ghahramani et al., 2011).Steep slopes are common landscapes in headwater catchments in the study area.

MATERIALS AND METHODS
Field monitoring was conducted on a hillslope in the Tanzawa Mountain, Kanagawa Prefecture, Japan.Three erosion plots of 2 m wide and lengths of 5, 10 and 20 m were installed on the hillslope approximately 1.5 m apart, and three sediment traps were installed approximately 30 m to the west in order to examine longer slope length effect on sediment and litter production.Weekly monitoring was conducted from 2006 to 2010.Detailed explanations of the site characteristics, experimental setup and measurements are presented in the Supplementary Material 1.The spatial connectivity of sediment and litter was assessed as sediment and litter yield (g) ratios for pairs of consecutive plot lengths, i.e., length of small plot to the next larger plot (e.g., Le Bissonnais et al., 1998;Moreno-de las Heras et al., 2010).We considered ratios of 5 m/10 m and 10 m/20 m plots.Decreasing transport connectivity occurs for ratios > 1 and increasing transport connectivity occurs for ratios < 1. Precipitation depth for monitoring periods ranged from 0.5 to 587.5 mm, thus to examine the influence of precipitation on the slope length effect, precipitation periods were grouped into 3 clusters by applying the K-means clustering method (Supplementary Material 2).The number of clusters was determined by their silhouette value which indicates a better separation of data for different cluster numbers.

Sediment and litter flux
During monitoring of plots, 104 precipitation periods (weekly base) were observed.Figure 1a shows mean annual sediment production (gm −2 ) for monitored plots.In three of the four monitored years, sediment flux increased markedly from the 5 m to 10 m plot scale, but then decreased markedly at the 20 m scale.As shown in Figure 1c, in four years of monitoring, organic litter production (gm −2 ) decreased as slope length was increased.Variation of sediment production was different among plots (Figure 1b) but variation of litter production decreased linearly with slope length (Figure 1d).Sediment and organic litter production (gm −2 ) were different from the plots on hillslopes with longer slope length and in the absence of litter dams.On hillslope mean sediment and organic litter production of all monitoring years decreased with hillslope length (Figure 2).

Sediment and organic litter transport connectivity
From the 5 m plot to the 10 m plot, the slope length effect was positive for 86% of events (scale ratio less than 1) and negative for 14% of events (scale ratio larger than 1).Between the 10 m plot and the 20 m plot, the slope length effect was positive for 9% of events (scale ratio less than 1) and negative for 91% of the events with sediment production ratio more than 1 (Figure 3a, b).For litter transport from 5 m plot to 10 m plot, slope length effect was positive for 72% and negative for 28% of precipitation periods and from 10 m plot to 20 m plot was positive and negative for 43% and 53% of precipitation periods respectively.Mean values are shown in Figure 3.In hillslope and measured data by sediment traps showed that from trapper No1 to No2, slope length effect on sediment and organic litter transport were positive for 61% and 34% of observed precipitation periods respectively.Between trapper No2 and No3, slope length effect on sediment and organic  litter transport were positive for 33% and 64% of observed precipitation periods respectively (Figure 4).Along hillslope where longer distances in absence of litter dams were examined, mean sediment and litter production decreased with slope length, and the degree of decrease for sediment production was less than that of the longest plot (20 m).

Slope length effect influenced by precipitation
In all precipitation groups (clusters), sediment increased from the 5 m plot to the 10 m plot but then decreased between the 10 m and 20 m plots; this pattern was also pronounced for litter movement (Figure 5).In cluster 3 (Supplementary Material 2) in which all dimensions of clustering were highest, the length scale appeared to have the highest effect on sediment and litter transport without changing the overall pattern.High coverage by a litter layer during any kind of storm did not change the positive effect of the length scale.According to Figure 5, in all precipitation groups (clusters) for which the value of the length scale effect ratio was less than 1, sediment and litter production increased from the 5 m plot to the 10 m plot and this positive effect of length scale increased as the magnitude of the precipitation event increased (precipitation depth, average intensity, I 30min , I 24hr ).In cluster 3 (extreme storm events  Figure 5. Scale effect on sediment transport for different precipitation magnitudes.Values of precipitation factors: (i) depth (mm), (ii) average intensity (mmhr −1 ), (iii) I 30min and (iv) I 24hr are increasing from cluster 1 to 3. associated with a typhoon), sediment and litter production showed a decrease from the 10 m plot to the 20 m plot.In cluster 3, for which all dimensions of the cluster were highest, length scale appeared to have the strongest negative effect on sediment and litter transport.

Pattern of slope length effect
In the absence of rill erosion, sediment production has been reported to be unaffected by length scale or to be negatively affected due to re-infiltration of overland runoff (e.g., Wilcox et al., 2003;Parsons et al., 2006;Sidle et al., 2007;Moreno-de las Heras et al., 2010).The results for our experimental plots for slope length without litter accumulation were consistent with the results of Moreno-de las Heras et al. (2010), who demonstrated positive scale dependency of soil flux in a semi-arid landscape.Le Bissonnais et al. (1998) linked the positive scale effect in slopes with 1 and 20 m length, to microtopography, whereby runoff flow depth increases with increase in length due to accumulated upslope runoff (Dunne et al., 1991).
The nature and intensity of the scale effect vary significantly in different contexts (Cerdan et al., 2004), with different temporal dynamics of rainfall-runoff events (Van de Giessen et al., 2000), under different surface conditions (Le Bissonnais et al., 1998), and in varying states of soil degradation (Moreno-de las Heras, 2010).Some studies have emphasized that the scale dependency of sediment transport is related to rill erosion processes (Foster et al., 1977).However, in our plots with steep slope gradients and in the absence of rill erosion, positive slope dependency was observed for short slope lengths between the 5 m and 10 m plots.In this part of the slope, there was no litter accumulation.Sediment production from the 5 m plot to the longer 20 m plot did not vary linearly with slope length (Figure 1) because of changes in soil surface conditions that led to negative length dependency of soil erosion (Figure 1).
Our experimental results have shown inconsistencies in slope length effect in terms of the transport of sediment (e.g., in unmanaged forests).Over short distances (plots), the slope length effect was positive, but for long distances (hillslope) the slope length effect was negative.A negative slope length scale effect occurred with the presence of litter dams, and this length scale pattern also applies along hillslopes.With the transport of organic litter, the negative slope length effect was found for both short and long distances (Figure 4).

Connectivity of sediment and organic litter transport
With an increase in slope length from 5 to 10 m, the amount of transported sediment increased by about 2 times (scale ratio = 0.55); in other words, the mean sediment transport distance downslope was more than 5 m (Figure 3a).With increase in slope length from 10 to 20 m, the amount of transported sediment sharply decreased and the 10 m plot to 20 m plot ratio became more than 1 (= 4.4), indicating that sediment transport distance was less than 5 m.Even during heavy precipitation events, the length scale effect did not change.For example, during the storm on 6-7 September 2007, which had an associated precipitation depth of 588 mm, sediment transport length in the 20 m plot did not exceed 5 m with a 5 m plot to 10 m plot ratio of 17.
With increase in slope length from 5 m to 10 m, litter transport increased (scale ratio = 0.86).With increase in slope length from 10 to 20 m, transported litter decreased and scale ratio became more than 1 (= 1.49), meaning that the length of litter transport was less than that of the plot with 10 m length.The 10 m/20 m scale ratio of sediment was much higher than that of the litter scale ratio, and thus litter was transported longer distances than sediment.Findings indicate that litter production patterns are more consistent than that of sediment.
From the 5 m to 10 m plots, even with high coverage of litter, a positive effect of length scale on sediment and litter production was observed under monitored storm events (on average), and these high ground cover values still could not change the positive effect of slope length.The length scale appeared to have the strongest negative effect on sediment and litter transport from the 10 m plot to 20 m plot and during heavy storm events (Figure 5).This negative effect is thought to be associated with the occurrence of accumulated organic litter (litter dams), where large storm events generate more overland runoff with larger stream power capable of mobilizing more organic litter and depositing it behind and over litter dams.

Implications for erosion modeling
In existing modeling methodologies, runoff in small watersheds and consequent erosion is independent of scale (Wilcox et al., 2003), and point measurements are extrapolated to larger areas under the assumption that spatial variability or heterogeneity can be neglected, and that broadscale spatial patterns are static (Cammeraat, 2002).Horton (1945) quantified the effects of slope length on sediment production and demonstrated that erosion increases with slope length.Later models, such as the USLE (Wischmeier and Smith, 1978) and RUSLE (Renard et al., 1998), have applied this relationship.In these models, slope length is one of the main factors in predicting soil loss, but is also one of the most variable factors among landscapes (Liu et al., 2000).The transport-distance approach (Wainwright et al., 2008), which considers slope length effects, can be appropriate for erosion modeling.However, to achieve accurate results, the ecological context and inconsistencies in the scale effect should be considered.

CONCLUSIONS
The slope length scale effect was shown to differ depending on local soil surface characteristics of organic litter accumulation and upslope conditions which determined sediment transport connectivity length.The effect of slope length did not differ according to the magnitude of storm events but did differ depending on the existence of the litter dams.Slope length had a negative effect along a hillslope with rich organic litter coverage but, with the appearance of the litter dams, non-linearity began to occur in the dependency of sediment production on the length scale.The overall results showed inconsistencies associated with sediment transport with increase in the slope length in a steep forested slope with rich organic litter and sparse understory.Inconsistencies occurred due to the features of the soil surface.At the plot scale, litter accumulation (litter dam) defused the sediment production response so that along a boundary across the contours, overland runoff and sediment flux connectivity could be cut.Such processes create uncertainty in sediment flux estimations in steep forested landscapes.

Figure 2 .
Figure 2. Sediment and organic litter production of hillslope.

Figure 3 .
Figure 3.In plots; (a) Scale ratio of sediment transport, (b) variability of scale ratio of sediment transport, (c) scale ratio of litter transport, (d) variability of scale ratio of litter transport.

Figure 4 .
Figure 4.In hillslope; (a) Scale ratio of sediment transport, (b) variability of scale ratio of sediment transport, (c) scale ratio of litter transport, (d) variability of scale ratio of litter transport.