Sites and bioengineering works
The three study sites (Francon, Saignon and Brask catchments) are located in the Southern French Alps within the Durance watershed (Fig. 1 A). The Francon experimental catchment is located 10 km south-east of Digne-les-Bains. The Saignon and the Brask experimental catchments are located 10 km north-east of Sisteron (Fig. 1 B).
In the Francon catchment, the mean annual precipitation is 822 mm with peak intensity in May and September during storm events. The Saignon and the Brask catchments show a similar seasonal rain pattern with a mean annual precipitation of 787 mm
(Vallauri, 1997). All the studied terrains are composed of highly erodible Jurassic black marls and are prone to severe linear hydric erosion, characterized by active eroded gullies.
Fig. 1 Location of the study sites
(A) Location of the Alpes-de-Haute-Provence departement within France. (B) Location of the experimental catchments.
In some gully beds in the studied catchments, 773 experimental bioengineering works were built by the French forest national service (ONF) in 38 gullies in spring 2002, 2004 and 2008 respectively on the Saignon, the Brask and the Francon experimental catchments. In each gully, a set of works was built starting from the gully outlet with a single structure every 2 m, as recommended by Rey (2009). The works were installed along the whole length of gully floors with a slope of less than 40%, thus determining the number of works installed in each gully. They are made of a dead-wood sill, upstream of which a 1.20 m single row of 50 cm long S. purpurea cuttings is planted perpendicularly to the flow (Fig. 2). These cuttings are sticks 20 cm up from the ground level. The grain size distribution of sediment trapped in such vegetative barriers covers a wide spectrum of particle size. From studies conducted in the area, we known that the detritic material found in gully beds displays 44% of clay and silt together, 33% of sand, 21% of splinters between 2 mm and 5 cm and finally 2% of marly platelets the dimensions of which
0 60 km
exceed 7.5 cm in length, 5 cm in width and 2.8 cm in thickness (Mathys, unpublished data).
The study focuses on 77 bioengineering works from 5 gullies. At the date of this study (2010), all the cuttings in the sampled works had resprouted and formed living tillers.
History of sediment deposit upstream Salix barriers
Fig. 2 Three dimensional schematic view of bioengineering works in a gully bed
A bioengineering work is composed of a dead-wood sill upstream of which a single row of Salix cuttings is implanted. Salix tillers then resprout from the cuttings a few months after their implantation. The sediment deposited by human work when building the deadwood sill, as well as sediment trapped by the cuttings and tillers during rain events, are shown in different colors.
Segments of Salix tiller barriers have been chosen to maximize the homogeneity of the barrier features (height of trapped sediment and plant traits). The heights of sediment trapped by the two homogeneous segment barriers are shown by white bold arrows.
The monitoring of sediment deposit history was conducted on 15 Salix barriers in one gully located in the Saignon catchment, 20 barriers in one gully in the Brask catchment and 42 barriers in 3 gullies in the Francon catchment (Table 1). In this case, the whole Salix barrier (cuttings and tillers) was considered (Fig. 2). The sediment deposit heights (hbarrier) were measured with a permanent wooden stick implanted 10 cm upstream of the Salix barrier, in the middle of the accumulated sediment deposit (Fig. 3). The buried lengths of the wooden sticks were measured after each productive rainfall event,
determined on-site with a rain gauge recording total rainfall and event characteristics.
We consider that the Salix barrier has reached a threshold when the sediment trapping activity by the tillers becomes efficient. As the cuttings are sticks 20 cm up from the floor, the 20 cm line in hbarrier (Fig. 4) distinguishes the sediment trapping by the cuttings (below 20 cm) and by the tillers (above 20 cm). This efficiency threshold can thus easily be identified by a threshold value of 20 cm in hbarrier. We checked that productive rains occurred regularly during the study period.
Relationship between sediment deposit heights and morphological traits of tiller barrier segments
Sampling the homogeneous tiller barrier segments
The investigation of the links between tiller morphology and sediment deposit height was conducted on selected segments of the barriers, in order to minimize barrier heterogeneity, both in terms of tiller morphology and sediment deposit (Fig. 2). Indeed, within a single barrier, we can see one segment which has trapped sediment and another which has not, due to heterogeneity in the micro-topography as well as in sediment flow intensity. The study focused on barrier segments with a quantity of sediment upstream, which shows that these segments have been efficient in trapping sediment and that sediment flow has passed through the barrier. According to Rey and Burylo (in press), the amount of sediment produced in these eroded gullies is significantly higher than the maximum amount which can be trapped by the barriers.
Consequently, if a sediment deposit is found upslope of a segment barrier, it means that this is the maximum amount that the barrier can trap at the time we observe it. Indeed, if the barrier morphology allows sediment trapping, then, since sediment production is greater than the barrier trapping capacity, sediment is trapped up to the maximum sediment height. If the plant barrier morphology changes and does not allow sediment trapping any more, rainfall events can lead to a release of previously trapped sediment.
As rainfall events are more frequent than the time-pattern according to which plant morphology changes (seasonal growth), we consider that the sediment height observed is the maximum sediment height. In spring 2010, 8 segments were studied at the
Saignon catchment, 20 segments at the Brask catchment and 21 segments from the three gullies in the Francon catchment (Table 1).
Table 1 Number, type and location of Salix barriers and tiller barrier segments selected for the study The history of sediment deposit height (hbarrier) was conducted on whole Salix barriers. The investigation of the link between Salix tiller morphology and sediment deposit height (hsegment) was conducted on similar tiller barrier segments, selected from living Salix barriers which were still visible (not buried under sediment) in spring 2010. Therefore this part of the study was conducted on tiller barrier segments identified in a reduced number of Salix barriers, compared to the history of the sediment deposit height record.
Measurements of sediment deposit heights upstream of homogeneous tiller barrier segments
The sediment trapping efficiency of tiller barrier segments was evaluated according to the sediment deposit height (hsegment) from a horizontal line (basal line) starting from the top of the cuttings, if the Salix cuttings are not buried under sediment (Fig. 3 A). Mostly in the Saignon catchment, Salix cuttings were buried under sediment deposits. In this case, hsegment was measured from the lowest level of the scale formed by the barrier (Fig. 3 B). This sediment deposit height was considered to be the maximum at the time of measurement. Indeed, according to rainfall intensity and sediment load, sediment height can either increase or decrease.
Numbers of entities Francon Brask Saignon
Gullies 3 1 1
History of sediment
deposit (hbarrier) Salix barriers 29 8 5 20 15
Salix tillers morphology and sediment deposit (hsegment)
Salix barriers 9 5 3 18 6
segments 13 5 3 20 8
Morphological traits of tiller barrier segment
The morphology of the tiller barrier segments was investigated in the same segments as those which were selected for hsegment measurement. Plant morphological traits were chosen from traits that have been shown or have a potential to influence sediment trapping (Bochet et al., 2000; De Baets et al., 2009; Isselin-Nondedeu and Bedecarrats, 2007; Burylo et al., 2012). If the cuttings were still visible (i.e not buried under sediment), the basal diameter of the tillers was measured at the top of the cuttings (basal line), where the tillers have started resprouting (Fig. 3 A). If the cuttings were buried, the basal diameter was measured at the basis of the tillers (basal line), where they leave the floor (Fig. 3 B). The baseline shift in cases where the cuttings were buried
Fig. 3 Schematic longitudinal section of bioengineering works The sediment flow runs from the upper left corner to the bottom right corner of the diagrams. The height of sediment deposit trapped by the whole Salix barrier (hbarrier) is measured with a permanent wooden stick. The height of sediment deposit trapped by a homogeneous segment of Salix tillers is indicated by the hsegment arrow. The basal line points out the level where basal tiller morphological traits were measured. (A) Case where the Salix cuttings are still visible (not buried). (B) Case where the Salix cuttings are buried under sediment.
applied to all basal morphological traits. The basal stem density was measured by the number of stems at the base of the tillers divided by the width of the segment barrier (Fig. 2). The basal shoot branching index was measured by the number of stems 10 cm
above the base of the tillers divided by the number of stems at the base of the tillers (figure 3). Only the first 10 cm were considered because it has been shown that the mean sediment trapping height per year varies between 5 and 10 cm (Rey and Burylo, submitted). Finally the SOP (Ʃ (basal stem diameter / barrier length) was determined (De Baets et al., 2009).
Variations in hsegment and morphological trait values between tiller barrier segments of various ages were investigated using the Kruskal and Wallis test analysis because of non-Gaussian distribution. Significant differences between ages were assessed by post hoc test using the Wilcoxon test (Sprent, 1992) because of non-Gaussian distribution.
Principal component analysis (PCA) was performed to relate hsegment to the morphological traits of Salix tiller barrier segments. We performed this analysis on all the 49 tiller barrier segments, regardless of the age-site group they belong to. An analysis of covariance (ANCOVA), with the tillers age as covariate, was used to test the influence of the age and the morphological traits of the tillers on hsegment. This analysis allows points out the morphological traits of tiller barrier segments, which explain variations in hsegment within a group of tiller barrier segments as if they showed constant age-site features.