Ecological Restoration of Headwaters in a Rural Landscape (Normandy, France): A Passive Approach Taking Hedge Networks into Account for Riparian Tree Recruitment

Abstract

The safeguard of riparian ecosystems is a major field of study in the understanding and maintenance of the ecological health of rivers. Vegetation communities found on these ecotones ensure essential functions such as limitation of river bank erosion and protection of rivers from pollutants. The aim of our study is to investigate the potential for natural regeneration of trees on river banks after passive restoration. We have also studied the influence of landscape on recolonization through the analysis of the influence of hedge networks. Our study takes place on headwaters in Normandy (France) on Vallée-Aux-Berges, a stream, which has been passively restored for the last 6 years. As passive restoration removes stresses (heavy trampling and grazing) caused by cattle on river banks, we expect it to help the growth of natural plant communities. The condition of this stream—from the start of restoration work to the present—is compared to another one in the same catchment considered to be ecologically healthy. Our results suggest that passive restoration leads to an increase in tree cover on river banks and contributes to the improvement of the banks' physical integrity. Landscape structure seems to be a major factor for this recolonization: the more the stream is surrounded by hedge networks, the more the recolonization by trees on banks is effective. These results indicate that the influence of landscape structure should be considered in future restoration management in similar headwaters.

Introduction

Intensification of agriculture has drastically modified the structure of rural landscapes by changing field size, destroying hedgerow networks, and increasing nutrients and fine sediments fluxes to rivers (Sarriquet et al. 2006). In this context, restoration ecology is a necessity to avoid those perturbations on rivers, especially in headwaters. Indeed, the importance of headwaters rises from their control on both structure and operation of superior order rivers, as illustrated in the River Continuum Concept (Polis et al. 1997; Gomi et al. 2002; Power & Dietrich 2002; Lowe & Likens 2005). This concept highlights the strong influence of debris from riparian vegetation on headwater streams. Farther downstream, in-stream production increases, reducing dependence on terrestrial material as an energy source. Because processes occurring upstream depend primarily on terrestrial production, and downstream regions are influenced by upstream processes for which resources are limited, protection efforts should focus on headwaters (Saunders et al. 2002). It is therefore necessary to protect the integrity and the ecological services provided by rivers (Arthington et al. 2010).

Because of the position of river banks at the interface between aquatic and terrestrial environments, their preservation is essential to prevent potential degradation of riparian ecosystems (Naimann et al. 2005). Vegetation colonizing this ecotone fulfills several functions: it filters pollution and limits eutrophication (Correl 1997), creates root systems that stabilize banks (Gray & Barker 2004), plays a part in energy cycling (flux of nitrates) (Pinay & Décamps 1988) (e.g. it filters light, is involved in trophic networks, etc.), and provides physical structure for faunal habitats (Copeland et al. 1996). Vegetation can also have an important impact on the dispersion of nutrients, plant propagules, or wood debris (Perucca et al. 2009).

Solutions frequently employed by practitioners of riparian restoration, such as installation of woody plants, are rather laborious and expensive to setup. Moreover, restoration practitioners often do not consider the potential for the natural functioning of these ecosystems to simplify the management styles of river bank vegetation and generally do not adopt a landscape perspective (Jungwirth et al. 2002). In order to make restoration programs sustainable, they must be integrated with an interdisciplinary approach and not be restricted to the river. Many restoration projects do not have ecological objectives, and surveys in cases of success or failure are rare (Bash & Ryan 2002). Authors agree, however, on the need for a scientific base and monitoring funds before beginning any project of restoration of rivers (Bash & Ryan 2002; Shields et al. 2003).

To be ecologically successful, projects must involve restoration of natural river processes (e.g. channel movement, river-floodplain exchanges, organic matter retention, biotic dispersal). Restoration using hard-engineering methods should not be the first method chosen as it often constrains the channel (Palmer et al. 2005). In this study, we focused more particularly on a passive restoration project of a first-order stream. Passive restoration, with minimal intervention, takes into account concepts of restoration ecology and has many ecological and economic advantages (Clewell & Aronson 2007). Passive restoration may be the first and the most important step in riparian restoration, especially in smaller order streams where flow regimes are still intact (McIver & Starr 2001). The question of the effectiveness of passive approaches is highly discussed in the literature. Prach and Hobbs (2008) suggested that a minimum intervention approach is advisable. Using spontaneous succession as a restoration tool has the potential to save both time and effort and allows the development of a less contrived ecosystem. In riparian ecosystems, passive efforts like livestock exclusion are often all that is required to achieve restoration success (Briggs 1995). Kauffman et al. (1995) recorded a significant recovery of cottonwood and willow after fencing of a riparian area. Briggs (1992) found that of 27 riparian revegetation projects, 19 achieved their objectives, largely because of the exclusion or management of livestock. A recent study led in northern Utah has shown that passive restoration is an effective management approach for restoring stream habitat and has the potential to minimize interactive effects of habitat degradation (Hansen & Budy 2010). However, passive restoration seems to be less successful when many factors contribute to riparian degradation, such as the occurrence of invasive plants, presence of roads, channel incisions, and upstream interruption of natural flow (McIver & Starr 2001).

The main goal of our study was to restore a stream highly impacted by livestock. We restored it passively using exclusionary fencing and feeding troughs and footbridges. The cumulative effects of trampling and grazing of young seedlings by cattle has been already shown as a primary factor in the deficit of tree germination (Dembélé et al. 2006). Grazing also impacts both the erosion resistance of river banks and the erosive forces applied to the banks (Trimble & Mendel 1995). By using fences to exclude cattle, this type of restoration enhances the emergence of spontaneous growth of ligneous species within the herbaceous layer. Trees and shrubs influence the morphological characteristics and the functioning of rivers, in particular by the shade they provide, thus allowing the maintenance of stable temperatures. Moreover, ecological engineering studies have shown the importance of root systems on banks susceptible to erosion (Gray & Barker 2004). De Baets et al. (2005) highlighted the important role of grass roots in increasing topsoil resistance to erosion by concentrated flow. Roots also reduce erosion and the crumbling of river banks by supporting cohesion of the substrate (Abernethy & Rutherfurd 2001). Even dead, root systems can have an important function on surface drain water by creating biopores. This improvement of drainage decreases risks of massive movement to the banks (Tabacchi et al. 2000). These holes can also provide optimal larval habitats for the dragonfly Hadrothemis camarensis, as showed by Copeland et al. (1996).

The aim of our study is to have a better understanding of the influence of the landscape structure, in particular the importance of hedge networks, in river restoration projects. Indeed, both proximity and extent of land use patches interplay to influence the degree of changes in riparian structure (Ferreira et al. 2005). The question which underlies our work is as follows: how can we take into account the natural processes of riparian woody plant recruitment for easier and cheaper management strategies of river bank vegetation? In the context of an increase in intensive farming, which involves the alteration or complete removal of hedge networks, it is important to show how this bocage network could contribute to ecological restoration practices on rivers. Thus, we studied variations in ligneous plant recruitment after a passive restoration campaign, and carried out in a 6-year follow-up on the brook Vallée-Aux-Berges (VAB) (Normandy, France). We also studied this stream in comparison to another one that is ecologically healthy and has been historically protected from cattle pressure. Goals of this study include determining if ecological passive restoration, as an alternative to banks revegetation, is possible in this landscape context. It would thus be possible to validate this method and to extend it to other riparian ecosystems.

Methods

Study Site

The study was carried out in the catchment basin of the Oir (Normandy, France), a Sélune River tributary on which ecological surveys have been carried out for 20 years evaluating the connections between the quality of salmon populations and the quality of the habitat (Fig. 1). This area includes two tributaries of the Oir River, VAB and the control river, The Roche. Both occur in landscapes used for pasture and cultivation.

Figure 1. Map of the study site with Vallée-Aux-Berges and The Roche (full lines) and the catchement boundaries (dotted lines).

On the VAB river, we identified the presence of two small forest patches and a large hedge network. The largest forest (Clérice wood) is situated at the upstream zone of this study area, and the other is located in the downstream zone. Zone B on this river is restored by exclusion fencing, as are the other zones (A, C, and D), but the zone B landowner regularly mows vegetation to a very low level. There is thus another treatment in this area in addition to the exclusion fencing: this simulates grazing but avoids trampling. In an effort to restore the Oir River and its tributaries, several local organizations and public agencies collaborated on this riparian restoration project (SAGE [Amenagement Shema and Water Management of Sélune] and ONEMA [The French National Agency for Water and Aquatic Environments]). Their goal was to avoid degradation from both grazing and trampling to contribute to sustainability of the salmonids populations in these streams. Our goal is to suggest a new approach to ecological restoration practices in that context that takes into consideration local constraints.

Historically, and due to current landowner practices, some tributaries have been protected by fences in the Oir River catchment basin. On the basis of this management history, and also by evaluating the flora composition, we consider these river banks to be ecologically healthy. They can, therefore, be used as a reference in the context of restoration to prevent river bank erosion due to the direct access of cattle to the river. Specifically, in this study, The Roche is fully comparable to the VAB. The majority of The Roche has always been protected from the impact of cattle by using exclusionary fencing. It is a first-order stream (Strahler 1957) characterized by pastures and is in the same catchment area as the VAB. Furthermore, the surrounding hedge density cover between these two rivers is similar. The Roche is surrounded by 10.3 ha (82.5%) of grassland, whereas VAB is surrounded by 12.9 ha (75.1%) of grassland and 2 ha (11.7%) of crops (Figs. S1 & S2). There are no crop areas surrounding The Roche.

Monitoring of Woody Species

The banks corresponding to the study area (VAB) were divided into four sectors, from sector A (upstream) to B, C, and D (downstream) (Fig. 1). The emergence of ligneous trees was noted from 2004 to 2006 and again in 2009 and 2010. Sampling was carried out along the river banks. A total of 78 plots of 15 m² (15 m long, 1 m large) were sampled on VAB during the 5 years of monitoring. The tree seedlings were counted and their abundance was estimated according to the Tansley's scale, from 1 to 5 (Tansley 1926), 1: rare; 2: occasional; 3: frequent; 4: abundant; 5: dominating (very abundant). The same sampling method was used on the Roche in 2010 on four sectors (“A” to “D” as for VAB).

To study the influence of the type of hedges on seedlings, we focused on the most common ligneous species on these streams: alder (Alnus glutinosa [Aglu]), hazel (Corylus avelana [Cave]), oak (Quercus robur [Qrob]), and willow (Salix atrocinerea [Satr]). An inventory of all potential seed-bearers for these four species was carried out in a radius of 50 m around each plot containing seedlings of alder, oak, willow, and hazel. At each point, we recorded the co-ordinates, identified the species present, the percentage cover of species (scale of Tansley), the sector considered (A, B, C, or D), and the type of landscape structure (HE, distant hedge; SH, surrounding hedges near the stream (parallel); HP, hedge perpendicular to the bank; R, riparian forest). Hedges were considered as distant (HE) when they were not directly connected to the river contrary to SH, HP, and R.

Data Analysis

Various indices can be used to identify differences in the emergence of tree seedlings and species assemblages (Burel et al. 1998). First, specific richness (S) of the ligneous taxa was calculated on the whole study site and on each sector A, B, C, and D. The Shannon index (H′) was also used to characterize richness more precisely on each sector during the monitoring period. This index is often used to characterize biodiversity (Bryja et al. 1998; Michel et al. 2006), particularly in comparative studies, and we hypothesize that it should reflect the response of the woody recruitment after the initiation of passive restoration. The Shannon index was calculated for each zone in the VAB from the initial state to the present. We compared sites and years of restoration using the Hutcheson's test (Hutcheson 1970; Cheng 2004) due to the low value of n species (n < 30). A factorial analysis was then used to highlight the variation of the tree composition in each zone of the VAB river banks and compared to those observed on the Roche.

Finally, a generalized linear model (GLM) (binomial family, logit bond) was built to test the influence of the presence and distance of mature trees from hedge networks on the recolonization of river banks after passive restoration.

All statistical analysis was performed with the statistical software R (The R Foundation for Statistical Computing, version 2.8.1) and statistiXL 1.8. All maps were created with ArcGis.

Results

Temporal and Spatial Variation of Woody Species Diversity on the Banks of VAB

After the exclusion of the VAB stream banks, woody species richness doubled in just 3 years of passive restoration (2004–2006). Indeed, there were 6 species in 2004 and 13 in 2006. Following that, only Ulex europaeus appeared between 2006 and 2010. Figure 2 represents the variation in tree distribution over the restoration period. Alnus glutinosa was strongly dominant in 2004 and decreased during the restoration period with increases in Quercus robur and Corylus avelana and the appearance of Fraxinus excelsior. The “other” category includes Castanea sativa, Crataegus monogyna, Populus nigra, Prunus avium, Cytisus scoparius, and Rubus fructicosus (Fig. 2). The proportion of these dominant species is similar between VAB and the control river after 5 years of restoration.

Figure 2. Abundance index of dominant woody species (Alnus glutinosa [Aglu]; Corylus avelana [Cave]; Fraxinus excelsior [Fexc]; Quercus robur [Qrob], and Salix atrocinerea [Saur]) on banks at the restored brook VAB between 2005 and 2010 and on The Roche in 2010.

The Shannon diversity index in the four zones of VAB (from upstream to downstream: A, B, C, and D) significantly increased during restoration between 2004 and 2009 (ZA: p < 0.001; ZB: p = 0.53; ZC: p < 0.01; ZD: p < 0.01) (Fig. 3). This index indicates a strong variation along the upstream–downstream gradient (p < 0.01). The time axis represents the species diversity variation within each zone during the 6 years of restoration (Fig. 3). The longitudinal axis (Fig. 3) represents changes in species diversity along a gradient from upstream to downstream. Moreover, in 2004, no species of tree was identified in ZA, whereas the diversity of woody species increased significantly in this zone in 2005 and 2006. A significant increase of this diversity index in 2005 and 2006 was also observed downstream, in ZC and ZD. At the same time, ZA showed a sharp increase in this index between 2004 and 2005, then a more moderate increase between 2005 and 2006. However, following the increase of this diversity index during the first 3 years of restoration (i.e. from 2004 to 2006) in these three zones (A, C, and D), no significant variation was observed between 2006 and 2009. Finally, zone B is the only one which presented no significant increase in the index of species diversity over the 6 years of restoration.

Figure 3. Box plot of mean Shannon diversity index of zones A, B, C, and D on the VAB brook from 2004 to 2010. The line within each box represents the median value and graphics on the right-hand side represents the average value. Differences between boxes during years of restoration are represented as follows: **p < 0.01; ***p < 0.001. Error bars denote the SE.

The factorial analysis highlights specific variations of ligneous taxa of the four zones since the beginning of restoration in 2004 (Fig. 4). Each point represents the specific composition of one zone at the indicated year. Zones A, C, and D showed a high response to passive restoration. All points for these zones in the period before restoration occur on the left of the first axis, while after restoration, points for these three zones are on the right. This indicates that the restoration had a similar effect on the entire stream. The Roche data point is situated on the right-hand side of the first axis. This figure indicates a slow change in the species composition in zone B 6 years after restoration.

Figure 4. Distribution of the sample scores on the first two factorial analysis axes. Plots represent the tree assemblage of each bank before restoration in 2004 and after restoration in 2010 (acronyms are letter of the zone and the year of sampling, e.g. for ZA from 2004 to 2010 is notified as follows: A04; A06; A09; A10). La Roche is considered as a reference brook (unimpacted and unrestored).

Effects of Hedge Networks at the Landscape Scale for the Spontaneous Recruitment of Ligneous Plants on VAB Banks

The hedge density surrounding the two streams is similar: 80.08 m/ha for the control stream and 86.71 m/ha for VAB.

Table 1 shows the results of linear model relating to the influence of hedges on seedling presence on banks. We observed that distant hedges significantly influenced the emergence of willow trees only. However, proximity to surrounding hedges, forested riparian areas and perpendicular hedges seemed to facilitate the recruitment of other trees on the river bank. More precisely, the presence of mature trees in forested riparian areas seemed to increase seedling recruitment in Q. robur, A. glutinosa, and C. avelana.

Table 1. Results of the GLM for the effect of hedge networks and the presence of mature trees in them for the spontaneous emergence of young trees on the VAB banks after passive restoration. Significant effects are represented in bold

	Distant Hedges		Surrounding Hedges		Riparian Forest		Perpandicular Hedges
	f	p	f	p	f	p	f	p
Alnus glutinosa	2.631	0.105	4.254	0.037	11.9	0.001	7.426	0.007
Corylus avelana	1.062	0.303	7.877	0.005	3.989	0.034	3.618	0.047
Quercus robur	0.499	0.48	3.187	0.062	3.634	0.043	4.586	0.031
Salix atrocinerea	9.878	0.002	4.887	0.027	1.465	0.226	2.197	0.138

Discussion

Our results indicated a strong variation in the diversity of tree species during the 6 years after restoration. In just 5 years, the passive restoration method used here allowed a return to a diversified woody state in the riverine system. This is in accordance with other studies that have shown that the seed bank is able to initiate recovery after perturbation on an African riparian ecosystem (Vosse et al. 2008) and where the majority of tree recruitment is observed from the seed bank within the first few years of restoration (Reinecke et al. 2008). Indeed, the species richness of woody species doubled in only 3 years, showing a strong resilient potential for this type of river bank. The analysis of the Shannon diversity index of trees indicated that diversity is higher downstream near the junction of the VAB and the Oir River. We observed a strong dominance of alder trees during the early stages of restoration. This was the pioneer species in 2004, and its occurrence drastically fell in 2006. This could be a promising mode of restoration because alder trees close to water are an important habitat component for larval species, for example the Clouded Apollo (Parnassius mnemosyne) (Meier et al. 2005). Alnus glutinosa is adapted to full sun and is often found in pioneer stages of channels (Rivière 2007). It creates deep and dense root networks that provide soil protection. In the study area, the establishment of Alnus glutinosa was followed by a higher individual quantity of other species such as Quercus robur or Fraxinus excelsior.

We have no experimental data concerning the source of recruitment of these species during restoration. Recruitment may have come from the seed bank or from the spread of propagules by wind, water, or animals. During a study led in 2005 on this tributary, Bernez et al. (2005) showed that the height of banks was one of the key factors structuring herbaceous communities. River banks are deposit areas for propagules transported by water. Variations in water level are also a limiting factor for the establishment of many observed, woody species that require drier conditions. Tree communities are considered as objective indicators of ecological responses to flow alteration (Bejarano & Sordo 2011). An experiment in situ would allow us to understand the relative importance of these processes.

The analyses show that zone B does not present the same variation as the remainder of the brook. This is certainly due to the management method used by the owner of the meadow who regularly mows the bank to a very low height. As a result, young emergent seedlings cannot grow to maturity. From another study led in an abandoned Mediterranean cropland, it appears that, after reforestation of native shrubs and trees, and once seedlings have become established, mowing is beneficial if applied just before the first dry season (Benaya et al. 2005). As a result, we suggest the owner to stop mowing this zone for a few years in order to increase recruitment of trees as in the other zones of this brook.

In this study, we also considered the importance of the landscape network for riparian restoration projects. River bank proximity to mature ligneous taxa from various types of hedges seems to be an explanatory factor for the banks' colonization by the seedlings of trees. Indeed, the landscape context is very important for riparian diversity, but this level of scale is rarely considered in ecological restoration. Maekawa et al. (1997) have shown that the increase in the abundance of two species in a Japanese riparian landscape contributed to the diversity by resulting in the increase of the number of vegetation types. Wozniack et al. (2009) tested the impact of landscape changes such as the intensification of agriculture, and they have shown that these changes negatively affect potential values for protected and endangered species. Here we observed that the presence of and proximity to perpendicular hedges and surrounding hedges are very important, in particular for alder, oak, and hazel trees. The distance from sources is very important for seedling recruitment. The presence of herbivores is also an important factor that can influence the natural recolonization of banks by trees. For example, oak trees produce large seeds with high nutritional content that are subject to intense predation pressures. Many studies agree that predation is one of the main factors limiting natural regeneration in many oak species (Gomez et al. 2003). So, to maximize the success of germination of trees, a short distance from source is needed. Our observations of hazel, oak, and alder trees conform to those results. However, for willow trees, no correlations were found. Distant hedges have a significant effect on willow trees, as opposed to other hedges closer to the bank. Willows constitute a high biodiversity value in riparian landscape because they often provide woody nesting cover for a variety of birds (Young & Clements 2003). The natural regeneration of willow (Salix spp.) has the additional advantage of a vigorous growth rate and production of massive root systems that can rapidly stabilize stream bank sediments (Shields et al. 1995). This is the reason that the planting of willows (Salix sp.) is often recommended or implemented for trapping sediments and organic debris (Karle & Densmore 1994) or decreasing the invasion of exotic species into wetlands (Kim et al. 2005). Many species of Salix are proposed in classical river management guidelines, without any consideration on the origin of the taxa. However, we see from our passive ecological restoration project that no plantings were needed to reach this result.

Our results are in accordance with those of McIver and Starr (2001) who compared both active and passive restoration methods in the Columbia River basin. They revealed that passive restoration may be the first and most important step in riparian restoration, especially in smaller order streams where flow regimes are still intact. Therefore, allowing for the possibility of spontaneous succession as an ecological restoration tool creates the potential to reach a better ecological state for the restored river, save both time and effort, and allow a less directly human-influenced system to develop. Reinecke et al. (2008) observed that active restoration may be necessary when invasive plants have become established and require human intervention. But even in this case, studies should be conducted at larger temporal scales, such as this study, for 5 years and more. Such studies may reveal that ligneous plants could, in the medium or long term, be good competitors for invasives that occur in the early stages of colonization (Williamson et al. 1996).

The potential for spontaneous succession, either moderately manipulated or not, still needs to be more fully exploited. Such results will help scientists to interact with practitioners. It is helpful for hydrobiologists to expand their view to other scales and integrate landscape scale into their studies and assessments. Our study shows that farmers' practices have a strong influence on river bank restoration and that we have to work together for a success. Furthermore, we highlight the need for taking landscape network into account for ecological restoration projects of such first-order streams. This is a new view for ecological restoration practice, and it involves another way to approach river management.

It also helps ecological restoration practitioners to consider if this landscape scale approaches could be relevant and sustainable for the future; for the first time, the SERI congress at Perth, Australia in 2009 presented three sessions on restoration at the landscape scale.

Implications for Practice

• Natural regeneration by ligneous taxa may be a useful and low-cost restoration tool in this landscape context.

• Landscape network must be integrated in stream restoration studies to evaluate its potential role in such projects.

• Stream restorations can only be successful if farmers are part of the project: they are the key to restoration success.

Acknowledgments

This study has been funded, since 2004, by the SAGE-SÉLUNE (Programs of development and management of waters) and the ECOGER-PAPIER program (ecology for the management of ecosystems and their resources—agricultural landscapes, flow of pollutants, ecological impact in river). It was prolonged by an agreement INRA-ONEMA from 2009 to 2011.

The authors would like to thank F. Renault, A. Pingray, L. Muchembled, and C. Rouger for their assistance in the field. We would like to thank C. Jones for her helpful editorial advice concerning the English.