ELSEVIER Geomorphology 19 (1997) 99-116 Large woody debris and river geomorphological pattern: examples from S.E. France and S. England a,l a ax n,,^,i.ii b,* H. Piégay al, A.M. Gumell Laboratoire de Geographie Rhodanienne, URA 260 du CNRS, IS rue Chevreul, 69362 Lyon Cédex 07, France School of Geography, University of Birmingham, Edgbaston, Birmingham Bl5 2TT, UK Received 1 April 1996; revised 30 October 1996; accepted 30 October 1996 Abstract The study of accumulations of dead wood within the fluvial environment has been mainly undertaken in mountain streams and rivers within the Northwestern United States, and particularly in hydrosystems which have experienced little riparian vegetation cutting or disturbance by man. Appraisals of the spatial variability in the physical character of accumulations of dead wood has mainly highlighted the volumes of large woody debris (LWD) accumulations and the local channel morphological properties induced by their presence. The spatial variability in the accumulation and processing of organic material forms one of the central concepts of the River Continuum Concept, which characterises the occurrence and processing of organic material, of which LWD is an important component, according to a longitudinal gradient along a river's course. Some studies have extended the concept by illustrating the importance of the lateral dimension, particularly in large rivers with extensive floodplains, and by relating the occurrence of dead wood to fluvial morphodynamics. However, to date there has been no synthesis of the relationship between LWD and the geomorphic pattern of the river channel. Although the research literature shows that the routine clearance of wood from water courses is not an environmentally-sympathetic strategy, within Europe LWD accumulations are usually seen as a river management problem and are routinely cleared from river channels. This paper addresses these physical and applied aspects of the role of LWD. It presents an analysis based upon semi-natural hydrosystems in S.E. France and S. England. The forested corridors discussed are currently or have recently been maintained. They are essentially young and so produce relatively small amounts of woody debris in relatively small-sized individual pieces in comparison with the rivers studied in North America. Using observations from these example river corridors, the relationship between rivers of a particular size and geomorphic pattern and the dynamics of dead wood is described and evaluated. Major constrasts in the role of LWD are found between small, single thread rivers, and larger, piedmont, braided and wandering rivers. Some points of synthesis concerning the ecological, hydraulic and morphological impacts of dead wood are drawn from these examples, and are used as a basis for proposing some simple maintenance rules. Keywords: large woody debris; river pattern; riparian zone management; European rivers Corresponding author. Fax: +44(121)4145528. E-mail: a.m.gurnell@bham.ac.uk 1 Fax: +33-72722185. 0169-555X/97/S17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S0169-555X(96)00045-1 100 H. Piégay, A.M. Gurnell / Geomorphology 19 (1997) 99-116 1. Introduction River valleys across much of the World were once extensively wooded with the consequence that fallen trees, branches and leaves formed a significant component of the river environment. Clearance of wood from rivers to maintain the capacity of the channel to transmit water and so to alleviate flooding; to avoid blockage and damage to structures such as bridges, piers and weirs; to improve navigation and; (mistakenly) to improve fish passage have led to far-reaching changes in the morphology, stability and ecology of river systems (e.g. Sedeli and Frog-gatt, 1984). The adverse impact of woody debris clearance has been magnified by the widespread deforestation and drainage of floodplains and the engineering of river channels to completely new forms through straightening, dredging and bank stabilisation (Brookes, 1988; Petts, 1990). In the United Kingdom, the long history of woody debris clearance from river channels is well illustrated by the reference to the routine clearance of weeds, trees and bushes in historical sources (e.g. Laws of Sewers, 1732). In France, since the floods of 1993, the authorities have clearly stressed their desire to manage riparian vegetation. The Barnier Law of 2 February 1995, which relates to the protection of the environment, reaffirms in article 23 that the riparian owner is required to undertake a regular clearing regime to reinstate the water course to its natural size and depth, to maintain banks by pruning woody vegetation, and to clear dams and debris, whether floating or not, in order to maintain the natural flow. The analyses presented in this paper are based upon two premises: First, it is already recognised within the international research literature, that the removal of debris dams is not consistent with the maintenance of a healthy aquatic ecosystem. Indeed, the contrary is true (Bisson et al, 1987; Piégay and Maridet, 1994; Gurnell et al., 1995). Whereas floodplain deforestation has a long history in Europe, tree and driftwood clearance has been more recent in the Pacific Northwest of the United States. Here settlement did not commence until the 19th century, and so the extensive research undertaken in this region has clearly documented the devastating impact of deforestation and driftwood clearance on the aquatic environment (Maser and Sedeli, 1994). Very significant research programmes undertaken in this and other regions of North America over the last twenty years have provided the core of several recent reviews (e.g. Gurnell et al., 1995; Sedeil et al., 1988; Sullivan et al., 1987). These reviews illustrate that the routine clearance of wood from water courses is no longer an environmentally-acceptable component of river management, but they also show that much research is required outside of North America if the management of wood in rivers is to be appropriate to local environmental conditions. Second, the dynamics of large woody debris (LWD: defined as wood pieces greater than 10cm diameter and 1 m length, Platts et al., 1987) and its role in enhancing flood risk varies with the type of river, so influencing the assessment of the hazard. It also varies with the nature of human occupation of the floodplain, so impacting on the assessment of the risk. These varying aspects of both the hazard associated with LWD and the accompanying risks to flood-plain occupants implies that a flexible approach to LWD management is required to maximise environmental benefit and minimise associated risks. This paper contributes to the evaluation of the geomorphological significance of LWD in European rivers. It provides a context for this evaluation through an analysis of the international literature on LWD. The geomorphological significance and role of LWD within European rivers is then emphasised through the description of some case studies from S.E. France and S. England concerning rivers of different geomorphological pattern or type. Finally, some recommendations for management based upon geomorphological river types are presented. 2. Previous research The emphasis of previous research on LWD can be illustrated by an analysis of the content of 104 research papers assembled by the present authors from the open literature. Of these 104 papers, only one was published before 1975, but there has been a steady rise in the number of publications since then. H. Piégay, A.M. Gurnell / Geomorphology 19 (1997) 99-116 101 There is a strong regional bias in the published studies. 23 are review papers which mainly focus on North American research, and of the remaining 81 papers concerning individual research programmes, 67 are based in North America of which 34 (50% of the North American research publications) are located in the Pacific Northwest or in Alaska. This not only indicates the North American bias in the published research but also the very strong regional focus within North America and the focus on catchments where much of the forest cover is relatively undisturbed. The publications address a range of aspects of LWD character and impact. 49% provide details of the volume of LWD and the character and stability of debris accumulations. The impact of woody debris on physical habitat diversity and stability is also a major feature of the published case studies (47%). The influence of debris accumulations on sediment movement (27%), organic matter retention and decomposition (21%) and flow hydraulics (20%) relate to their significance in creating suitable habitat for fish (20%) and macroinvertebrate (10%) populations. Many studies consider the impact of forest management on debris accumulations through empirical analyses but the precise role of LWD is also investigated by debris addition or removal experiments in as many as 28% of the investigations. Whilst the general environmental significance of LWD is highlighted in virtually all of the studies, in 15% of the publications research has led to very specific recommendations about the management of in-channel debris and riparian vegetation. In summary, the content of these 104 publications on LWD shows it to have at least three levels of impact on woodland river environments: (1) LWD directly impinges upon the distribution of stream power, leading to influences on the hydraulics of in-channel flows and the distribution of overbank flows; and thus on the transport and storage of sediments and organic material within the river channel system and on the floodplain. Comparative studies between streams where debris has been retained or removed, suggest that as a result of removal, sediment yield can increase by an order of magnitude and sediment movement results in the development of bars and benches which partly replace the storage role of woody debris accumula- tions, although enhanced sediment yields persist because the bed material remains relatively more mobile than in channels where woody debris has been retained. (2) Influences on flow hydraulics and sediment storage and transport lead to secondary impacts on the geomorphology of woodland river channels including the average condition and variance in channel dimensions; the magnitude and distribution of pools and riffles; and the overall increased stability of river channels. (3) River channels containing accumulations of LWD are able to store and transmit sediments, including organic material, in a well-regulated manner; and they consist of channels which present a high physical habitat diversity both within and between debris accumulations. As a result there are major impacts of LWD accumulations on the ecology of woodland river channels. The storage, break-down and regulated release of organic matter provides temporally and spatially regulated food sources for aquatic biota. Furthermore, the complex physical structure of woodland river channels provides a variety of habitat patches which can support a wide variety of organisms at different stages of their life cycles. As a result, the abundance of LWD has been positively correlated with a high diversity in both macroinvertebrates and fish; and the removal of debris has caused a reduction in the diversity, density and/or biomass of invertebrates and fish. However, the regional focus of much of the literature, has resulted in a large proportion of the research results reflecting mountain rivers where hy-drosystems have experienced little riparian vegetation cutting or disturbance by man. Furthermore, although longitudinal gradients in the character of LWD accumulations have been identified, and their lateral role has also been recognised, the relationship between the character of accumulations and the geo-morphological pattern of the river has not been evaluated. By focusing on some European examples, the emphasis presented in this paper is on semi-natural hydrosystems in forested corridors that are currently or have recently been maintained. The forests are frequently young and so produce relatively small amounts of debris in pieces which are individually of relatively small-size in comparison with the forested river corridors studied in North America. 102 H. Piégay, A.M. Gurnell/ Geomorphology 19 (1997) 99-116 3. The accumulation and transfer of LWD in rivers of different size and geomorphological pattern A variety of factors govern the physical processes in rivers and thus their morphology. Primary factors are the volume and temporal distribution of water and sediment delivered to the channel and the nature of the materials through which the river flows. Secondary factors include the nature of the riparian vegetation, land use, and direct modification of the channel (Church, 1992). Since riparian vegetation and the associated supply of organic material to the river forms only a secondary factor influencing channel morphology, there is an enormous variety of river channel types within which the role of LWD should be evaluated. Church (1992) provides a classification of river channel types within three groups: small, intermediate and large channels, where channel size reflects the ratio of channel to boundary particle size. "Large rivers are ones in which purely fluvial processes and geological constraints determine the morphology" and thus the primary factors are dominant, whereas within the range from intermediate to small channels, secondary factors, particularly individual roughness elements such as rocks and pieces of wood become increasingly important influences on channel mor-phodynamics. In particular, the size of the channel in comparison with the size of the organic debris deliv- ered to it will have an important influence on the interaction between the two. The subdivision of river types into three groups is mirrored by the River Continuum Concept (RCC, Vannote et al., 1980) which "provides a framework for integrating predictable and observable biological features of flowing water systems with the physical-geomorphic environment" (p. 135). The RCC is a concept developed mainly by biologists based on the energy equilibrium theory of geomorphologists (Vannote et al., 1980). It associates biological invertebrate strategies and coarse to fine organic matter dynamics with a hydrological and geomorphological continuum, and hypothesises analagous functioning of biological and physical systems in terms of energy utilisation. The authors argue that allocthonous supplies of coarse particulate matter including LWD are the main source of organic matter in small streams, whereas primary production characterises larger rivers located downstream. Thus, the RCC proposes some broad characteristics of lotic communities related to three groupings of stream size: headwaters (stream orders 1-3); medium-sized streams (orders 4-6) and larger rivers (order > 6). The RCC is extended by Sedeil et al. (1989) through the incorporation of the lateral dimension, particularly in the case of large rivers where the ratio of floodplain width to river channel width may be large, and where interaction between river and flood-plain may be complex and both geomorphologically Fig. 1. Location of the Highland Water catchment. H. Piégay, AM. Gurnell / Geomorphology 19 (1997) 99-116 103 France r-v 100 km «cr /xíy- Venera */ Giffre R. 1 f~*J? ---------------------- j< ti/i í^^t^ ,py AinR~ ^t§4,Ťo> DrômeR- a JÄ •• D ^C^Kj ^UbayeR. ArděcheR. ----~^^/^4->-V"' * - ■"■- y J/»1'-. J$Ĺ l' . *> J S Kit Fig. 3. Unstructured accumulation of LWD within the Upper Highland Water. %-, Vi'&it! is a survey of the dimensions of the largest piece of debris present in the debris dams. In the summer of 1991, a survey of 76 dams within the same stretch of the Highland Water, revealed a mean length for the largest piece of wood within each dam as 5.3 m, a length which is significantly larger than the 3.2 m average channel width and which illustrates the way in which many LWD key pieces lie partly on the bank and partly in the channel. Thus, in these small- to medium-sized channels, LWD accumulates largely in the form of transverse dams within or, in the case of some long pieces of debris, spanning the main channel. In smaller streams (e.g. order 1 to 2), the small width of the channel in comparison with the size of much of the debris and the low power of the flow, can lead to relatively unstructured, accumulations of LWD, with medium to large pieces of wood remaining where they fall Fig. 4. Compact accumulation of smaller debris pieces within the Upper Highland Water. H. Piégay, A.M. Gurnell/ Geomorphology 19 (1997) 99-116 105 Fig. 5. An accumulation of LWD forming an active dam within the main Highland Water channel. In the case of the main Highland Water, clearance of dams during January 1990 may have reduced the number below that which is representative of a period without significant dam management, and so frequencies for the same stretch in 1984 are included to represent a period of negligible dam management. The percentages of dams of the three types are similar for the two dates. The major contrast between the two small streams and the medium-sized stream is the larger proportion of complete dams in the former, representing the influence of the smaller channel size in proportion to tree size. There is little difference in the proportion of active dams between the streams of different size. Table 1 also lists estimates of the biomass of debris accumulations for the three sample stretches of river in summer 1991. The lower estimate for the main Highland Water in comparison with the smaller rivers may be influenced by an incomplete recovery from the debris clearance undertaken in January 1990. Of particular significance for management is the mobility of debris dams within these small- to medium-sized rivers. Table 2 summarises the degree to which dams of different types mapped within 5.0 km of the main Highland Water in September/October 1982 persisted or changed according to a resur-vey in January 1984. Table 2 lists changes in the frequency and type of debris dams identified within 50 100 m sections of the main Highland Water between 1982 and 1984. Although there was only a 8% increase in the total number of dams during this period, 33% of 1984 dams were newly created, 24% of dams had been removed, whereas 18% had changed type. Of the three different dam types, partial dams were the most unstable with a gain of 44% resulting from the creation of new dams, 40% resulting from dam removal, and only 13% and 9% gains and losses, respectively, resulting from a change of dam type. In contrast, active dams were the most stable with a gain of only 7% attributable to dam creation, a loss of only 11% attributable to dam removal, but 25% and 43% gains and losses, respectively, resulting from a change in dam type. This suggests that once major dams accumulate they are relatively stable, and may play an important role in regulating the movement of individual pieces of debris. A further indication of the mobility of debris in this 5.0 km stretch of the main Highland Water is listed in Table 3, where the detailed recovery in the frequency of dams of different types is recorded from immediately after debris dam clearance (January 1990) over the following 16 months (to May 1991), in comparison with the condition of the dams after a long period without management (January 1984). The rate of recovery of the dams is rapid, although this recovery was undoubtedly accelerated by the major input of debris during the severe storm on 24 January 1990. The relatively large number of the partial and complete dams which developed immediately after clearance, also illustrates the mobility of LWD pieces when active dams are not there to trap them. Increased river flows during Autumn 1990 appear to have mobilised much of the LWD, reducing the number of partial and complete dams and relocating the debris within active dams. This pro- 106 H. Piégay, A.M. Gurnell / Geomorphology 19 (1997) 99-116 Table 1 Frequency and biomass of debris dams in small and medium-sized rivers River Stretch Order Biomass (kg • km"') " Number (%) of different dam types Partial Complete Active Total 1984 Main Highland Water 3-4 - 1991 Main Highland Water 3-4 5100 Upper Highland Water 1-2 6500 Bagshot Gutter 1-2 9800 109 (64%) 41 (24%) 21 (12%) 58(61%) 23(24%) 14(15%) 7 (19%) 17 (46%) 13 (35%) 0 (0%) 35(85%) 6(15%) 171 95 37 41 Biomass is expressed to 2 significant figures. vides further evidence for the importance of the more stable active dams in controlling overall debris mobility. 4.2. Larger, alluvial rivers of mountain and piedmont zones The alluvial river channels of mountain and piedmont zones are characterised by high energy, strongly varying river flows, abundant coarse sediment load and wooded corridors whose ligneous biomass is frequently well developed. These forested hydrosystems are subject, according to their geomorphic pat- terns, to specific types of debris dynamics. Six water courses in South East France are taken as illustrative examples 4.2.1. Braided river systems Braided rivers like the Giffre (Alpes du Nord), the Ubaye and the Drôme (Alpes du Sud) or the Ouvéze (piedmont of the Prealpes du Sud) (Fig. 2) have an approximately rectilinear active band. The riparian margins usually consist of pioneer vegetation which is restricted in its development by frequent floods. Thus, these hydrosystems are unable to supply large amounts of dead wood but more usually supply Table 2 Changes in the frequency, location and character of debris dams in 100 m sections of a 5.0 km stretch of the main Highland Water, September/October 1982 to January 1984 Type of change observed in 100 m sections Number of dams of different type Partial Complete Active Total 1982 total Dam Additions, 1982-Changed dam type Dams created Total additions Dam Losses, 1982-84 Changed dam type Dams removed Total losses Net change 1984 total 90 40 28 158 12(13) 10(25) 7(25) 29 (18) 40(44) 10 (25) 2 (7) 52 (33) 52 (58) 20 (50) 9(32) 81 (51) -8 (9) -9 (23) -12(43) -29 (18) -28 (31) -7 (18) -3(11) -38 (24) -36 (40) -17 (43) - 15(53) -68 (43) + 16 (18) + 3 (8) -6(21) 13 (8) 106(118) 43(108) 22 (78) 171 (108) > of 1982 totals given in parentheses. H. Piégay, A.M. Gurnell/ Geomorphology 19 í 1997) 99-116 107 Table 3 Debris dam frequency at different dates along a 5.0 km stretch of the Highland Water Date Dam type partial complete active total Autumn 1982 90 40 28 158 Jan. 1984 106 43 22 171 Jan. 1990 26 8 15 49 Jul. 1990 68 45 10 123 Nov. 1990 52 21 24 97 May 1991 58 23 14 95 pieces of flexible vegetation and small pieces of dead wood. They are also unable to retain those pieces of wood which are in transit because of the low trapping efficiency of the channel as a result of the low sinuosity, and the lack of secondary channels narrow enough to stop the transit of wood. Debris accumulations are extremely rare. The deposits of LWD tend to accumulate preferentially in the braided zone and not in the forest fringe (Fig. 6). For example, isolated trunks are often deposited on gravel bars during the falling limb of floods (Fig. 7). The transfer time of wood is relatively short as a result of the relative rarity of potential depositional sites and the high energy of the river systems. Nevertheless, some of the wood may be transported only a short distance. For example, on the River Drôme there is a low but significant correlation between the number of logs located within a 500 m reach and the number of 500 m reaches in the 3 km stretch upstream that are characterised by an eroding wooded bank (r2 = 0.13; p < 0.0001; Landon et al-, 1995). Dead wood may also have a distinctive morphological role within the active band of the channel. Under certain circumstances it provides locations for sedimentation and thus the accumulation and stabilisation of gravel banks, as has been shown by Malanson and Butler (1990), Naka-mura and Swanson (1993), Robison and Beschta (1990). Contributions of dead wood have been observed to differ between intramontane and piedmont rivers. In intramontane rivers, the contributions of dead wood are primarily delivered from the numerous tributaries along the study reaches. The wood is often profoundly altered during transport so that it consists of torn and jagged pieces. In the piedmont rivers, the riparian zone can also act as a significant source of LWD. As a result, the logs are not so heavily altered and trees, complete with branches NO Mass of large woody debris * t.ha"1 SE 0 25 50 75 100 125 * Masses were estimated on plots of 10*20m 150 175 200 225 250 275 300 325 350 375 400 Distance (m) Fig. 6. Distribution of large woody debris along a cross-section of the River Ubaye at the Enchastrayes braided site. 108 H. Piégay, A.M. Gurnell / Geomorphology 19 (1997) 99-116 Fig. 7. Distribution of debris jams and floated-in logs on the gravel bars of the active band of the Drome river, downstream of Crest. and root systems, can sometimes be observed in the river. On the Drôme, two main types of dead wood have been observed: open accumulations of branches and broken or isolated logs on the active tract; and a discontinuous line of woody debris located on eroded, forested banks. The discontinuous line of debris is very different from that described below for wandering rivers. It exists only sporadically and comprises a Table 4 Mass of LWD observed at seven sites on the Rivers Giffre, Ubaye and Ouvěze River Giffre Ubaye Champanastais Meolans Enchastrayes Ouvěze Study site Verchaix Millieres Violes Saint-Michel Total area studied (m2) 3600 2800 3800 Total LWD mass (t) 2.6 0.038 0.012 Total LWD mass (tha"1) 7.22 0.14 0.3 LWD location Active channel (% total mass) 99.0 73.0 0.0 Pioneer vegetation (% total mass) 1.0 0.0 0.0 Post-pioneer vegetation (% total mass) - 25.0 100.0 Mature forest (% total mass) 0.0 2.5 0.0 Floodplain zone along the active 0.0 0.0 0.0 channel (% total mass) Locally-produced LWD (i.e. local 0.0 25.5 100.0 blow-down not transported by river flows; % total mass) 6500 8000 0.39 1.8 0.6 2.25 48.0 93.3 43.0 6.7 2.5 0.0 6.5 0.1 0.0 0.0 9.0 0.1 1560 1320 6.8 10.4 43.5 78.6 0.0 0.0 0.0 - - 100.0 100.0 - 62.0 98.0 8.0 0.0 H. Piégay, A.M. Gumell / Geomorphology 19 (1997) 99-116 109 relatively low mass of debris in comparison with the mass accumulated within the active tract of the channel. Three main factors were found to influence the number of wood pieces (Y) located within 500 m lengths of the Drôme river bed: the braiding index (ratio of total wetted channel length to valley length, Leopold and Wolman, 1957) in the 500 m length (X[); the number of wooded islands (X2) in the 500 m length; and the number of 500 m lengths of eroding wooded banks in the 1 km section immediately upstream (X3). The following multiple regression model illustrates the relationship between these variables: ľ = 11.99X, + 10.44X2 + 6.21 X3 - 22.6 (r2 = 0.31; p < 0.0001; n = 203) X, and X2 are indicators of the retentiveness of the reach, whereas X3 is an indicator of the potential supply of wood to the reach. The role of large floods is an important determining factor in controlling the spatial distribution and dynamics of debris. There is often a discontinuity between the timing of the supply of debris to the river and the occurence of rare floods which are able to transport the supplied wood, so influencing the nature and location of depositional sites. Whereas the active band of the channel is normally the main location of LWD, during large floods dead wood is deposited as a line of debris along the junction between the active band of the channel and the adjacent riparian shrub zones. This debris line can contain large masses of LWD. For example, on the River Ouvěze after a flood whose recurrence interval has been estimated to be greater than 1 in 100 years (Chastan et al., 1993), 4000-5000 t ha-1 of LWD were observed along the channel-riparian shrub contact zone compared with 5-40 t ha-1 within the forest. The mass and location of LWD differs to some extent between this river and other examples of -r 1050 m Motion reach (Lower Ain valley) Pradons reach (middle Ardáche valley) 350 lo 300 m Dd°J tonne Fig. 8. Debris jam mass distribution in the meander peduncules of Mollon (Ain) and Chauzon (Ardeche). 110 H. Piégay, A.M. Gurnell / Geomorphology 19 (1997) 99-116 Table 5 Mass of LWD observed at two sites Arděche (see also Fig. 3) on the Rivers Ain and Mollon (Ain) Total area sampled (ha) Total debris jam mass (t) Total debris mass (t • ha~') Number of debris jams surveyed Average debris jam size (t) Smallest debris jam observed (t) Largest debris jam observed (t) % debris jam mass located on the channel banks (% total) Debris line mass located on the channel bank (tonnes per linear metre of bank) 0.5 Chauzon (Arděche) 85.2 10.0 325.5 371.7 3.8 37.1 93 110 3.5 3.4 0.1 0.02 23.0 59.0 45.0 74.0 0.9 rivers of this geomorphic type (Table 4; Piégay and Bravard, 1997). 4.2.2. Wandering rivers In wandering rivers the sites for preferential deposition of woody debris are not within the main channel, but are situated along the adjacent flood-plain (Fig. 8, Table 5). The wood is deposited at the edges of the main channel and along axes of over-bank flows within the forest (Piégay, 1993), the mass of LWD deposited being much larger in the former location. The LWD does not form a debris line along the axes of overbank flows, but is preferentially deposited along flow axis concavities. Debris deposition occurs in the form of debris 100 80- ■ 60 40 20 0 1995 -+- -+- -+- 10 20 30 40 50 60 70 80 90 100 110 120 *1 pixel - 0.13 m2 Distance from the channel to the forest (m) Fig. 9. Upstream - downstream distribution of the surface occupied by LWD in the Mollon plot at the edge of a meander bend on the Ain River in 1992, 1993, 1994 and 1995 (from a GIS-based analysis of raster data). H. Piégay, A.M. Gurnell / Geomorphology 19 (1997) 99-116 111 Fig. 10. Canalisation of overbank flow by the debris line on the edge of the Mollon meander bend, River Ain, during a 1 in 7 year flood. accumulations or jams, that is heterogeneous accumulations of logs in transit, that progressively accumulate in response to debris inputs. These jams are not isolated but form a debris line which is more or less continuous according to the site. On the Chau-zon and Mollon sites (located on the Rivers Arděche and Ain), 74 and 45%, respectively, of the debris is located on the floodplain, amounting to 500 and 900 kg of debris per linear metre of channel. The total mass of debris is considerably higher than that observed in braided river courses. However, the amount varies from one meander bend to the next. Three factors provide an explanation for this variability: the degree of connectivity between the main channel and the floodplain, the planform of the meander, and the availability of dead wood immediately upstream of the meander bend. A four year study of a 4000 m2 area at the edge of a meander bend on the Ain has shown that the mass of debris remained approximately constant from one year to the next although the bank receded by 15m each year (Fig. 9). The concave meander site was a preferential depositional zone but was equally a debris source since, on average, 20 trees disappeared each year. The lines of woody debris were mobile, progressing in proportion to the migration of the bank. In these wandering systems, accumulations of dead wood have a morphological role. The distribution of debris is discontinuous and is canalised during overbank flows along preferential flow axes (Fig. 10). Marginal erosion occurs in these locations so that a single channel develops as a result of the coalescence of erosion scars. Dead wood, therefore, enhances the battering and erosion of the meanders of forested hydrosystems during overbank flows. These observations are different from those reported by Hickin (1984) and Keller and Swanson (1979), who worked on rivers of a similar geomorphic pattern. Hickin (1984) suggested that the debris line could stop the lateral migration of the channel, but we believe that the riparian morphodynamic reaction to LWD deposition depends on the supply rate, and thus human influence could explain the differences between our observations and those of Hickin. Keller and Swanson (1979) showed that an obstruction of the main channel could accelerate the process of cutoff. Such a process has not been observed on the River Ain, but the Ain is larger than the Squamish and is also probably subject to a more restricted 112 H. Piégay, A.M. Gumell / Geomorphology 19 (1997) 99-116 supply of LWD. The supply of less LWD and the predominantly smaller piece sizes make obstruction of large channels by LWD very unlikely in the European context. 5. A relationship between the geomorphic pattern of the river and the character of dead wood accumulations Our observations of the role of LWD on some European rivers illustrates that this role varies greatly between rivers of different size, power and geomorphic pattern. The size of the river channel plays a major role in the rate and form of LWD retention and so influences the speed of migration of the material. The riparian forest has greatest influence on small rivers firstly because it supplies LWD to an aquatic surface of limited extent, and secondly because the size of the pieces of wood may exceed the channel width in some cases, so encouraging debris jams to develop. The interaction between LWD accumulation, decomposition and transfer and a downstream continuum of increasing channel width is the key underlying the River Continuum Concept (RCC; Vannote et al., 1980) and is confirmed by our observations of LWD accummulations in small and medium-sized European rivers. In the small rivers wood accumulates largely where it falls because the power of the river is rarely sufficient to move the debris and because a large proportion of the debris exceeds the river width, making it difficult for the river to transport it. In medium sized rivers the debris is more readily moved because of the relatively higher stream power and the greater proportion of debris pieces which are less than the channel width. Thus in the Highland Water catchment, the proportion of partial dams were observed to increase and complete dams decrease with increasing channel size. In the larger European rivers studied, the small and restricted supply of debris pieces delivered from the riparian forests made complete obstruction of the channel by accumulations of LWD extremely unlikely. However, the location, structure and function of LWD accumulations associated with large rivers appears to be very variable and so does not always follow a simple longitudinal pattern as implied by the River Continuum Concept. This seems to reflect the retention capacity of the main channel rather than the size of standing trees and whether the trees are managed or unmanaged. It seems that wandering sections of river have a greater capacity to stop the migration of LWD than braided sections. The variability in the role of LWD can be illustrated at a number of spatial scales. For example, an individual tree trunk within a dynamic meander may be localised in its influence, whereas downstream of a trunk within a braided channel considerable volumes of dead wood can be readily retained. At the scale of the entire channel cross section, the weak overall retention capacity of braided channels in comparison with wandering channels of similar size, also illustrates that LWD dynamics in large rivers can be spatially very variable according to river pattern and may not always follow a downstream progressive continuum. There is a longitudinal discontinuity in the retention capacity, the speed of debris transfer and the local volumes of debris delivered to the river system according to the geomorphic pattern of the river. Furthermore, the amount of wood can actually be much greater in large rivers than in smaller ones when the large rivers are characterised by high lateral channel mobility and thus by very significant local inputs of dead wood. Based on our European examples, we have illustrated that in large piedmont rivers the functional and structural characteristics of dead wood differ according to the geomorphic pattern of the river. Distinct differences have been observed between braided and wandering river patterns and also between rivers of different power (specifically between intramontane and piedmont rivers). Importantly, the role of LWD in these larger European rivers has been noted to differ from that reported in North American studies. These differences are thought to be mainly attributable to the restricted supply rate and size of LWD pieces delivered from the managed riparian forests of the European rivers that have been investigated. In greater detail, the geomorphic pattern of a river channel integrates a range of different controlling factors which operate in both longitudinal and lateral directions to control the LWD mass in a river cross section. Fig. 11 synthesises these different factors, H. Piégay, AM. Gurnell / Geomorphology 19 (1997) 99-116 113 EXOGENIC FACTORS Potential supply from upstream reach or tributaries Stream power Mass of Large Woody Debris ENDOGENIC FACTORS BED CHARACTERISTICS MARGIN CHARACTERISTICS ECOTONE CHARACTERISTICS Sinuosity rate Overflow frequency (connectivity) Gravel bar? Lateral mobility of the channel Active channel width Biostructure type (tree or shrub species and ages, morphology of roots.) Cover type (trees. shrubs) Management strategies (clearing, snagging,,,) Micromorpholoav Influence type : - Retention capacity - CWD supply Fig. 11. Influences on the mass of LWD in river systems. both endogenic and exogenic, and illustrates the importance of channel size, bed morphology, and bank and riparian zone character. Geomorphic pattern reflects stream power which depends on two closely interelated primary variables changing in a downstream direction, the bankfull flow and the slope. In addition, a variable and more subtle explanation is required to explain the spatial variability in the functional and structural character of dead wood. The character of the LWD depends on local biomor-phological conditions associated with the vegetation (age, resistance of individual plants to flow and erosion) and the connectivity between the river and its floodplain which impacts upon the mobilisation and retention of wood and varies with geomorphic pattern of the river. The mobilisation of material is itself a function of the character of the riparian vegetation (age, degree of management) and the energy of the flow, which if it is strong is capable of fragmenting the vegetation mosaic and augmenting the input of wood. It is, therefore, necessary to take into account the transverse dynamics and locally diverse character of stretches and of processes, if one wishes to under- stand the geographical diversity of the functional and structural characteristics of LWD. 6. What conclusions can be drawn for woody debris management? LWD plays a critical geomorphological and ecological role in pristine temperate rivers flowing through forested river corridors (Triska, 1984; Sedeli and Luschessa, 1982). It has also been shown to be important in arid (Minckley and Rinne, 1985) and tropical rainforest (Spencer et al., 1990) river systems. Therefore, the reappearance of LWD in heavily-modified European river systems can be seen as an indicator of the restoration of rivers to a more "natural" state. Indeed, agricultural change in Europe may lead to increased woodland cover along river corridors and thus an increase in the delivery of LWD to river systems. Environmentally-sensitive strategies are required for LWD management in European rivers. Dead wood has long been recognised as a risk factor in the context of flooding and erosion along 114 H. Piégay, AM. Gurnell / Geomorphology 19 (1997) 99-116 Table 6 Ecological, hydraulic and morphological impacts a of LWD according to river size and pattern of European rivers (a qualitative and comparative assessment based upon the observations presented in this paper) Small- Large Piedmont rivers medium rivers braided wandering Creation of terrestrial habitat - + + + Creation of aquatic habitat + + + + Increase in inundation + - = frequency Diversification of the flood- = — + + plain mosaic Increase in bank erosion ----- — + + Increase in sedimentation + + + + Level of impact: + + very significant; + significant; = moderate; — weak;-----very weak. man-modified river courses. Since the twelfth century, French legislation has recommended preservation of the free flow of water in river channels through the maintenance of riparian vegetation and river bed clearance. Similar long-term maintenance of rivers has also occurred in the United Kingdom (Brookes, 1988). However, as has been suggested for Danish rivers (Iversen et al., 1993), it is now necessary to redefine river maintenance strategies for both ecological and economic reasons, including the development of guidelines for the management of LWD. Dead wood is associated with variable scenic/aesthetic, ecological and morphological impacts and differing hydraulic risk according to the geomorphic river type (Table 6). Maintenance decisions within a particular geomorphic type should be based on three groups of factors: erosion/flooding vulnerability; ecological criteria; and economic criteria. In general, the ecological gains represented by LWD retention (Table 6) and the significant costs involved in removal imply that minimal removal is the best strategy. Thus the fine-tuning of LWD management strategies is of considerable economic and ecological interest. A primary intervention logic can be based upon river size, where maintenance is focused on small streams where the length of trunks is frequently greater than the channel width and thus the hydraulic impact of LWD is at its greatest. Here removal of large debris pieces at locations where flooding has significant economic consequences is highly cost-effective. However, retention of LWD debris elsewhere is not only of direct ecological and economic benefit, but it provides a buffer to slow movement of debris pieces rather than allowing them to move freely downstream to accumulate at more flood-sensitive sites. In larger rivers LWD is unlikely to create major obstructions to flow and so does not represent a significant influence on floodplain inundation frequency except where structures such as narrow bridge openings and weirs impede the transfer of the debris. At these locations removal is highly beneficial to Fig. 12. Detail of a main debris jam located on the margin of the channel in the active band of the Drôme river, downstream of Crest. H. Piégay, A.M. Gumell/ Geomorphology 19 (1997) 99-116 115 flood alleviation, but elsewhere debris should be left because its environmental benefits and high removal costs heavily outweigh any small hydraulic benefits (Table 6). In particular, LWD has a very feeble hydraulic influence within braided systems but has considerable significance for both aquatic and terrestrial habitat diversity. For example, a vegetation maintenance strategy has been proposed recently on the Drôme River (Piégay and Landon, 1997). Three classes of reach have been distinguished based on a maintenance gradient from no intervention to riparian and bed clearing. The sectorisation takes into account potential risks and the ecological interest of each reach. In the most ecologically interesting reaches, it is proposed that debris jams are left in the channel (Fig. 12). Acknowledgements J.P. Bravard, P.G. Salvador and, particularly, N. Landon are thanked for their help in the field and their constructive criticisms of the text. Agence de ľEau Rhône-Mediterranée-Corse are gratefully acknowledged for funding the research on the French Rivers. The assistance of Richard Davis, Neil Harris, Kevin Hewtson, Chris Hill and Stephen Tooth in carrying out surveys of woody debris accumulations in the Highland Water catchment is also gratefully acknowledged. The comments of two anonymous referees were very valuable in improving this paper. References Bisson, P.A., Bilby, R.E., Bryant, M.D., Dolloff, CA., Grette, G.B., House, R.A., Murphy, MX., Koski, K..V. and Sedeli, J.R., 1987. Large woody debris in forested streams in the Pacific Northwest: Past, present and future. In: E.O. Salo and T.W. Cundy (Editiors), Streamside Management: Forestry and Fishery Interactions, Proc. Symp., 1986, Seattle, WA. College of Forest Resources, Univ. of Washington, Contrib. 57, pp. 143-190. Brookes, A., 1988. Channelized Rivers: Perspectives for Environmental Management. Wiley, Chichester. Chastan, B., Guard, O. and Lavabre, J., 1993. Les difficultés ďestimation rapide des debits observes en crue: exemple de la crue de 1'Ouvěze du 22 septembre 1992. Rev. Géogr. Lyon, 68: 153-158. Church, M., 1992. Channel morphology and typology. In: P. Calow and G.E. Petts (Editors), The Rivers Handbook, Vol. 1. Blackwell, London, pp. 126-143. Gregory. K.J. and Davis, R.J., 1992. Coarse woody debris in stream channels in relation to river channel management in woodland areas. Regul. Rivers Res. Manage., 7: 117-136. Gregory, K.J. and Davis, R.J., 1993. Debris dams in river channels: to clear or not to clear? Geogr. Rev., 6: 2-6. Gregory, K.J., Gurnell, A.M. and Hill, CT., 1985. The permanence of debris dams related to river channel processes. Hydrol. Sei. J., 30: 371-381. Gurnell, A.M. and Gregory, K.J., 1995. Interactions between semi-natural vegetation and hydrogeomorphological processes. Geomorphology, 13: 49-69. Gurnell, A.M., Gregory, K.J. and Petts, G.E., 1995. The role of coarse woody debris in forest aquatic habitats: implications for management. Aquat. Conserv. Mar. Freshwater Ecosyst., 5: 143-166. Hickin, E.J., 1984. Vegetation and river channel dynamics. Can. Geogr., 28: 111-126. Iversen, T.M., Kronvang, B., Madsen, B.L., Markmann, P. and Nielsen, M.B., 1993. Re-establishment of Danish streams: restoration and maintenance measures. Aquat. Conserv. Mar. Freshwater Ecosyst., 3: 73-92. Keller, E.A. and Swanson, F.J., 1979. Effects of large organic debris on channel form and fluvial process. Earth Surf. Process., 4: 361-380. Landon, N., Piégay, H. and Bravard, J.P., 1995. Mission d'exper-tise gémorphologique pour le compte de la Commission Locale de l'Eau du SAGE Drôme: Propositions pour une gestion physique équilibrée du lit de la Drôme. Commission Locale de l'Eau du SAGE, 105 pp. Laws of Sewers, 1732 (Second Edition). Printed by E. and R. Nutt, and R. Gosling for W. Sears, at the Lamb on Ludgate Hill. Leopold, L.B. and Wolman, M.G., 1957. River channel patterns: braided, meandering and straight. U.S. Geological Survey Professional Paper 282-B, pp. 39-85. Malanson, G.P. and Butler, D.R., 1990. Woody debris, sediment, and riparian vegetation of a subalpine river, Montana, USA. Arct. Alp. Res., 22: 183-194. Maser, C and Sedeli, J.R., 1994. From the Forest to the Sea: the Ecology of Wood in Streams, Rivers, Estuaries, and Oceans. St Lucie Press, Delray Beach, FL, 200 pp. Minckley, W.L. and Rinne, J.N., 1985. Large woody debris in Hot-Desert streams: an historical review. Desert Plants, 7: 142-153. Nakamura, F. and Swanson, F.J., 1993. Effects of coarse woody debris on morphology and sediment storage of a mountain stream system in western Oregon. Earth Surf. Process. Land-forms, 18: 43-61. Petts, G.E., 1990. Forested river corridors: a lost resource. In: D. Cosgrove and G.E. Petts (Editors), Water, Engineering and Landscape; Water Control and Landscape Transformation in the Modern Period. Belhaven Press, London, pp. 12-34. Piégay, 1993. Nature, mass and preferential sites of coarse woody debris deposits in the Lower Ain valley (Mollon reach) France. Regul. Rivers Res. Manage., 8: 359-372. 116 H. Piégay, A.M. Gurnell / Geomorphology 19(1997) 99-116 Piégay, H. and Bravard, J.P., 1997. The reactions of a mediterranean riparian forest to a major hydrological event, the 1 in 400 year flood (22.09.1992) in the Ouvěze river, Drôme-Vaucluse, France. Earth Surf. Process. Landforms, in press. Piégay, H. and Maridet, L. (1994) Formations végétales arborées riveraines des cours d'eau et potentialités piscicoles (revue bibliographique). Bull. Fr. Péche Piscicult., 333: 125-147. Piégay, H. and Landon, N., 1997. To promote an ecological management of riparian forests; the Drôme river case, France. Aquat. Conserv. Mar. Freshwater Ecosyst., submitted. Platts, W.S., Armour, C, Gordon, D., Bryant, M., Bufford, J.L., Cuplin, P., Sherman, J., Lienkamper, G.W., Monsen, G.W., Nelson, R.L., Sedeli, J.R. and Tuhy, J.S., 1987. Methods for evaluating riparian habitats with application to management. Gen. Tech. Rep. Int. 221. USDA Forest Service, Intermoun-tain Research Station, Ogden, UT, 177 pp. Robison, E.G. and Beschta, R.L., 1990. Coarse woody debris and channel morphology interactions for undisturbed streams in southeast Alaska, USA. Earth Surf. Process. Landforms, 15: 149-156. Sedeli, J.R. and Froggatt, J.L., 1984. Importance of streamside forests to large rivers: the isolation of the Willamette River, Oregon, USA, from its floodplain by snagging and streamside forest removal. Verh. Int. Ver. Limnol., 22: 1828-1834. Sedeli, J.R. and Luschessa, K.J., 1982. Using the Historical Record as an Aid to Salmonid Habitat Enhancement. In: E. Neil and B. Armantrout (Editors), Proceedings of a Symposium on Aquisition and Utilization of Aquatic Habitat Inventory Information, Portland, OR, pp. 210-223. Sedeli, P.A., Bisson, P.A., Swanson, F.J. and Gregory, S.V., 1988. What we know about large trees that fall into streams and rivers. In: G. Maser, R.F. Tarrent and J.F. Franklin (Editors), From the Forest to the Seas: a Story of Fallen Trees. Gen. Tech. Report No. 229. USDA Forest Service, Pacific Northwest Station, Portland, OR, pp. 47-81. Sedeli, J.R., Richey, J.E. and Swanson, F.J., 1989. The river continuum concept: a basis for the expected behaviour of very large rivers? In: D.P. Dodge (Editor), Proceedings of the International Large River Symposium. Can. Spec. Publ. Fish. Aquat. Sei., 106: 49-55. Spencer, T., Douglas, I., Greer, T. and Sinun, W., 1990. Vegetation and fluvial geomorphic processes in south-east Asian tropical rainforests. In: J.B. Thornes (Editor), Vegetation and Erosion. Wiley, Chichester, pp. 451-469. Sullivan, K., Lisle, T.E., Dolloff, CA., Grant, G.E. and Reid, L.M., 1987. Stream channels: the link between forests and fishes. In: E.O. Salo and T.W. Cundy (Editors), Streamside Management: Forestry and Fishery Interactions, Proc. Symp., 1986, Seattle, WA. College of Forest Resources, University of Washington, pp. 39-97. Triska, F.J., 1984. Role of wood debris in modifying channel geomorphology and riparian areas of a large lowland river under pristine conditions: A historical case study. Verh. Int. Ver. Limnol., 22: 1876-1892. Vannote, R.L., Minshall, G.W., Cummins, K.W., Sedeli, J.R. and Cushing, C.E., 1980. The river continuum concept. Can. J. Fish. Aquat. Sei., 37: 130-137.