Channel response to multiple damming in a meandering river, middle and lower Aragón River (Spain)

Channel response to multiple damming in a meandering river, middle and lower Aragón 1 River (Spain) 2 3 4 Ibisate, A.1; Díaz, E.1; Ollero, A.2; Acín, V.3 & Granado, D.3 5 6 1 Dpt. Geography, Prehistory and Archaeology, University of the Basque Country, UPV/EHU 7 c/ Tomás y Valiente s/n, 01006 Vitoria-Gasteiz, Spain 8 9 2 Dpt. Geography and Land Management, University of Zaragoza 10 c/ Pedro Cerbuna s/n, 50009 Zaragoza, Spain 11 12 3 ECOTER S.C. Ecology and Land, Zaragoza, Spain 13 14 askoa.ib[email protected], Tlf. +34945013996 Fax +34945013309 15 16 17 18 Abstract 19 20 Small hydropower plants (SHP) affect river flow and sediment transport, and thus, impact river 21 morphology. Eight hydropower schemes were studied along the meandering middle and lower reaches of 22 Aragón River (Spain) to assess their effects on channel morphology and sediment dynamics from 1927 to 23 2010. GIS tools were used to measure changes in fluvial surfaces, channel planform and lateral and 24 vertical dynamics. Three periods (early, middle and late 20th century) were analysed to discern the effects 25 of the main pressures, such as changes in land use, large reservoirs upstream, and SHP. Results were 26 combined with field and topographical measurements and hydrological analysis. 27 28 Active channel width and channel migration suffered a clear reduction in the whole period. They started 29 as a consequence of land cover changes in the drainage basin, but their speed increased after a large 30 reservoir was built upstream. More recent changes occurred since most of the SHP were put into 1 operation in the 1990's, especially in their short-circuited reaches and in the four more downstream ones. 2 These changes are interpreted as a consequence of reduced discharge, transitory sediment trapping, and 3 reactivation of sediment transport after weirs became filled as well as by the impact of flood hydrology. 4 5 6 7 8 Keywords: channel adjustment; human impact; reservoirs; dams; small hydropower station; fluvial 9 geomorphology; meandering river, Aragón River 10 11 12 13 Introduction 14 15 Water discharge, especially high flows, and sediment supply drive fluvial dynamics and thus determine 16 river channel form (Simon, 1989; Downs, 1995; Ibisate et al. 2011). Therefore, human activities affecting 17 discharge or sediment supply have important impacts in river morphology and adjustments (Kondolf, 18 1997; Rinaldi & Simon, 1998). These impacts include changes in land use (García-Ruiz et al., 1997 19 Liébault & Piégay, 2002), reservoirs (Petts, 1979; Brandt, 2000), weirs (Gurnell, 1997), gravel mining 20 (Petit et al., 1996; Kondolf, 1994), or mining waste (Knighton, 1991). Additionally, other in-channel 21 actions as channel modifications (Simon & Rinaldi, 2000), or channelizations (Brookes, 1988) directly 22 modify channel form and thus river ecosystems. 23 24 Small hydropower plants (SHP, here those smaller than 50 MW) typically divert part of the water, use it 25 to produce electricity, and revert it to the river downstream. They can thus impact river hydromorphology 26 by affecting water discharge and sediment transport, and lead to imbalances in river morphogenetic 27 processes. The number of SHP has increased worldwide as a consequence of the need for renewable 28 energy sources (Alonso-Tristán et al., 2011). Hydropower activities in fluvial systems modify natural 29 processes, features and forms in rivers (Sear, 1995; Kondolf & Batalla, 2005; Csiki & Rhoads, 2010). 30 Modifications caused by hydropower activities occur not only at the site of the infrastructures, but also 1 upstream and downstream (Baker et al., 2011). 2 3 Although there is a fairly comprehensive information on the morphological impacts of hydropower 4 associated to large dams (Rollet, 2007), much less is known about the impacts of weirs (Degoutte, 2006; 5 Couvert, 1999) and SHP on hydromorphological and ecological functioning (Warner, 2012; Malavoi, 6 2003, 2009). Some studies explore the ecological effects of small hydropower plants on 7 macroinvertebrate communities as a consequence of river habitat fragmentation (Fu et al., 2008; Cortes et 8 al. 1998), or on fish (Santos et al., 2006; Hatfield et al. 2007; Lewis et al., 2004). SHP seem to have a 9 smaller impact than that of large reservoirs, but because many rivers are affected by multiple SHP, their 10 cumulative effect could be important (Yang et al., 2011). Therefore, for river managers it is important to 11 define the hydromorphological effects of these infrastructures and to perform restoration plans that could 12 mitigate or recover fluvial dynamics. 13 14 In this study we analyse temporal changes in channel form, in an attempt to distinguish the morphological 15 adjustments caused by different factors that are governing fluvial processes: land use, large reservoirs and 16 several SHP on a Spanish river, the Aragón. 17 18 Study site 19 20 The Aragón River is a 190 km-long tributary of Ebro River that drains an 8609 km2 basin in the north of 21 the Iberian Peninsula (Fig. 1). The Yesa reservoir, located in the middle reach of the Aragón, has a strong 22 effect on water and sediment discharge downstream (López-Moreno, 2006; López-Moreno et al., 2003). 23 The study area, located downstream from this reservoir, presents a gravel-bed meandering channel. 24 According to Díaz & Ollero (2005), this 88 km-long section can be divided into four functional sectors, 25 sensu Amoros & Petts (1993), differing in valley setting, planform and slope. The first sector, from Yesa 26 dam to the confluence with the Irati River, has a slope of 0.3%, a sinuosity index of 1.6, an average daily 27 discharge of 44.9 m³/s, and a restricted floodplain excavated in its own terraces. The second sector 28 extends from the confluence with the Irati River to Gallipienzo. Its mean daily discharge is around 64 29 m³/s (Peralta et al., 2003), the slope is 0.1% and the sinuosity index 1.4, but runs through a wide valley, 30 up to 1600 m, which allows lateral mobility in the floodplain. The third sector goes from Gallipienzo to 1 Murillo, and the river, with an estimated mean daily discharge of 65 m³/s, runs through a narrow valley, 2 150-320 m wide, of tertiary conglomerates. The fourth and last sector goes from Murillo to the 3 confluence with the Arga River, has an average annual daily discharge of 68.6 m³/s, a slope of 0.1%, and 4 meanders freely through a floodplain wider than 3 km. 5 6 Fig. 1. Study site area and location of the hydropower stations within the Aragón River basin 7 8 The river has been long affected by impacts both on the drainage basin and in the channel. Since the 9 beginning of the 20th century many defences (groynes, riprap and bank structures) have been constructed 10 along the river to reduce channel mobility and bank erosion and to protect from floods. These 11 infrastructures encouraged human activities, mainly agriculture, in the floodplain. Additionally the river 12 has been affected by two reservoirs, Yesa (446.9 hm³) built in 1959 and Itoiz (417 hm³) built in 2004 in 13 the Irati River, one of the main tributaries of the Aragón (Fig. 1). The drainage area of Yesa is around 14 25% of the total of Aragón basin while in Itoiz is a 5.9%, and both reservoirs are used mainly for 15 irrigation purposes. These reservoirs, but especially Yesa, which is older, have drastically altered the 16 dynamics of the Aragón River, as a result of reduced discharge (Fig. 2A) and sediment trapping (López-17 Moreno; 2006, Acín et al., 2011). From 1913 to 1956 the river still presented an almost natural regime, 18 with a mean discharge of 76.7 m³/s and highest discharges during March. Mean daily discharges 1 decreased from 1959 to 2000 and especially during late winter and spring, which also affected to the 2 highest flows that shifted to January. The annual mean discharge was 64.2 m³/s. 3 Table 1. Main characteristics of studied SHP (U: upstream; S: short-circuited; D: downstream) in the 4 middle and lower reaches of Aragón River. Mean annual discharges are estimations based on the 5 discharges upstream and flow data of arriving tributaries. (Data source: Ebro Water Agency and own 6 data) 7 Hydropower station Construction date Functional reaches Mean annual Q (m³/s) Maximum Q diverted in canal (m³/s) Weir height (m) Reaches length (m) U S D Sangüesa I 1968 (reformed in 1981) 1 44.9 14 2.5 1132 1948 536 Sangüesa II 1967 (reformed in 1986) 2 64 16 3 825 132 1012 Cáseda 1927 2 64 8 2 1125 4000 950 Gallipienzo 1919 (reformed in 1982) 3 65 40 5 1683 669 1051 Murillo 1995 4 68.6 60 5 1408 2243 1718 Santacara 1994 (reformed in 2005) 4 68.6 70 2 1146 4202 1447 Mélida 1995 4 68.6 77 2 489 3595 1715 Caparroso 2000 4 68.6 70 3 1069 3222 1128 8 Moreover, 8 small hydropower plants (Fig. 1, Table 1) have been built along 85 km of the middle and 9 lower reaches of the Aragón River, which also abstract water and modify flowing discharges in the short-10 circuited reaches. Some of these plants were built in the early 20th century, but were reformed and made 11 more effective at trapping water in the 1980's and 1990's. The rest were built in the last decade of the 20th 12 century. From 2000 to 2010 the discharge decreased considerably to a 42.9 m³/s of mean annual discharge 13 for the total amount of both the diverted and the short-circuited flows and a 22.7 m³/s when only the 14 short-circuited reach was considered (Fig. 2A). Moreover flows present a very high variability linked to 15 hydropower requirements, which can be higher than 15 m³/s in half an hour (Fig. 2B). Nevertheless, 16 unlike in the large reservoirs, most of the water diverted by the hydropower schemes returns downstream 17 to the river. 18 19 Fig. 2. Hydrology of the Aragón River at Caparroso. A) Mean daily discharge for three periods: 1913-20 1959 (before Yesa reservoir), 1959-2000 (after Yesa reservoir) and 2000-2010 (after Caparroso SHP). B) 21 Example of daily discharge variation during 24 hours on 1st of February of 2008 (Data source: Ebro 1 Water Agency). 2 3 Seven of these SHP are diversion power plants. The meanders are cut by means of a weir that diverts part 4 of the water into a bypass channel, so called the diversion channel. The water is returned downstream into 5 the natural channel, so called the short-circuited channel (Fig. 3). Only in one case, Sangüesa II, the 6 hydropower plant is a run-of-river power plant with no diversion channel. Thus, in most hydropower 7 schemes three reaches can be differentiated: upstream from the weir; the short-circuited reach (Stroffek et 8 al., 1996); and downstream from the confluence with the returning bypass the river receives again the full 9 water discharge. The length of the upstream and downstream reaches was defined from the comparison of 10 aerial photographies before and after the construction or reconstruction of the weir and field 11 reconnaissance in different flow conditions; upstream was defined by the dam effect created by the weir 12 through the delimitation of the pool, the disappearance of riffles and the change in the width of the water 13 surface, whereas the downstream reach was delimitated by the end of the effect of the rapid formed by the 14 water released from the artificial channel. The dimensions of the reaches thus varied from one 15 hydropower station to the other, depending on weir size, the diverted discharge and the characteristics of 16 the river channel and hydropower station. The upstream reach ranged from 825 m in Sangüesa II to 1683 17 m in Gallipienzo. The downstream reach length ranged from 536 m in Sangüesa I to 1718 m in Mélida 18 (Table 1). 19 20 Fig. 3. Location of the four small hydropower plants of the lower sector of the Aragón River. The three 21 reaches and their influence length are marked. 22 23 The effect of the four hydropower stations settled in the lower reach of Aragón River is especially 24 relevant (Fig.3). All four were constructed in the mid 1990’s (except Mélida the rest were built starting 25 from old weirs), have a similar discharge concession, around 70 m³/s (Table 1) and affect forty eight 1 percent of the lower Aragón River reach taking into account the dam effect upstream, the short-circuited 2 reach, and the downstream influence. For this reason a deeper analysis has been performed on this lower 3 reach of the river. 4 5 6 Data and Methods 7 8 We followed a stepwise procedure to analyse the geomorphologic effects of hydropower plants in the 9 Aragón River. First, based on aerial photographs, we analysed the temporal changes in channel 10 dimensions, riparian surface and planform in the reaches affected by all 8 SHP. Then, we complemented 11 the study with field surveys and hydrological analyses. Finally a deeper study was performed in the 4 12 downstream SHP. 13 14 The variations in active channel and riparian surface were studied by comparing aerial photographs and 15 orthophoto maps using GIS tools. Three aerial flights of 1927, 1957, 1992 and an orthophoto map of 2010 16 were used. Of those collected, 1927 is the first aerial photography available at a scale approximately of 17 1:9000; the flight of 1957, at a scale of 1:33000, was taken just after the Yesa reservoir was built; the 18 1992 flight, at a scale of 1:8000, was taken just before the construction of most of the hydropower 19 stations; and the orthophoto of 2010, at a scale of 1:5000, was the last available. The aerial photographs 20 of 1957 were already georeferenced by the Government of Navarre. We corrected geometrically the aerial 21 photographs of 1927 and 1992 by means of the second-order polynomial transforming technique. In each 22 image and average of 12 permanent ground control points were matched, spatially well-distributed but 23 with special concentration in the features of interest. The average root mean square (RMS) error for the 24 1927 photography was 8.3 m and 6.6 m for the 1992 one. All the images were rectified using the cubic 25 convolution resampling (Chuvieco, 2002). 26 27 Four basic fluvial features lying within the fluvial corridor were identified to measure variations in active 28 channel and riparian surface: water surface, sediment bars (unvegetated bars which are submerged at high 29 discharges), pioneer vegetation (bars with pioneer vegetation or terraces with grass), and riparian forest 30 (depositional forms covered by riparian forest and inundated only during the largest floods) (Surian, 1 1999, Comiti et al., 2011). Both water surface and unvegetated sediment bars were considered as the 2 active channel, and therefore taken as a single feature to analyse their evolution. These four 3 morphological features were identified through the visual analysis of aerial photographs and the latest 4 orthophoto map. The surfaces of each morphological feature and their evolution were delimitated using 5 GIS tools for the three reaches of each hydropower station: upstream, short-circuited and downstream. In 6 the four stations located in the lower reach of Aragón River (Murillo, Santacara, Mélida and Caparroso) 7 we calculated for the four years, 1927, 1957, 1992 and 2010, the average active channel width (Surian, 8 1999; Rinaldi, 2003) for each 500-m reach, and considered the width at bankfull stage (Parker et al., 9 2007; Zanoni et al., 2008). Additionally the eroded and constructed surfaces (Piégay et al., 2005) were 10 calculated with GIS for three periods 1927-1957, 1957-1992 and 1992-2010. These mobilized surfaces 11 were measured considering the modified surface and channel migration between the active channel in the 12 aerial photography at the beginning of the period and the position of the active channel at the final of the 13 period. Thus, we identified the surfaces gained (erosion areas, sensu Ghoshal et al., 2010) and lost 14 (deposition areas) by the lateral shift of the river. The results were expressed in ratios considering the 15 changes in surfaces in relation to the length of each reach of all SHP. 16 17 The comparison of topographic maps of 1998 and 2008, at a scale of 1:5000, allowed the analysis of river 18 bed elevation for the four more downstream SHP. Contour lines location was compared in both maps and 19 distances calculated. Longitudinal topographic profiles were done in the reaches upstream from the weir 20 by means of a topographic level (error 0.02 m) to analyse the filling of the river bed of sediments. 21 Besides, an exhaustive field reconnaissance of the channel on both river banks was carried out, looking 22 for evidences of channel incision (exposed roots, exposed foundations in bridges or weirs, river bed 23 excavation, lineal and upstream erosion symptoms) or aggradation (buried elements, buried base of 24 riparian trees) (Surian, 1999; Rinaldi, 2003). 25 26 The hydrological impact of hydropower stations was studied based on the data collected at the gauging 27 station of Caparroso, which continuously measures discharge at both the derived (bypass channel) and the 28 short-circuited reach (natural river channel). The discharge data started in 1912 and the derived discharge 29 started to be measured when the hydropower station came into operation in 2000. Thus all the period 30 analysed by the aerial photographs can be covered and in addition 10 years of both natural river (short-1 circuited) flows and diversion data are available to assess the hydrological alterations deriving from these 2 infrastructures. The impact of hydropower activity was analysed for both mean annual discharges and 3 flood hydrology. The number of days in which the maximum diversion discharge is exceeded was 4 counted for the four more downstream SHP. Using Gumbel's law probability distribution for the gauging 5 station of Caparroso, we calculated the 5and 10-year return-period flood discharges (Q5 and Q10), and 6 Q2, considered the channel-forming discharge (Wolman & Miller, 1960; Gregory & Park, 1974). 7 8 Critical discharge for bedload entrainment (Schoklitsch, 1950) was calculated based on the grain size for 9 the short-circuited reach of each hydropower station. The frequency of days above the critical discharge 10 for bedload entrainment was analysed for the data provided by the gauging station of Caparroso. The 11 number of days that this discharge is exceeded was counted both in the short-circuited channel and the 12 total discharge (the sum of the diversion discharge and the discharge flowing through the short-circuited 13 reach) to evaluate the impact of the diversion. 14 15 16 Results 17 18 Changes in active channel and riparian surfaces and channel adjustments 19 20 The Aragón River became narrower from 1927 to 2010: the surface of the active channel decreased and 21 the surface of riparian forests increased, as can be appreciated in the examples for Mélida and Murillo 22 hydropower stations (Fig. 4). This trend was more evident in the downstream SHP (Murillo, Santacara, 23 Mélida and Caparroso), where the average channel width decreased in 79% (Fig. 5). This overall trend 24 can be split in three periods (1927-1957, 1957-1992, and 1992-2010), according to the main pressures in 25 the basin. 26 1 From 1957 to 1992 a relevant stabilization and reduction of lateral movement was appreciated in all 2 reaches (Fig. 8), with residual erosion areas contrasting with the wide deposition areas. In average 9, 8 3 and 14 ha/km were deposited in their upstream, short-circuited and downstream reaches respectively, 4 while only 0.5, 0.3 and 0.5 ha/km were eroded. Again, the reaches initially most prone to migrate 5 (Santacara, Cáseda, Caparroso and Mélida) suffered the highest stabilisation rates. In all cases the 6 deposition of surfaces dominated. From 1992 to 2010 deposition decreased considerably compared to 7 previous periods, as only 0.9, 1.4 and 2.3 ha/km were average deposited in their upstream, short-circuited 8 and downstream reaches respectively, while erosion increased with 0.9, 0.9 and 1 ha/km eroded 9 respectively. All the SHP constructed in that period showed a drastic reduction of migration rates 10 irrespective of reach type (Fig. 8). 11 1 1 Fig. 8. Evolution of erosion and deposition rates (ha/km), for the upstream, short-circuited and 2 downstream reaches in all SHP for the 1927-1957, 1957-1992 and 1992-2010 periods. Erosion ratios are 3 given in negative values and deposition ratios in positive values. 4 5 Comparison of the 1998 and 2008 topographical maps showed that in the upstream reaches contour lines 6 shifted downstream (2832 m in Santacara and 123 m in Mélida), thus showing aggradation of the river 7 bed (Table 2). This process was confirmed by field surveys (Fig. 9), although in this case the average 8 length of the aggraded reach was 200 m. Mélida, the most recent SHP, was relatively less filled. All 9 short-circuited reaches presented an upstream shift of their contour line (703 m in Murillo, 776 m in 1 Santacara, 963 m in Mélida and 265 m in Caparroso), showing an incision process. Finally, in the 2 downstream reach the response was an aggradation for Santacara (123 m of shift downstream) and 3 incision in Mélida (912 m of shift upstream). The Santacara downstream reach matches with the upstream 4 reach of Mélida, being influenced by the weir of Mélida, in process of being filled. 5 6 Table 2. Shift in meters of the contour line within each reach from the comparison of topographical 7 cartography of 1998 and 2008 for Murillo (MUR), Santacara (SAN), Mélida (MEL) and Caparroso 8 (CAP) SHP. The lack of data means that no contour line crosses those reaches. 9 MUR SAN MEL CAP Upstream + 2832 +123 Shortcircuited - 703 -776 -963 -265 Downstream +123 -912 10 11 12 Fig. 9. Longitudinal profiles for the upstream reaches. Distances measured from the weir in an upstream 13 direction (m). 14 15 Field evidences (exposures in roots, bridges and other infrastructures) showed an incision ranging from 16 0.5 to 2 m in the short-circuited reaches (Table 3), progressive erosion from the weir and upstream 17 erosion starting from the confluence of the diverted waters. Only 8% of the evidences were detected 18 below the confluence, and no evidence of incision occurred upstream from the weir. The deepest incision, 19 deeper than 1.75 m, occurred in Gallipienzo, the oldest SHP, whereas the smallest incision occurred in 20 Sangüesa II, the only run-of river SHP. 21 Table 3. Average incision in meters in the short-circuited reaches affected by hydropower stations (SAN 1 I: Sangüesa I; SAN II: Sangüesa II; CAS: Cáseda; GAL: Gallipienzo; MUR: Murillo; SAN: Santacara; 2 MEL: Mélida; CAP: Caparroso). 3 SAN I SAN II CAS GAL MUR SAN MEL CAP 1 0.7 1 1.75 0.9 1.3 0.9 1.1 4 5 Hydrological changes 6 7 Mean daily discharges decreased in the three periods analysed (Fig. 2). At the short-circuited reach in 8 Caparroso the general pattern of the hydrological regime was maintained, but the average discharge 9 decreased in 45% (from 40% in wet years to 52% in dry ones). The percentage of water diverted varied 10 throughout the year from 34% in October to 62% in November (Fig. 10). River discharge exceeded the 11 water concession of SHP only in 39, 37, 31 and 27 days for Murillo, Santacara, Mélida and Caparroso 12 respectively. 13 14 Fig. 10. Hydrology in Caparroso. A) Short-circuited and diverted mean annual discharge (m³/s). B) Flood 15 event in April 2007. C) Peak flow series from 1913 to 2011. D) Days above critical discharge. Data for 16 the short-circuited reach are compared to total discharge since 2000 when Caparroso SHP started running 17 (Data source: Ebro Water Agency). 18 1 Peak flow series at Caparroso gauging station (Fig. 10) showed a reduction of the magnitude of floods 2 after Yesa reservoir was built. Q2, Q5 and Q10 discharges also decreased (Table 4), although these results 3 present low significance due to broad confidence intervals. Since Caparroso SHP started operating, 4 changes were small, and limited to lower magnitude events. Q2 discharges decreased in the three periods, 5 being this decrease slightly more accentuated for the last period. However, Q5 and Q10 of total discharge 6 increased between 1957-1999 and 2000-2011. Meanwhile, in the short-circuited reaches Q5 decreased 7 whereas Q10 increased. Therefore, the comparison between the discharges flowing these days through the 8 short-circuited section and the total discharge, when the diverted discharge is added, showed scant 9 differences, which were even smaller as the discharge increased. Peak flow series also showed 10 moderately large floods in 2003, 2007 and 2010, with 928, 1027 and 797 m³/s respectively. 11 12 Table 4. Q2, Q5 and Q10 discharges and confidence intervals for each period at Caparroso gauging station: 13 before reservoirs and SHP (1913-1956), since Yesa (1957-1999) and after Caparroso SHP. 14 Q 2 (m³/s) 95 % Confidence interval Q 5 (m³/s) 95 % Confidence interval Q 10 (m³/s) 95 % Confidence interval 1913-1956 592 ± 83 1096 ± 148 1275 ± 203 1957-1999 549 ± 55 718 ± 99 836 ± 136 2000-2011(short-circuited discharge) 434 ± 161 698 ± 314 877 ± 441 2000-2011 (total discharge) 483 ± 160 752 ± 312 930 ± 439 15 16 The critical discharge for bedload entrainment (276.7 m³/s, and 1.05 yr return period) was exceeded only 17 a few days per year, differences between the short-circuited reach and the total discharge being minor 18 (Fig. 10). Critical discharge evolution throughout all the data series showed a reduction of the number of 19 days since the mid 90’s, with a slight difference between the total and short-circuited discharges. 20 21 22 Discussion 23 24 The main morphological changes in the Aragón River consisted on channel narrowing, decreased lateral 25 movement, and reduced surface of water and sediment bars, and an increase in the cover of riparian 26 forests. Similar evolution has been described in other rivers of the Ebro basin (Martínez-Castroviejo et al. 27 1991; Beguería et al. 2006; Ollero, 2010; Martín-Vide et al. 2010), what suggests a regional trend towards 1 reduced river dynamism. Discharge and flood intensity have steadily decreased in the Ebro basin during 2 the last decades together with an increase of the contribution of low flows to the total annual discharge 3 (Beguería et al. 2003; López-Moreno et al. 2006), despite no consistent trend in the distribution of 4 precipitation (González Hidalgo et al., 2009; De Luis et al., 2009; Lorenzo, 2012). Reduced discharge and 5 flood intensity has been thus attributed to changes in land cover due to reforestation and abandonment of 6 farmland (López-Moreno et al. 2008), which also resulted in decreased sediment supply (López-Moreno 7 et al. 2003; García-Ruiz et al. 2010). Additional factors influencing reduced sediment supply in the Ebro 8 basin include gravel mining (Granado et al., 2008; Martín-Vide et al. 2010), and reservoirs (López-9 Moreno 2006). 10 11 Although river dynamism decreased in the Aragón River at least since 1927, when the first aerial 12 photographs were taken, the rate of decrease has varied, and differences between periods can shed light 13 on the main causes underlying this trend. Between 1927 and 1957 the width of the active channel 14 decreased along the studied area, whereas the pioneer vegetation and riparian forests increased. 15 Nevertheless, channel migration was significant, with important areas of erosion and deposition. As no 16 large reservoirs existed in the basin during this period, the reduced dynamism must be attributed mainly 17 to changes in land cover, especially increase in forested areas. Forest cover usually results in low 18 discharge and decreased sediment yields, and reforestation has been shown to reduce the dynamism of 19 rivers elsewhere (Liébault & Piégay, 2002). Bank protection, which started in 1924 (Jimeno, 1996) and 20 use of riparian areas for agriculture could have played an additional role. The only two SHPs in operation 21 during this early phase were Gallipienzo and Cáseda, but their impact on river dynamism seem to have 22 been minor, as their evolution was very similar to the rest of the reaches studied. This lack of impact 23 suggests that sediment transport was little affected by their dams, probably because during this period 24 large floods were still frequent and the sediment yield was high enough to fill the low dams in a short 25 period of time. 26 27 From 1957 to 1992 the trend towards reduced river dynamism accelerated, with a loss of active channel 28 area of 1.7 ha/km/y, when it had previously been of 1.1 ha/km/y. This acceleration coincides with the 29 building of the Yesa reservoir, which started operating in 1959 and has an impoundment runoff index (IR) 30 (i.e., reservoir capacity respect to mean annual runoff) of 0.3. Many papers have described a reduction of 1 river dynamism as a consequence of the construction of large reservoirs (e.g., Graf, 2006; Gordon & 2 Meentemeyer, 2006). Two are the main effects of large reservoirs on dynamism: sediment starvation and 3 modification of the flood regime (Kondolf, 1997; Kondolf & Batalla, 2005). During 1960-85 the Yesa 4 reservoir lost a 4.4% of its storage capacity as it accumulated 20.67 km³ of sediments (0.82 km³ per year, 5 López-Moreno, 2006). Nevertheless, it seems that the main effect of the Yesa reservoir in our study sites 6 is due to a reduction of daily flows and flood frequency. This effect of reduced flood frequency is 7 consistent with the decrease in erosional areas measured after the building of the Yesa reservoir along our 8 study reaches. It is worth noting that since the middle of the 20th century the decrease of river dynamism 9 was slower upstream from Yesa reservoir. There the active channel was reduced 0.12 ha/km/y between 10 1946 and 1977 and 0.15 ha/km/y from 1977 to 2003 (Granado & Ollero, 2008). 11 12 From 1992 to 2010 river dynamism continued to decrease along the study area, but especially so in the 13 lower Aragón sector, where new SHPs were built. There, active channel width reduction increased 14 progressively downstream, from 19% in Murillo to 59% in Caparroso. This increase seems not to be a 15 result of a cumulative effect of SHPs on river dynamism, as progressive active channel reduction affects 16 mainly short-circuited reaches, not so clear in downstream ones. Vadnais et al. (2012) also found that in a 17 complex systems dams and reservoirs do not presented a cumulative effect on fluvial morphology. 18 However, a cumulative effect was validated in Gállego River due to dams and gravel extractions (Martín 19 Vide et al. 2010). We can only speculate about the factors responsible of this progressive change, as we 20 are not aware of any gradual change in environmental conditions between Murillo and Caparroso, apart 21 from increasing distance from the Yesa reservoir. Downstream reaches adjustments seem to be related 22 with the local conditions of the valley characteristics, as more inserted channels (Murillo and Mélida) 23 show a more moderate width reduction. Contrasting with this trend of reduced river dynamism, erosion 24 increased during this period, probably as a result of several flood events that were registered, especially in 25 2003, 2007 and 2010. Channel erosion usually is associated to large floods (Kochel, 1988, Turowski et 26 al., 2012), so the occurrence of these events might be the responsible of the increased erosion during this 27 period. 28 29 Fieldwork in all SHP and analysis of topographic maps in the four lowermost SHP to measure changes in 1 river bed elevation yielded generally consistent results, and showed aggradation of the river bed in 2 upstream reaches as a consequence of sediment retention in the dams, and degradation in short-circuited 3 reaches. Low dams trap sediments, and thus, can result in downstream sediment starvation and channel 4 incision (García-Ruiz & Puigdefábregas, 1984; Kondolf, 1997; Doyle et al., 2005). Nevertheless, in 5 absence of active management their effects are short-lived: as the dam becomes filled sediment transport 6 is resumed downstream (Brandt, 2000; Baker et al., 2011; Csiki & Rhoads, 2010) 7 8 In summary, SHP in the middle and lower Aragón River seem to accentuate a long trend to reduce river 9 diversity and lost of geomorphological value of the free meandering river planform, derived from changes 10 in land uses and from the building of the large Yesa reservoir. Although the time of operation of most 11 SHP is still too short to discern the magnitude of the changes, we expect sediment starvation to be 12 temporary as the dams fill up. The impacts of altered hydrology will last for as long as the SHP are in 13 operation, but it is no clear how long will it take for the river to adjust to the new conditions. Furthermore, 14 changes continue to occur in the basin, such as the building of the Itoiz reservoir, and thus it is expected 15 that the River Aragón will continue to change as a result of the highly dynamic landscape. 16 17 Acknowledgments 18 This study is based on the results obtained in the project titled “Efectos de las minicentrales 19 hidroeléctricas en los sistemas fluviales de los ríos Aragón y Arga (tramos medio y bajo): diagnóstico, 20 tendencias y propuestas de gestión” (Effects of the small hydropower stations in the fluvial systems of 21 Aragón and Arga Rivers (middle and lower reaches): diagnosis, tendencies and management proposals). 22 We want to thanks to Gestión Ambiental de Navarra, S.A. for the funding of this research, to Josu Elso, to 23 Camino Jaso for their cooperation in the project and all the researchers that collaborated in the 24 geomorphological study (Javier Fuentes, Laura E. Gonzalo, Daniel Ballarín, Daniel Mora, Rafael 25 Sánchez, Iban Sánchez, Horacio García, Sergio Domenech). Finally we would like to thank the Spanish 26 Ministry of Economy and Competitiveness for its support through the project SCARCE (Consolider-27 Ingenio 2010 CSD2009-00065). The authors also acknowledge support the Consolidated Research Group 28 of the Basque Government IT 288-07, the Research and Formation Unit “UFI11/09. Cuaternario: 29 Cambios Ambientales y Huella Humana” and the Research Group of the Aragón Regional Government 30 "Clima, Agua, Cambio Global y Sistemas Naturales". We also want to thank Kelly Janes from the 1 University of California, Berkeley for the review of the text in English, Arturo Elosegi for his useful 2 comments and the review of the English and Hervé Piégay for his relevant advices and comments as well 3 as the anonymous reviewers that helped to an important improvement of the paper. 4 5 References 6 7 Acín, V., E. Díaz, D. Granado, A. Ibisate, A. & A. Ollero, 2011. Cambios recientes en el cauce y la 8 llanura de inundación del área de confluencia Aragón-Arga (Navarra). Geographicalia 59-60: 11-25. 9 Amoros, C. & G.E. Petts, coords., 1993. Hydrosystèmes fluviaux. Masson, Paris. 10 Baker, D.W., B.P. Bledsoe, B.P., C.M. Albano & N.L. Poff, 2011. Downstream effects of diversion dams 11 on sediment and hydraulic conditions of Rocky Mountain streams. River Research and Applications 27: 12 388-401. 13 Beguería, S., J.I. López-Moreno, A. Lorente, M. Seeger & J.M. García-Ruiz, 2003. Assessing the Effect 14 of Climate Oscillations and Land-use Changes on Streamflow in the Central Spanish Pyrenees. Ambio 15 32(4): 283-286. 16 Beguería, S., J.I. López-Moreno, A. Gómez-Villar, V. Rubio, N. Lana-Renault & J.M. García-Ruiz. 2006. 17 Fluvial adjustments to soil erosion and plant cover changes in the central Spanish Pyrenees. Geografiska 18 Annaler 88 A (3): 177-186. 19 Brandt, S.A., 2000. Classification of geomorphological effects downstream of dams. Catena 40: 375-401. 20 Brookes, A., 1988. Channelized Rivers: Perspectives for Environmental Management. John Wiley & 21 Sons Ltd. 342 p. 22 Chuvieco, E., 2002. Teledetección ambiental. La observación de la tierra desde el espacio, Ariel. 586 p. 23 Comiti, F., M. Da Canal, N. Surian, L. Mao, L. Picco, M.A. Lenzi, 2011. Channel adjustments and 24 vegetation cover dynamics in a large gravel bed river over the last 200 years. Geomorphology 125: 147-25 159. 26 Cortes, R., M.T. Ferreira, S.V. Oliveira & F. Godinho, 1998. Contrasting impact of small dams on the 27 macroinvertebrates of two Iberian mountain rivers. Hydrobiologia 389 (1-3): 51-61. 28 Couvert, B., P. Lefort, J.L. Peiry, P. Belleudy, 1999. La gestion des rivières. Transport solide et 29 atterrissements. Guide méthodologique. Les études des agences de l’eau nº 65. Agence de l’Eau Rhône-30 Méditerranée-Corse, Paris. 31