Artificial steps to stabilize mountain rivers: a post-project ecological assessment

RIVER RESEARCH AND APPLICATIONS

River. Res. Applic. 25: 639–659 (2009) Published online 4 February 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rra.1234

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS: A POST-PROJECT ECOLOGICAL ASSESSMENT F. COMITI,a* L. MAO,a,b M. A. LENZI a and M. SILIGARDI c a

Department of Land and Agroforest Environments, Universita` di Padova, Italy b Department of Geography, University of Hull, UK c Agenzia Provinciale per la Protezione dell’Ambiente, Trento, Italy

ABSTRACT This paper compares the ecological response of reaches treated with traditional (check dams, CDs) and morphologically based (artificial steps, ASs) grade-control works installed in a mountain river, where an unmodified reach provided reference conditions. Coarse particulate organic matter (CPOM) retentiveness was measured at the reach scale using artificial leaf release experiments. Macroinvertebrate sampling was carried out by a Surber net at pool- and run-type morphological units within the three reach types, and hydraulic and sediment characteristics were measured at each Surber sampling site. In addition to calculating a variety of parameters to describe hydrodynamic and macroinvertebrate characteristics within reaches and morphological units, biotic (Extended Biotic Index, IBE) and ecological (Fluvial Functionality Index, IFF) index scores were also calculated. Results averaged at the reach scale show that CPOM retention, macrobenthos taxa richness and diversity were higher in the presence of a natural bed morphology, whereas both types of grade-control works alter CPOM dynamics and macroinvertebrate communities. However, ASs showed CPOM retention dynamics and macroinvertebrate richness and diversity that were closer to the unmodified reach than CDs. Analysis at the unit scale demonstrated strong links between hydraulic variables and macroinvertebrate parameters such as abundance, diversity and proportion of shredders (SHR). Overall, ASs seem to provide a good trade-off between the need to limit channel incision whilst maintaining aquatic ecosystems. Copyright # 2009 John Wiley & Sons, Ltd. key words: check dams; fluvial morphology; bed incision; macroinvertebrates Received 9 December 2008; Accepted 16 December 2008

INTRODUCTION Most of the mountain rivers in the European Alps feature control structures (check dams (CDs), bed sills, ripraps, sidewalls) which are meant to render the streambed and the banks stable even during infrequent, high-energy flood events. Steep (i.e. mean slope S > 2–3%) channels have a tendency to erode and incise because of their large sediment transport capacity compared to the available sediment supply (e.g. Montgomery and Buffington, 1997; Mao and Lenzi, 2007). Such natural dynamics conflict with human occupation of valley bottoms, and have forced local people to attempt to stabilize mountain channels for centuries. However, the physical instability of channel beds has been recognized to be the most important driver of fluvial ecosystems in the Alps (Petts et al., 2000). Therefore, a true river ecological restoration (see e.g. Kondolf, 1995; Palmer et al., 2005) would imply the reestablishment of destabilizing transport processes which are not acceptable from a social (i.e. river stakeholders) perspective. High-gradient streams often exhibit a natural step pool architecture, which likely represents a self-adjustment towards higher channel stability (Abrahams et al., 1995; Lenzi et al., 2006a; Church and Zimmermann, 2007) because most flow energy is dissipated in hydraulic jumps (i.e. spill resistance) at each step pool unit (Wilcox et al., 2006; Comiti et al., 2007). The same physical process has long been ‘utilized’ in the construction of staircase-like sequences of grade-control works such as CDs to inhibit bed incision in steep channels (Comiti et al., 2005), but ‘classic’ CDs are high (>2 m, but up to 10 m) concrete structures which cause unnatural impoundments and channel *Correspondence to: F. Comiti, Department of Land and Agroforest Environments, Universita` di Padova, Italy. E-mail: [email protected]

Copyright # 2009 John Wiley & Sons, Ltd.

640

F. COMITI ET AL.

widening upstream, severe disruption of the longitudinal river continuum (sediments, organic matter, biotic communities; Fu¨reder et al., 2002, Monaghan et al., 2005) and a strong visual ‘artificialization’ of the mountain landscape. Indeed, the most relevant source of environmental degradation in mountain channels comes from their physical alteration rather than from chemical pollution, which has a more notable impact on lowland rivers (Giller and Malmqvist, 1998). Such adverse physical environmental impacts have led over the past two decades to the substitution of stone ramps for CDs in several stream ‘restoration’ projects. In addition, recent torrent control projects across the Alps have tended to implement lower CDs than previously, sometimes including bed sills built with boulders and logs to emulate a more natural morphology. These new interventions are fundamentally different from placements of logs, boulders or steps to enhance habitat diversity in rivers degraded by human impacts, as is commonly done in the Pacific Northwest of North America and in Northern Europe (e.g. Hilderbrand et al., 1997; Linlokken, 1997; Roni et al., 2005a, 2005b, 2006; Chin et al., 2007). In steep Alpine Rivers, artificial boulder steps have the primary function of grade-control structures and are thus designed to be stable even during intense flood events (recurrence interval, RI, up to 50–100 years). The adoption of artificial steps (ASs) designed according to more natural morphological criteria, besides giving an improved visual impact over traditional structures, has proved to be cost-effective (Lenzi, 2002) and sufficiently stable (Lenzi and Comiti, 2003; Lenzi et al., 2004). Even though these works cannot be viewed as restorative interventions in a truly ecological sense, it is reasonable to hypothesize that they have less adverse impacts on the fluvial ecosystem than traditional high, widely spaced CDs because their dimensions and spacing resemble more closely a natural stepped channel. However, to the Authors’ knowledge, a comparison of the ecological impacts due to traditional and morphologically based grade-control structures has not been undertaken so far, hampering an integrated appraisal of their overall performance that combines engineering, socio-economic and ecological aspects (Palmer et al., 2005). Indeed, post-project monitoring activities, even though widely viewed as crucial to improve river restoration practice (Roni et al., 2005a), are rarely carried out. The purpose of the present paper is a comparative assessment of the ecological response to traditional (concrete CDs) and morphologically based (ASs with cemented boulders) grade-control works installed in a high-gradient mountain river (Maso di Spinelle, Italian Alps) more than a decade ago. Engineering and socio-economic advantages of the latter solution in the study river have already been established (Lenzi, 2002; Lenzi and Comiti, 2003), whereas its ecological benefits are still unknown. If these can be determined, bearing in mind problems and uncertainties of post-project assessment (Roni et al., 2006), the choice of ASs might be advocated on a sounder ecological basis.

METHODS Study basin and channel The Maso di Spinelle is a perennial boulder-bed tributary of the Brenta River (Autonomous Province of Trento, Northeastern Italy, Figure 1) whose characteristics are reported in Table I. Precipitation in the basin occurs mainly as snowfall from November to April, with runoff being dominated by snowmelt in May and June. Annual peak discharges are usually associated with cyclonic storms in early autumn. Forests (mostly conifers, Picea abies and Larix decidua) cover 53% of the basin. Riparian vegetation consists of alder (Alnus incana) and willow (Salix spp.) mixed with spruce. Basin geology comprises metamorphic (schists, gneiss) and intrusive (granite) rocks. Quaternary deposits dominate along the channel, with talus often resting on moraine and fluvioglacial sediments rich in very large boulders (intermediate diameter >3 m). Some of these boulders induce flow diversions during large flood events, leading to bank failures. In the 1980s, following major damage during an extreme (recurrence interval RI > 100 years) flood event in 1966, when the streambed incised by 10–15 m and widened from 10 to 50 m, 13 reinforced concrete CDs were built along the most degraded channel reach (hereafter called the CD reach, Figures 1 and 2a). CD height and spacing are approximately 8 and 90 m, respectively, with a mean channel slope (crest-to-crest) of about 10%. The bed slope on the ‘step tread’ between two consecutive structures is about 2%. Bankfull channel width is 18 m. In 1996 and 1997, after an intense (RI
50–75 years) flood event, further grade-control works were deemed necessary upstream of the CD reach. About 30 cemented boulder steps were built along a 400 m-long segment Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS

641

Figure 1. The lower section of the Maso di Spinelle River (Eastern Italian Alps, Italy), locating the three different reaches analysed (U, unmodified reference reach; AS, artificial steps reach; CD, concrete check dams reach). The river flows from north to south. This figure is available in colour online at www.interscience.wiley.com/journal/rra

(hereafter called AS reach, Figures 1 and 2b). A morphological criterion incorporating bed grain size distribution and channel slope was adopted to design step height and spacing (Lenzi, 2002), following the flow resistance maximization proposed by Abrahams et al. (1995). The control works successfully resisted an intense flood event in 1998 (RI ¼ 20–25 years), with only minor bed adjustments (Lenzi and Comiti, 2003). Bankfull width here is approximately 27 m, and ASs vary in height between 1 and 2.5 m, are spaced 10–24 m apart with a crest-to-crest channel slope ranging from 11 to 21%. The bed slope between ASs is around 3–5% but where steps are closely spaced the alluvial bed can display a reverse slope (i.e. pools occupy most of the spacing length). Just upstream of the AS reach, a channel segment not affected by control works (hereafter called the unmodified, U, reach, Figures 1 and 2c) remains and is used as reference for natural bed conditions in the present study. The Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

642

F. COMITI ET AL.

Table I. Main characteristics of the study basin (Maso di Spinelle) Characteristics Catchment area (km2) Mean annual precipitation (mm) Minimum elevation (m.a.s.l.) Maximum elevation (m.a.s.l.) Length of the main channel (km) Mean channel slope (%) Bed morphology

Value 45 1200 909 2561 10.0 13.8 Cascade, step pool

Figure 2. Views of the study reaches in the Maso River: (a) the check dams (CD) reach; (b) the artificial steps (AS) reach; (c) the reference unmodified (U) reach featuring cascade and step pool morphologies. This figure is available in colour online at www.interscience.wiley.com/ journal/rra Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS

643

channel in this reach is dominated by a cascade morphology (sensu Montgomery and Buffington, 1997) with occasional step pool units stretching across the entire cross-section. The abundance of large boulders leads to frequent high steps (up to 1.5–2 m) as well as to a chaotic bed arrangement which makes the flow very turbulent and aerated. Bankfull width is approximately 14 m, and channel slope is 10%. The ecological effects of the two different types of grade-control works (i.e. CDs and ASs) were evaluated both at the channel reach and channel unit scales, based on investigations of CPOM (i.e. coarse particulate organic matter) retentiveness (reach scale) and macroinvertebrate communities (reach and unit scale) in relation to physical (i.e. hydraulic, morphological and sedimentological) characteristics. Two indices (biotic and ecological) were also estimated at the reach scale. Physical and ecological measurements are described below separately for each scale of analysis. Hydraulics, sediments and channel geometry measurements Channel reach scale. Discharge Q and reach-averaged flow velocity V were measured by the salt dilution (Elder et al., 1991; D’Agostino, 2004) and salt tracer (Calkins and Dunne, 1970; Comiti et al., 2007) methods, respectively. Two conductivity metres with data loggers were placed at the upstream and downstream ends of the study reaches. At least three slugs of salt mixture were injected for each flow measurement and the average values were then used as Q and V values. Longitudinal profiles (Figure 3) and cross-sections (one for each segment, Figure 4) for the CDs and the unmodified reaches were surveyed by a laser distance metre with clinometer. A more detailed topographic survey was already available for the AS reach (Lenzi and Comiti, 2003) because it had been the subject of a detailed analysis of flow velocity and flow resistance (Cattelan, 2005; Lenzi et al., 2006b). As a result, 16 cross-sections were available. The representative cross-sections presented in Figure 4 were selected to depict the ‘average’ channel geometry, avoiding deeper pools and wider/narrower sections. Several morphological parameters describing CD/step and pool dimensions were calculated from the longitudinal bed profiles (Table II). Bed surface sediments at the reach scale were also analysed based on pebble counts (sample size 100–200 pebbles) obtained using an integrated grid-by-number sampling scheme (Bunte and Abt, 2001) for all the study reaches.

Figure 3. Longitudinal profiles of the study reaches, with pools indicated by horizontal lines. This figure is available in colour online at www.interscience.wiley.com/journal/rra Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

644

F. COMITI ET AL.

Figure 4. Representative cross-sections of the study reaches. Continuous and dashed lines represents low flow and bankfull stage, respectively. Cross-sections were surveyed in locations without large pools and/or channel constrictions. This figure is available in colour online at www.interscience.wiley.com/journal/rra

Table II. Main characteristics of the study reaches Characteristics

Reach length (m) Reach slope (%) Slope between control structures (%) Channel width (bankfull, m) Channel width (low flow, m) Surface grain size D50 (m) Surface grain size D84 (m) Step/check dam distance (m) Average step drop (m) Average pool depth (m) Average pool length (m) Total pool length/reach length Macroinvertebrates sampling sites CPOM tests IBE (Extended Biotic index) IFF (Fluvial Functionality Index)

CD

AS

U

CDP

CDNP

ASU

ASD

93 10 2 18 16.5 0.05 0.20 90 7.9 2.0 9.8 0.1 — 3

63 2 2 18 16.5 0.05 0.20 — — — — — 6 3

51 11 4.5 27 22.5 0.08 0.25 13 0.79 1.0 4.9 0.4 6 3

33 21 3.5 27 17.5 0.14 0.32 11 2.14 1.2 6.9 0.6 — 3

Y Y

Y Y

110 10 10 14 12 0.34 1.16 5 0.5 0.4 4.4 0.4 6 3 Y Y

CD, check dams (CDP crest-to-crest thus including a pool, CDNP begins downstream of the pool); AS, artificial steps (ASU upstream sub-reach, ASD downstream sub-reach); U, unmodified reference reach. For reaches CD and AS, size of steps and pools refer to the artificial ones (i.e. concrete or boulder structures) only. Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS

645

Channel unit scale. At each macroinvertebrate sampling site (see channel unit scale macroinvertebrate sampling for sampling design), flow velocity, water depth and surface grain size measurements were taken using a 1D electromagnetic current metre, a stadia rod and a calliper, respectively. Three measurements of all variables were taken at each replicate macroinvertebrate sampling site. Stream-wise velocity (averaged over 10 s) was measured at 0.2 h (ub, near bed velocity) and at 0.8 h (us, near surface velocity), and the arithmetic mean was used as the profile mean flow velocity u. The intermediate axis of all surface clasts > 4 mm was measured within and adjacent to the Surber area (see later) for each site (range: 8–46 particles, average: 22 particles), and presence of finer gravel or sand was noted. Ecological measurements Channel reach scale. Channel retentiveness of CPOM was estimated using leaf release experiments (Lamberti and Gregory, 1996; Siligardi et al., 2000; Lepori et al., 2005). The test procedure was applied to morphologically homogeneous reaches and involved the sudden injection of 500 Gingko biloba leaves, a tree species that is not present in the study area. Before injection into the upstream end of each test reach, the leaves were rehydrated for 15–20 min in a bucket of stream water. Following release, Gingko leaves were counted at fixed time intervals (1, 2, 3, 4, 7, 10, 15, 20, 30 min) by 4–5 operators located 2–3 m apart across the downstream cross-section of the reach. The total leaves counted were subtracted from 500 to estimate total leaf retention in the reach. CPOM tests were performed during summer low flows in 2006 and 2007 (Panazzolo, 2006; Checchinato, 2008). One reach was analysed in the unmodified (U) segment, encompassing the sites where macroinvertebrates were sampled. Within the AS reach, upstream and downstream sub-reaches of contrasting AS geometry were analysed (ASU, ASD, Figure 3). In the channel segment controlled by CDs both crest-to-crest (i.e. including pool, CDP) and pool-to-crest (i.e. no pool, CDNP) CPOM measurements were carried out in order to discern the role of the large pools present below each structure. Macroinvertebrates were sampled within the unmodified (U) segment, the upstream artificial step reach (ASU), and in the pool-to-crest components of the CD reach. Suitable sites were not available to allow Surber sampling in the ASD reach or in the deep pools of the CD reach. The IFF (Fluvial Functionality Index, Siligardi et al., 2003), which is derived from the Riparian Channel Environmental Inventory (RCE, Petersen 1992), was estimated for each of the three reaches. The IFF is based on 14 parameters related to four broad aspects of river ecosystems: land use and riparian vegetation, morphological structure of the banks, riverbed structure and biological characteristics. The final score (ranging from 14 to 300, with higher values associated with higher functionality) is classified into five categories, from I (high functionality) to V (poor functionality). Channel unit scale. Macroinvertebrates were collected during low flows in July–August 2005 using a 500 mmSurber net (area 0.1 m2) within two types of morphological unit, runs and shallow pools, identified from water depth and flow type. Pools were characterized by flow deceleration with three dimensional turbulent eddies generated by oncoming jets, whereas runs featured a smoother water surface where flow acceleration dominated (Wilcox and Wohl, 2007). It was not possible to sample steps and deep (>0.4 m) pools using the Serber net. Three different sampling sites were investigated for each unit type (run, shallow pool) at each reach type (CD, AS, U). Three replicate samples were taken within each sampling site (Panazzolo, 2006) and were combined to characterize the macroinvertebrate community for that sampling site. Macroinvertebrates were identified to the taxonomic level required by the IBE method (Extended Biotic Index; Ghetti, 2001, see next section), i.e. genus for Plecoptera, Ephemeroptera, Odonata, Tricladia, Hirudinea and family for Trichoptera, Coleoptera, Diptera, Eteroptera, Crustacea, Gasteropoda, Bivalvia, Oligochaeta, Nematoda. Data analysis Physical variables. At the reach scale, the longitudinal distance between the two conductivity metres was used to determine V as travel distance over travel time (Comiti et al., 2007). Resistance to flow was quantified by the Darcy–Weisbach friction factor f (¼8 ghmS/V2), where S is the bed slope, g is the acceleration due to gravity and hm is the reach-averaged flow depth calculated from continuity (i.e. hm ¼ Q/(wV) where w is the channel width). Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

646

F. COMITI ET AL.

At the unit scale, using measurements of velocity and flow depth, the local Froude number, indicating the relative kinetic energy of the flow at each site (Kemp et al., 2000), was calculated as Fr ¼ u/(g h)0.5, where g is the acceleration due to gravity and h is water depth. Also, a parameter similar to the grain Reynolds number and so related to the characteristics of the flow around sediment particles near the bed, was calculated as Re ¼ (ub d)/n, with n being the kinematic viscosity of the water, and d the mean diameter of surface particles at each site. CPOM retention. In addition to estimating the final number and percentage of retained CPOM, the leaf counting method allows the establishment of time-retention curves, which give insights into the dynamics of retentive bed structures, using the following theoretical asymptotic model (Siligardi et al., 2000): y¼

at 1 þ bt

(1)

where y is the cumulative number of exported leaves, t the time elapsed from the moment the first leaf crosses the measuring section and a and b are coefficients estimated using the least-squares method. The average number of exported leaves at each time interval (i.e. 1, 2, 3, 4, 7, 10, 15, 20, 30 min) from the three tests were used to determine the retention curve for each studied reach. Exported leaf numbers were normalized by reach length to allow interreach comparisons per 100 m of streambed length. Average leaf velocity Vl was calculated from the instant of injection to the time of peaking export rate at the counting section. Macroinvertebrates samples. Macroinvertebrate communities were investigated both at channel unit (using data from each sampling site, 3 reaches, 2 channel unit types and 3 samples per unit type: total 18 sites), and at reach scales (by treating the 6 sites from each reach as a single sample). Macroinvertebrate abundance A (number of individuals per m2 of streambed area) was estimated as the total number of individuals N divided by the Surber-sampled area. The IBE (the Italian version of the Extended Biotic Index, Woodwiss, 1978; Ghetti, 2001) was calculated for the reach-scale communities. The method highlights modifications of macroinvertebrates communities by pollution and environmental alterations with reference to expected natural communities, and is calculated using the presence/ absence of sensitive macroinvertebrates taxa and the total number of taxa. IBE scores range from 14 to 1, with lower values indicating an increased degree of alteration), and IBE classes range from I (i.e. no pollution) to V (i.e. heavy pollution). Diversity indices (see Magurran, 2004) were calculated for each of the 18 samples (the number of taxa (s), Margalef’s richness index (m ¼ (s  1)/ln(N)), Shannon’s entropy (H ¼ S((ni/N)ln(ni/N), where ni is the number of individuals for each taxon), and Buzas and Gibson’s evenness index (Buzas and Hayek, 1996, E ¼ eH/s)). Other parameters describing the composition of the community within each sample were also derived, including EPT N and EPT s, which represent the proportion of the total number of individuals and the total number of taxa, respectively, belonging to the Ephemeroptera–Plecoptera–Tricoptera groups; and the proportion of shredders (SHR) out of the total number of individuals for each sample (SHR), based on the functional classification reported in Ghetti (2001). Statistical analysis. Non-parametric methods (Spearman’s rank correlations with reach hydraulic variables; Mann–Whitney U and Kruskal–Wallis tests for between-reach comparisons) were used to analyse CPOM retention (the 30 min number of leaves exported) due the small sample size available (three tests for each of the CDNP, CDP, ASD, ASU and U sub-reaches). Also at the reach scale, differences in macroinvertebrate communities were evaluated using similarity indices (i.e. Sørensen, Jaccard, Sokal and Sneath, see Magurran, 2004) based on presence/absence data. Differences in community diversity at the reach scale were tested by the Shannon H t-test (Poole, 1974). Non-metric multidimensional scaling (NMS, Kruskal, 1964) coupled with analysis of similarities (ANOSIM, Clarke, 1993) were used to analyse the composition of macroinvertebrate communities (i.e. number of individuals for each taxa) at the unit scale. The measure of similarity used for both calculations was the Morisita’s index, as recommended by Krebs (1989). Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

647

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS

Physical and ecological variables relative to the 18 macroinvertebrate sampling sites were preliminarily tested for normality using the Shapiro–Wilk test. To test for difference in physical and ecological variables between reaches, between units and for reach  unit interactions, analyses of variance (one-way and factorial ANOVAs) were performed. Finally, multivariate relationships among physical and ecological variables were explored using Pearson’s correlations and a principal components analysis (PCA) ordination analysis. All analyses were conducted using STATISTICA ver.7.1 (Statsoft Inc., Tulsa, US), except for calculation of diversity indices and ANOSIM which were conducted using PAST ver.1.82 (Hammer et al., 2001).

RESULTS Analysis at the channel reach scale CPOM retentiveness. Large variations in CPOM retentiveness, expressed as the final number of exported leaves adjusted to a channel length of 100 m, were found both within and between sub-reaches (Figure 5), but a Kruskal– Wallis test indicated a significant difference between sub-reaches ( p < 0.10), with the unmodified reach (U) displaying the lowest median value (highest CPOM retentiveness), closely followed by the CD reach including a large pool (CDP). When the large pool was excluded (CDNP), the number of exported leaves increased significantly (Mann–Whitney U test, p < 0.05), indicating high leaf retention in the large pool. The AS sub-reach with higher and closer steps (ASD) showed the lowest CPOM retention capacity.

Figure 5. Measured values of the total number of exported leaves per 100 m of channel length within 30 min. Three separate tests were performed for each sub-reach. This figure is available in colour online at www.interscience.wiley.com/journal/rra

Table III. CPOM retention model (Equation (1)): the coefficients a and b, determined by the least squares method, and their ratio (a/b), which indicates the theoretical asymptotic value Sub-reach CDNP CDP ASD ASU U

a

b

a/b

V

f

Vl

505.6 173.6 446.6 386.4 78.9

2.0 0.9 0.9 1.30 0.39

247.4 182.0 471.1 297.9 203.9

0.36 0.43 0.27 0.33 0.45

2.6 7.2 54.1 17.7 11.9

0.42 0.58 0.17 0.33 0.64

Reach-average flow velocity V and average leaf velocity Vl are expressed in m s1. The Darcy–Weisbach friction factor ( f ¼ 8ghmS/V2), which represents the hydraulic resistance to flow, is also reported. Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

648

F. COMITI ET AL.

Flow velocity and resistance also varied considerably among the sub-reaches (Table III). Spearman’s correlations indicated an inverse relationship between average flow velocity V and the total number of exported leaves (R ¼ 0.85, p < 0.05), whereas average leaf velocity Vl (Table III) was positively correlated with V (R ¼ 0.87, p < 0.05), with Vl > V in all the reaches except ASD (Table III). A strong inverse correlation between exported leaf number and leaf velocity was also observed (R ¼ 0.88, p < 0.05). The Darcy–Weisbach friction factor was inversely correlated with Vl (i.e. leaves move slower with increasing total channel roughness, R ¼ 0.70, p < 0.05), but was not significantly correlated with exported leaf numbers. Other correlations between CPOM retentiveness and sedimentological/morphological parameters (Table II) of the study reaches were not significant ( p > 0.10). Cumulative exports of leaves through time in each sub-reach described different trajectories (Figure 6a) according to the retentiveness of the corresponding reach. The estimated coefficients (a, b) of the CPOM retention curves (Equation (1)) for each sub-reach and their ratio (a/b), which is the theoretical leaf export asymptote, are reported in Table III. Differences in shapes of dimensionless curves obtained by normalizing the curves to their asymptotic values (Figure 6b), give clues regarding the CPOM retention process. Three types of trend are apparent (Figure 6b). The unmodified reach (U) features the lowest initial export of leaves and then a regular increase with

Figure 6. Time-retention curves in absolute (a) and relative (b) terms, built putting together the three tests for each sub-reach. In the latter values are normalized by the final (i.e. at 30 min) exported number of leaves. The check dam sub-reach without pool (CDNP) features the faster initial export of CPOM, in sharp contrast to the unmodified bed geometry (U). Artificial steps (ASD and ASU) display an intermediate behaviour, similar to the check dams with pool (CDP) Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

649

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS

90% of total leaf export achieved within 40% of the time required to reach the asymptotic state. The CD sub-reach without a large pool (CDNP) shows the fastest export of leaves in the early stages of the experiments with 90% of final leaves exported in 10% of the time. Between these extremes are sub-reaches characterized by relatively large pools formed below ASs and CDs (CDP, ASU, ASD), with 90% of total leaf export achieved in approximately 25% of the time required to reach the asymptote. Macroinvertebrate communities and ecological indices. A complete list of the macroinvertebrate taxa found in the three reaches is presented in the Appendix. Macroinvertebrate communities do not vary dramatically among the analysed reaches, with sensitive taxa typical of unpolluted, well-aerated flows (e.g. Perla, Epeorus, Rhytrogena) present in unmodified and modified segments. However, one taxon of Plecoptera (Dinocras), two taxa of Trichoptera (Psycomyidae and Odontoceridae) and one taxon of Diptera (Dixidae) were collected only in the unmodified reach, and the CD reach was the only segment lacking the reophilic Blephariceridae and Athericidae (Diptera) while hosting Ephemerella (Ephemeroptera), a taxon requiring finer sediments (i.e. sand and silt). The ASs reach featured an intermediate taxa composition. There was a higher similarity (SSør) between the ASs and unmodified reaches (SSør ¼ 0.84) than between the CD and unmodified reaches (SSør ¼ 0.75), with an intermediate level of similarity between CD and AS (SSør ¼ 0.80). Similar results were obtained with other similarity indices (Jaccard, Sokal and Sneath). Table IV summarizes macroinvertebrate indices obtained at the reach scale for the three main reach types (CD, AS, U). As previously noted in the methods section, because the macroinvertebrate classification was based on the IBE protocol (Ghetti, 2001), taxa richness and diversity indices can be used only for relative comparisons, not for species-based community analysis. Macroinvertebrate abundance was highest in the ASs and lowest in the unmodified reach, with the CDs segment featuring a slightly higher value than the latter. However, the unmodified reach featured the highest number of taxa, the highest Margalef’s richness and Shannon’s diversity, but the lowest evenness. The CDs exhibited the highest evenness and the lowest number of taxa and Margalef’s index. The ASs had an intermediate value for most of the indices, but displayed an evenness value very close to the reference reach. However, differences in diversity between reaches were not statistically significant ( p > 0.10, Shannon H t-test). The IBE (Table IV) assigned all reaches to the highest water quality category (class I), although the IBE score decreased from the unmodified through the AS to CD reach. The IFF index varied remarkably among the three reaches (Table IV), with the unmodified segment being the only one to attain class I, followed by the AS (class II) with the CD reach characterized by lower IFF values (classes II–III), largely as a result of the simplified crosssection geometry and bed morphology. Analysis at the channel unit scale Physical characteristics of channel units. The only physical characteristic of the 18 macroinvertebrate sampling sites (Table V) found not to be normally distributed ( p < 0.05) was Re, which was normalized by a logtransformation. All hydraulic and sediment variables apart from water depth were significantly different between pool and run units (one-way ANOVAs, df ¼ 1, p < 0.01) with runs featuring faster flows near the bed and surface, and coarser particles than pools. Differences between reaches (i.e. CD, AS, U) were not statistically significant Table IV. Reach scale analysis of macroinvertebrate communities (A abundance, s number of taxa, m Margalef index, H Shannon’s diversity index, E Buzas and Gibson’s evenness index) Reach

CD AS U

A (N m2)

398 542 372

s

21 25 27

m

3.04 3.49 4.00

Diversity H

2.33 2.327 2.384

Evenness E

0.49 0.41 0.40

IBE

IFF

Score

Class

Value

Class

11–10 11–12 12

I I I

181–210 229–241 290–290

II–III II I

Values of the ecological indices (IBE, Extended Biotic Index and IFF, Fluvial Functionality Index) are also reported, with IFF values reported for both banks. Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

650

F. COMITI ET AL.

Table V. Hydraulic and grain size characteristics at the macroinvertebrate sampling sites Site

CDP1 CDP2 CDP3 CDR1 CDR2 CDR3 ASP1 ASP2 ASP3 ASR1 ASR2 ASR3 UP1 UP2 UP3 UR1 UR2 UR3

Reach

CD CD CD CD CD CD AS AS AS AS AS AS U U U U U U

Unit

Re

ub

us

u

h

d

(m s1)

(m s1)

(m s1)

(m)

(m)

—

—

0.18 0.26 0.07 0.26 0.80 0.60 0.28 0.13 0.03 0.23 0.44 0.78 0.12 0.01 0.16 0.14 0.50 0.55

0.28 0.29 0.20 0.58 0.88 0.59 0.40 0.34 0.06 0.30 0.52 0.79 0.09 0.09 0.15 0.49 0.46 1.00

0.23 0.27 0.13 0.42 0.84 0.60 0.34 0.24 0.04 0.26 0.48 0.78 0.11 0.05 0.15 0.31 0.48 0.77

0.33 0.36 0.43 0.29 0.25 0.20 0.24 0.27 0.25 0.18 0.25 0.29 0.43 0.29 0.41 0.38 0.19 0.40

0.05 0.04 0.01 0.12 0.12 0.07 0.09 0.02 0.04 0.04 0.05 0.09 0.02 0.03 0.03 0.06 0.04 0.01

0.13 0.15 0.06 0.25 0.54 0.43 0.22 0.15 0.03 0.20 0.31 0.46 0.05 0.03 0.08 0.16 0.35 0.39

9000 9100 700 31 720 94 400 40 800 24 500 2730 1050 9890 21 120 66 690 1920 260 4480 8960 21 000 7150

P P P R R R P P P R R R P P P R R R

Fr

Morphological unit (P: pool, R: run); ub near bed velocity (measured at 0.2 h); us near surface velocity (measured at 0.8 h); u mean profile velocity; h water depth; d mean surface grain size; Fr Froude number of the flow; Re grain Reynolds number (relative to d). Reach types are check dams (CD), artificial steps (AS) and unmodified (U). All variables are normally distributed (Shapiro–Wilk test p > 0.05) apart from Re.

(df ¼ 2, p > 0.10) for any of the physical variables, although mean grain size d was larger in the CDs reach and smaller in the unmodified reach (Table IV). Although the unmodified reach featured the coarsest surface sediments at the reach scale (Figure 2), runs comprise short gravely patches just upstream of steps, whereas in CD and AS reaches the run units were longer and often featured pebble-sized sediment. A factorial ANOVA evaluated differences in the dimensionless hydrodynamic parameters (i.e. Fr and log Re) using as factors channel units, channel reaches and their interaction effects (reach  unit). Both Froude numbers and log-transformed grain Reynolds number were significantly different between pool and run units (Fr: df ¼ 1, F ¼ 24.53, p < 0.001; log Re: df ¼ 1, F ¼ 14.69, p < 0.005), whereas reach type (Fr: df ¼ 2, F ¼ 0.95, p > 0.10; log Re: df ¼ 2, F ¼ 1.91, p > 0.10) and interactions between unit and reach (Fr: df ¼ 2, F ¼ 0.35, p > 0.10; log Re: df ¼ 2, F ¼ 0.17, p > 0.10) were not significant. Therefore, physical variables measured at sampling sites can be assumed comparable among the three reaches. Macroinvertebrate communities at the channel unit scale. The hypothesis that macroinvertebrate communities at the channel unit-scale differ in taxa composition (i.e. number of individuals per taxa found at each sampling site) was tested by a NMS–ANOSIM analysis. Figure 7 shows the two-dimensional configuration obtained by NMS, which features a stress value of 0.10, indicating a good fit to the observed distance matrix, as was also suggested by a low scatter around the step-line representing the transformed input data D-hat (Kruskal, 1964) in the Shepard diagram (not shown). In Figure 8, most of the sites belonging to the unmodified (U) reach plot in the lower part of the graph showing negative values with respect to dimension 2, whereas CD sites plot in the upper area of the graph. ASs units show a larger scatter, but generally plot between the U and CD sites. It is also notable that pool and run sites display predominantly positive and negative values, respectively, in relation to dimension 1. Overall, the ANOSIM indicates (Table VI) that the three reach types differed significantly (mean rank within groups 67.9 and between groups 80.8), but the pairwise analysis illustrates that this is due to the large ‘distance’ between U and CD sites, whereas other combinations did not prove to be statistically significant. Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS

651

Figure 7. NMS analysis on taxa composition of the macroinvertebrate communities at the unit scale. 2D configuration of the 18 sampling sites (see Table V for site acronyms)

Table VII lists calculated parameters of macroinvertebrate communities at the 18 sampling sites. The number of taxa (s) was not normally distributed, even following transformation, and so was not included in the subsequent parametric analysis. A factorial ANOVA tested the hypothesis that reach type, reach unit and/or reach  unit interactions affect macrobenthos parameters. Reach type and interaction effects were not found to have a significant effect (reach: df ¼ 2, p > 0.10; reach  unit: df ¼ 2, p > 0.10) on macroinvertebrates abundance, but abundance was significantly

Figure 8. Principal components analysis: variables loadings (a) and site scores (b) with respect to the first two principal components. See Tables IX and X for eigenvalues and eigenvectors. This figure is available in colour online at www.interscience.wiley.com/journal/rra Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

652

F. COMITI ET AL.

Table VI. Results of the analysis of similarities (ANOSIM) on macroinvertebrate taxa composition among the three reaches (all), showing the statistic R and its significance ( p-value)

All CD–U AS–U CD–AS

R

p

0.17 0.47 0.04 0.04

0.10

Pairwise statistics represent a post hoc test. Large positive R (up to 1) signifies dissimilarity between groups. The significance is computed by permutation of group membership, with 10 000 replicates (CDs, check dams; AS, artificial steps; U, unmodified reach).

dependent upon channel unit (df ¼ 1, F ¼ 7.38, p < 0.05), with runs hosting a much higher number of benthic organisms than pools in all reaches. Similarly, reach type and reach  unit interactions did not have significant effects (df ¼ 2, p > 0.10) on diversity indices (m, H, E) or taxa composition indices (EPT s, EPT N), whereas channel unit was significant for m and H (i.e. runs have more diverse communities than pools, df ¼ 1, p < 0.05) and EPT N (runs host a higher proportion of EPT individuals df ¼ 1, F ¼ 4.00, p < 0.10). Finally, from a functional perspective, the proportion of SHR was significantly higher in pools (df ¼ 1, F ¼ 9.01, p < 0.05) than in runs, whereas reach type was again not significant (df ¼ 2, p > 0.10).

Relationship between physical variables and benthic communities In order to investigate links between macroinvertebrates and hydrodynamic factors, correlations between pairs of variables (Table VIII) were estimated and a PCA was applied to the physical (Table V) and macrobenthos Table VII. Macroinvertebrate characteristics calculated for the 18 sampling sites, with N number of individuals, A abundance per streambed area, s number of taxa, m Margalef richness index, H Shannon index, E evenness index, EPT N and EPT s proportion, respectively, of the total number of individuals and the total number of taxa from the Ephemeroptera–Plecoptera– Tricoptera group, SHR proportion of the total number of individuals that are shredders Site

CDP1 CDP2 CDP3 CDR1 CDR2 CDR3 ASP1 ASP2 ASP3 ASR1 ASR2 ASR3 UP1 UP2 UP3 UR1 UR2 UR3

Reach

CD CD CD CD CD CD AS AS AS AS AS AS U U U U U U

N

A

s

m

H

E

EPT N

EPT s

SHR

—

(N m2)

—

—

—

—

—

—

—

27 152 49 158 187 143 253 59 68 147 170 279 33 38 126 175 175 123

90 507 163 527 623 477 843 197 227 490 567 930 110 127 420 583 583 410

12 15 9 16 17 16 18 10 10 17 21 18 11 10 16 18 18 16

3.30 2.80 2.10 3.00 3.10 3.00 3.10 2.20 2.10 3.20 3.90 3.00 3.10 2.50 3.30 3.70 3.40 3.10

2.2 2.08 1.5 2.23 2.27 1.96 1.92 1.51 1.71 2.32 2.32 1.97 2.07 1.81 1.66 2.05 2.34 1.88

0.76 0.54 0.50 0.58 0.57 0.44 0.38 0.45 0.55 0.60 0.48 0.40 0.72 0.61 0.33 0.43 0.58 0.41

0.85 0.58 0.55 0.78 0.82 0.87 0.84 0.61 0.49 0.78 0.84 0.93 0.64 0.89 0.86 0.81 0.74 0.76

0.83 0.67 0.89 0.75 0.71 0.75 0.67 0.50 0.60 0.76 0.67 0.78 0.64 0.80 0.56 0.61 0.61 0.69

0.48 0.57 0.69 0.24 0.26 0.20 0.17 0.83 0.79 0.31 0.34 0.40 0.67 0.37 0.77 0.27 0.59 0.13

Reach types are check dams (CD), artificial steps (AS) and unmodified (U). All variables are normally distributed (Shapiro–Wilk test p > 0.05) apart from taxa richness s. Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

653

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS

Table VIII. Pearson’s correlation coefficients between physical (rows) and macroinvertebrate (columns) parameters (sample size n ¼ 18)

ub us u h d Fr Log Re

A

m

H

E

EPT N

EPT s

SHR

0.68 0.61 0.66 0.38 0.66 0.68 0.79

0.38 0.36 0.39 0.07 0.30 0.39 0.59

0.44 0.34 0.40 0.41 0.49 0.46 0.62

0.26 0.34 0.31 0.04 0.06 0.27 0.22

0.47 0.42 0.46 0.24 0.48 0.48 0.50

0.11 0.10 0.11 0.05 0.16 0.10 0.00

0.51 0.67 0.62 0.28 0.57 0.60 0.55

For explanation of physical and macroinvertebrate variable names see Tables V and VII. Significant correlations ( p < 0.05) are indicated by .

(Table VII) variables, using log-transformed values for Re and excluding the number of taxa (s) which was not normally distributed. Log Re was the physical variable with the highest correlation with macroinvertebrates abundance A (R ¼ 0.79, p < 0.001), Margelef richness m (R ¼ 0.59, p < 0.01), Shannon’s diversity H (R ¼ 0.62, p < 0.01) and EPT N (R ¼ 0.50, p < 0.05). Froude number Fr also had significant positive correlations with A and EPT N, and near bed velocity ub had generally higher correlations with physical parameters than near surface or vertically averaged values. Water depth was not significantly correlated with any macroinvertebrates variables. Finally, the relative abundance of SHR was significantly, negatively, correlated with both Fr and Re, and most strongly negatively correlated with the near surface velocity us (R ¼ 0.67, p < 0.005). PCA extracted four components with an eigenvalue greater than one (Table IX). Component 1 explained 52.2% of the variance in the data set (eigenvalue 7.3) and was inversely correlated with kinetic and turbulence characteristics of the flow (the highest correlation was with Fr and log Re, Table X). Macroinvertebrates richness, diversity and abundance were also inversely correlated with this factor, whereas the proportion of SHR in the macroinvertebrate community was inversely correlated with flow velocity (Figure 8a). The second component explained only 13.5% of the variance in the data set (eigenvalue 1.9) and appeared to be mainly correlated with macroinvertebrate evenness and diversity (Table X). The plot of the 18 sampling sites with respect to the first two components (Figure 8b) shows a clear distinction between run units, which have high scores on component 1, and pool units, which have low scores on component 1. More difficult is the interpretation of any relationship between the types of sampled site and component 2. Half of the run and pool units in the unmodified reach lie in the upper part of the graph, reflecting their higher macroinvertebrates richness as do half of those from the CD reaches, whereas four out of six sites in the AS reach plot in the lower area of the graph. In general, CD sites display higher scores on component 2 than sites from AS and unmodified reaches, indicating the former’s higher evenness and diversity, but there is no clear grouping due to reach type displayed in Figure 8b.

Table IX. PCA analysis: eigenvalues, % variance and %cumulative variance explained for the first six factors extracted (out of 14) Principal components 1 2 3 4 5 6

Copyright # 2009 John Wiley & Sons, Ltd.

Eigenvalue

% variance

% cumulative variance

7.31 1.88 1.29 1.14 0.92 0.54

52.20 13.46 9.25 8.18 6.59 3.88

52.20 65.66 74.91 83.10 89.68 93.57

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

654

F. COMITI ET AL.

Table X. PCA analysis: eigenvectors of the correlation matrix for the first four factors extracted (i.e. those with eigenvalues >1) Variable

Component 1

Component 2

Component 3

Component 4

ub us u H d Fr Log Re A M H E EPT N EPT s SHR

0.33 0.32 0.34 0.16 0.27 0.34 0.34 0.30 0.22 0.23 0.10 0.24 0.05 0.28

0.12 0.18 0.16 0.14 0.12 0.11 0.01 0.19 0.21 0.48 0.62 0.14 0.36 0.17

0.14 0.24 0.19 0.25 0.04 0.13 0.16 0.20 0.40 0.26 0.07 0.00 0.68 0.21

0.16 0.00 0.08 0.56 0.17 0.18 0.11 0.03 0.49 0.05 0.26 0.45 0.07 0.25

For explanation of the variable names, see Tables V and VII.

DISCUSSION The most widespread adverse impact on mountain streams of the Alps is commonly identified as the artificialization of channel bed and banks. Despite clear changes in bed morphology associated with grade-control works (e.g. slope reduction), modifications of stream ecological structure and functionality are often unclear. Our analysis indicates that reference (unmodified bed morphology) conditions in the Maso di Spinelle River support an overall higher macrobenthos richness and diversity than modified reaches, as well as higher fluvial functionality and IBE score (Table IV). Analyses at both reach (Sørensen similarity) and unit scale (NMS– ANOSIM) demonstrate that the reference benthic community of the unmodified reach differs significantly from that found in the CDs reach, but the contrast with the ASs reach is smaller and not statistically significant. The reach with the highest evenness value was the one controlled by CDs (Table IV), whereas the AS reach showed a lower evenness, which was similar to that of the unmodified reach. Indeed, the presence of lentic, depositional habitats in mountain rivers, which are a consequence of a lower average channel slope induced by grade-control works, may lead to an increase in diversity and/or evenness by enhancing the hydromorphological diversity of a stream (Bona et al., 2008). Nonetheless, an ordination of sampling units based on site-related hydraulic and macroinvertebrate parameters (PCA, Figure 9) did not clearly distinguish between modified and unmodified segments. Similarly, the variance in macroinvertebrate parameters at sampling sites was not statistically explained by reach type (factorial ANOVAs). Indeed, it seems that at the channel unit scale the local hydraulic/ sediment variables determine macroinvertebrate parameters such as abundance, diversity and functional composition (i.e. proportion of SHR), possibly suggesting that CPOM retention/availability, which showed significant contrasts between reach types, is not a limiting factor. Although counterintuitive, the visually rough hydraulic characteristics of the reference reach (Figure 2) features the lowest local Fr and Re values, because these variables reflect only stream-wise velocity, whereas in step pool channels both lateral and vertical velocity components can be dominant (Wilcox and Wohl, 2007), with very high turbulent intensities. In fact, most of the energy dissipation in these channels occurs as hydraulic jumps in the pools, where the longitudinal velocity is relatively low compared with the vertical component. Large turbulence intensities can be very effective in detaching invertebrates and thus increasing passive macroinvertebrates drift rates, and so these harsh three-dimensional hydrodynamic conditions may represent the limiting factors for macroinvertebrates productivity in step pool channels. This may explain why macrobenthos abundance is lowest in the reference reach (Table IV), notwithstanding its highest and most evenly distributed CPOM retentiveness (Figures 5 and 6). However, a 3D analysis of the flow field would be required for a more detailed investigation on macrobenthos distribution in similar streams. Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS

655

Interpretation of the relative results between CDs and ASs is more problematic. In this study, ASs featured the highest macroinvertebrate abundance in both pools and runs, possibly because bed geometry, sediment size and morphology represent a ‘hydraulically milder’ (i.e. less turbulent) physical environment than a natural step pool channel. The AS reach was more productive than CD associated with large pools, possibly because of higher CPOM availability along the entire channel bed length (i.e. not locally stored in deep pools). The main reason could be the bed slope between the grade-control structures, which is higher for the ASs (sub-reach ASU, 4.5%) than the CDs (2%) reach, since such a steeper bed gradient determines the formation of a ‘secondary’ step pool morphology between the artificial macro-steps, thus creating habitat units that are more similar to the reference conditions. The very short development of a sloping bed between the ASs in the downstream sub-reach (ASD) is the most probable cause for the very poor retentiveness of this channel segment, despite the presence of three pools. Indeed, this evidence, coupled with the positive correlation between leaf velocity and CPOM retentiveness (i.e. the faster leaves move the more they get captured), seems to suggest that the dominant mechanism for leaf retention is their entrapment by boulders and cobbles on the bed surface, which is likely more effective when leaves are moving rapidly over a rough channel bed with large clasts protruding into the main flow. Along with this spatially distributed mechanism, large pools are also known to be responsible for storing CPOM in natural channels (Siligardi et al., 2000), and a threshold pool size relative to the incoming flow energy can be envisaged, discriminating between poorly retentive (as in ASD) and ‘over-retentive’ (as in CDP) pools. As a tentative appraisal, the crest-to-crest CDs reach (CDP) was characterized by a pool length equivalent to 130 times the flow depth measured during low flows at the step crest, whereas the AS reach featured pool length/ flow depth ratios of
100 (ASD) and
83 (ASU), and the unmodified reach showed a much smaller ratio (49), suggesting a threshold value in the range of 100–150. Alternatively, because of the direct correlation between pool length and drop height (Lenzi et al., 2003), the threshold value could be expressed in terms of the drop height, which is also more convenient for grade-structure design. In this case, over-retentive pools (i.e. CDP) are characterized by drop height > 100 times the flow depth, whereas in the reference reach the ratio is
6, and AS feature values around 30 (ASD) and 13 (ASU). Within the ASs reach, it appears that low drop steps (ASU-type) bring about more natural CPOM dynamics compared to higher, short-spaced structures (ASD-type), even though the latter resemble more closely the ideal step pool geometry. Even though the macrobenthos was not sampled in the latter sub-reach because of the lack of suitable sites, both macroinvertebrate diversity and total abundance are probably smaller than in the former due to the limited extent of run-type habitats. In fact, the present results confirm what has already been noted in other mountain streams (e.g. Buffagni and Comin, 2000), that coarser units characterized by fast-flowing water such as riffles and runs present higher macroinvertebrate diversity and density than deeper, slower areas covered with finer sediments such as pools. It is interesting to note that the IFF (Table IV), rather than the IBE index, adequately ‘ranks’ the three reaches in relation to observations of macrobenthos communities and CPOM retentiveness. The IFF’s good capabilities for synthesizing overall river ecological status have already been noted by Ballestrini et al. (2004). Nonetheless, other new methods for evaluating habitat structures (e.g. HQA, HMS, Raven et al., 1998; Buffagni and Kemp, 2002; LRD, Buffagni, 2004) could also be applied and compared in the Maso River, as has already been successfully achieved in several mountain streams of the Western Italian Alps (Bona et al., 2008). CONCLUDING REMARKS Overall, this study indicates a relatively higher ecological performance of ASs compared to traditional CDs, in the sense that they display smaller contrasts with reference conditions than traditional CDs. Smaller, more frequent ASs appear to be a valid alternative to higher grade-control structures, although a reach controlled by ASs still differs from a ‘natural’ reach with self-formed step pool morphology, especially in terms of CPOM retentiveness. However, ASs similar to those in the Maso River do not represent an ideal solution for fish, being too high for the local salmonid species, even though some trout have been reported to successfully traverse the steps. Unfortunately, the presence of a long segment controlled by 8 m high CDs downstream of the ASs greatly limits the need to provide fish passage because the ASs reach is already disconnected from the downstream main watercourse (Pini Prato, 2004). Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

656

F. COMITI ET AL.

Boulder steps mimicking a natural step pool morphology are being increasingly adopted to ‘restore’ steep channel reaches worldwide, but the objectives of the intervention may differ substantially between geographic areas. In the European Alps, where steep mountain streams often represent a hazard for the human population, ASs are typically built as an alternative to concrete CDs to prevent bed incision and excessive sediment transport, and so they cannot be referred to as ‘restoration’ measures. Here, the use of boulder steps instead of concrete grade-control structures is acceptable as long as they perform the function for which they are installed. In order to perform the required function, steps must be high enough to dissipate flow energy in hydraulic jumps at pools, i.e. they should not be submerged during design flood events to avoid the possible onset of a skimming flow regime (Comiti et al., submitted). In this context, the ASs in the Maso River demonstrated a satisfactory flood control effectiveness over a 10-year period, in particular during the 1998 event (RI 20–25 years). In situations where population safety must be guaranteed for flood events with return interval of more than 50 years (e.g. densely urbanized basins), reinforced concrete structures may be required. However, to limit adverse impacts on CPOM retention and benthos colonization of both concrete and boulder structures, drop height should be limited (i.e. just enough to cause unsubmerged hydraulic jumps) and the spacing long enough to provide some length of sloping bed downstream of pools. Even though ASs do not represent a true ecological restoration technique, they may represent for steep mountain stream a valid trade-off between hazard mitigation effectiveness, visual blending in with the landscape and ecological functionality. However, further post-project appraisals in different settings are needed to confirm these preliminary results.

ACKNOWLEDGEMENTS

We are grateful to Professor A. Battisti for the availability of the Laboratory of Entomology (School of Agriculture, University of Padova). W. Panazzolo, A. Checchinato and all the students and colleagues who helped in the field are greatly thanked. Support for the research was provided by University of Padova projects ‘Channel adjustments and restoration in response to human alterations of wood and sediment fluxes in gravel bed rivers, N- 60A081729/08’ and ‘Real and potential volumes assessment and geomorphic effects of woody debris in mountain rivers of the Eastern Alps, N- PDR075378/07’. We are thankful to two anonymous reviewers and to Professor Angela Gurnell for providing excellent comments and thorough revisions which have greatly improved the original paper.

REFERENCES Abrahams AD, Li G, Atkinson JF. 1995. Step-pool streams: adjustment to maximum flow resistance. Water Resources Research 31(10): 2593– 2602. Ballestrini R, Cazzola M, Buffagni A. 2004. Characterising hydromorphological features of selected Italian rivers: a comparative application of environmental indices. Hydrobiologia 516(1–3): 365–379. Bona F, Falasco E, Fenoglio S, Iorio L, Badino G. 2008. Response of macroinvertebrate and diatom communities to human-induced physical alteration in mountain streams. River Research and Applications 10.1002/rra.1110 Buffagni A, Comin E. 2000. Secondary production of benthic communities at the habitat scale as a tool to assess ecological integrity in mountain streams. Hydrobiologia 422/423: 183–195. Buffagni A, Kemp JL. 2002. Looking beyond the shores of the United Kingdom: addenda for the application of River Habitat Survey in South European rivers. Journal of Limnology 61(2): 199–214. Buffagni F. 2004. Classificazione ecologica e carattere lentico-lotico in fiumacroinvertebratesmediterranei. Quaderno Istituto di Ricerca sulle Acque 122: 164 pp. Bunte K, Abt SR. 2001. Sampling surface and subsurface particle-size distributions in wadable gravel- and cobble-bed streams for analysis in sediment transport, hydraulics, and streambed monitoring. General Technical Report RMRS-GTR-74, USDA Forest Service, Rocky Mountain Research Station. 409. Buzas MA, Hayek LAC. 1996. Biodiversity resolution: an integrated approach. Biodiversity Letters 3(2): 40–43. Calkins D, Dunne T. 1970. A salt tracing method for measuring channel velocities in small mountain streams. Journal of Hydrology 11: 379–392. Cattelan M. 2005. Valutazione della velocita` media in un torrente sistemato tramite step-pool naturali. BSc Thesis, University of Padova, Italy. Checchinato A. 2008. Sistemazione di torrenti alpini con briglie di diversa tipologia: effetti sulla capacita` di ritenzione della sostanza organica. BSc Thesis, University of Padova, Italy. Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS

657

Chin A, Purcell A, Quan JW. 2007. Assessing geomorphological and ecological responses in restored step-pool systems. Geological Society of America Abstracts, 39(6): 107. Church M, Zimmermann A. 2007. Form and stability of step-pool channels: research progress. Water Resources Research 43: w03415. DOI: 10.1029/2006wr005037 Clarke KR. 1993. Non-parametric multivariate analysis of changes in community structure. Australian Journal of Ecology 18: 117–143. Comiti F, Andreoli A, Lenzi M. 2005. Morphological effects of local scouring in step-pool streams. Earth Surface Processes and Landforms 30(12): 1567–1581. Comiti F, Cadol D, Wohl E. In Press. Flow regimes, bed morphology and flow resistance in self-formed step-pool channels. Water Resources Research. Comiti F, Mao L, Wilcox A, Wohl E, Lenzi MA. 2007. Field-derived relationships for flow velocity and resistance in high-gradient streams. Journal of Hydrology 340: 48–62. D’Agostino V. 2004. Sull’affidabilita’ delle misure di portata nei torrenti montani con il metodo della diluizione salina. In Proceedings 29th Italian Congress of Hydraulics and Hydraulic Structures. Trento, Italy, 1005–1012. Elder K, Kattelmann R, Ferguson R. 1991. Refinements in dilution gauging for mountain streams. International Association of Hydrological Sciences Publications 193: 247–254. Fu¨reder L, Vacha C, Amprosi K, Bu¨hler S, Hansen CME, Moritz C. 2002. Reference conditions of alpine streams: physical habitat and ecology. Water, Air, and Soil Pollution: Focus 2: 275–294. Ghetti PF. 2001. Indice Biotico Esteso (I.B.E.). I macroinvertebrati nel controllo della qualita` degli ambienti di acque correnti. Provincia Autonoma di Trento, 205. Giller PS, Malmqvist B. 1998. The Biology of Streams and Rivers. Oxford University Press: New York, NY. Hammer Ø, Harper DAT, Ryan PD. 2001. PAST: Paleontological Statistics Software Package for Education and Data Analysis. Palaeontologia Electronica 4(1): 9. Hilderbrand RH, Lemly AD, Dolloff CA, Harpster KL. 1997. Effects of large woody debris placement on stream channels and benthic macroinvertebrates. Canadian Journal of Fisheries and Aquatic Sciences 54: 931–939. Kemp JL, Harper DM, Crosa GA. 2000. The habitat-scale ecohydraulics of river. Ecological Engineering 16: 17–29. Kondolf GM. 1995. Five elements for effective evaluation of stream restoration. Restoration Ecology 3: 133–136. Krebs CJ. 1989. Ecological Methodology. Harper & Row: New York. Kruskal JB. 1964. Multidimensional scaling by optimizing goodness of fit to a nonmetric hypothesis. Psychometrika 29: 1–27. Lamberti GA, Gregory SV. 1996. Transport and retention of CPOM. In Methods in Stream Ecology, Hauer FR, Lamberti GA (eds). Academic Press: San Diego, CA; 217–229. Lenzi M. 2002. Stream bed stabilization using boulder check dams that mimic step-pool morphology features in Northeastern Italy. Geomorphology 45: 243–260. Lenzi MA, Comiti F. 2003. Local scouring and morphological adjustments in steep channels with check-dam sequences. Geomorphology 55: 97–109. Lenzi MA, Comiti F, Marion A. 2004. Local scouring at bed sills in a mountain river: the Plima River, Italian Alps. Journal of Hydraulic Engineering, ASCE, 130(3): 267–269. Lenzi MA, Mao L, Comiti F. 2006a. Effective discharge for sediment transport in a mountain river: computational approaches and geomorphic effectiveness. Journal of Hydrology 326: 257–276. Lenzi MA, Mao L, Comiti F, Lucchini G. 2006b. La resistenza al flusso in corsi d’acqua sistemati con sequenze di briglie. Quaderni di Idronomia Montana 26: 375–386. Lenzi MA, Marion A, Comiti F. 2003. Local scouring at grade-control structures in alluvial mountain rivers. Water Resources Research 39(7): 1176–1188. Lepori F, Palm D, Malmqvist B. 2005. Effects of stream restoration on ecosystem functioning: detritus retentiveness and decomposition. Journal of Applied Ecology 42: 228–238. Linlokken A. 1997. Effects of instream habitat enhancement of fish population of a small Norwegian stream. Nordic Journal of Freshwater Research 73: 50–59. Magurran AE. 2004. Measuring Biological Diversity. Blackwell Publishing: Oxford; 256 pp. Mao L, Lenzi MA. 2007. Sediment mobility and bedload transport conditions in an alpine stream. Hydrological processes 21: 1882–1891. Monaghan MT, Robinson CT, Spaak P, Ward JV. 2005. Macroinvertebrate diversity in fragmented Alpine streams: implications for freshwater conservation. Aquatic Science 67: 454–464. Montgomery RD, Buffington JM. 1997. Channel-reach morphology in mountain drainage basin. Bulletin of the Geological Society of America 109(5): 596–611. Palmer MA, Bernhardt ES, Allan JD, Lake PS, Alexander G, Brooks S, Carr J, Clayton S, Dahm CN, Follstad Shah J, Galat DL, Loss SG, Goodwin P, Hart DD, Hassett B, Jenkinson R, Kondolf GM, Lave R, Meyer JL, O’Donnell TK, Pagano L, Sudduth E. 2005. Standards for ecologically successful river restoration. Journal of Applied Ecology 42: 208–217. Panazzolo W. 2006. Valutazione della risposta ecologica di un torrente montano a diverse tipologie di sistemazione idraulica: il caso del T. Maso di Spinelle (TN). MSc Thesis, University of Padova, Italy. Petersen R. 1992. The RCE: a riparian, channel, and environmental inventory for small streams in the agricultural landscape. Freshwater Biology 27: 295–306. Petts GE, Gurnell AM, Milner AM. 2000. Hydro-geomorphological controls on glacier-fed river ecosystems. In Perspectives on Glacial Systems, Munro D (ed.). Canadian Geophysical Union, National Hydrology Research Institute: Saskatoon. Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

658

F. COMITI ET AL.

Pini Prato E. 2004. Una proposta di valutazione indicizzata delle priorita` di intervento nella realizzazione di passaggi per pesci. Proceedings of 6th ‘‘Convegno Nazionale sui Passaggi per Pesci’’, Provincia di Modena, 33–43. Poole RW. 1974. An Introduction to Quantitative Ecology. McGraw-Hill: New York. Raven PJ, Holmes TH, Dawson FH, Everard M. 1998. Quality assessment using River Habitat Survey data. Aquatic Conservation: Marine and Freshwater Ecosystems 8: 477–499. Roni P, Bennett T, Morley S, Pess Gr, Hanson K, Van Slyke D, Olmstead P. 2006. Rehabilitation of bedrock stream channels: the effects of boulder weir placement on aquatic habitat and biota. River Research and Applications 22(9): 967–980. Roni P, Fayram A, Miller M. 2005a. Monitoring and evaluating instream habitat enhancement. In Monitoring Stream and Watershed Restoration, Roni P (ed.). American Fisheries Society: Bethesda; 209–236. Roni P, Hanson K, Beechie T, Pess G, Pollock M. 2005b. Habitat Rehabilitation for Inland Fisheries: A Global Review of Effectiveness and Guidance for Rehabilitation of Freshwater Ecosystems. FAO Fisheries Technical Paper 484 Food and Agriculture Organization of the United Nations, Rome, Italy. Siligardi M, Bernabei S, Cappelletti C, Chierici E, Ciutti F, Egaddi F, Franceschini F, Maiolini B, Mancini L, Minciardi MR, Monauni C, Rossi G, Sansoni G, Spaggiari R, Zanetti M. 2003. I. F. F. Indice di Funzionalita` Fluviale. Manuale ANPA, 223. Siligardi M, Ciutti F, Cappelletti C, Monauni C. 2000. Studio sulla capacita` di ritenzione a breve termine in un corso d’acqua alpino. Biologia Ambientale 14(2): 7–11. Wilcox A, Nelson JM, Wohl EE. 2006. Flow resistance dynamics in step-pool channels: 2. Partitioning between grain, spill, and woody debris resistance. Water Resource Resources 42: W05419. DOI: 10.1029/2005WR004278 Wilcox AC, Wohl EE. 2007. Field measurements of threedimensional hydraulics in a step-pool channel. Geomorphology 83: 215–231. Woodwiss FS. 1978. Comparative study of biological-ecological water quality assessment methods. Second practical demonstration Summary Report. Commission of the European Community.

Copyright # 2009 John Wiley & Sons, Ltd.

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra

659

ARTIFICIAL STEPS TO STABILIZE MOUNTAIN RIVERS

APPENDIX List of macroinvertebrate taxa sampled in the three reaches (CD, check-dams; AS, artificial steps; U, unmodified). List of macroinvertebrates

Plecoptera

Reach

Leuctridae Nemouridae Perlidae

Ephemeroptera

Trichoptera

Coleoptera Diptera

Gasteropoda Oligocheta Nematoda

Perlodidae Baetidae Ephemerelldae Heptageniidae Brachycentridae Hydropsychidae Psycomyidae Limnephilidae Odontoceridae Philopotamidae Polycentropodidae Rhyacophilidae Sericostomatidae Elminthidae Hydraenidae Athericidae Blephariceridae Chironomidae Dixidae Empididae Limoniidae Simuliidae Tipulidae Hydrobioidea Enchytraeidae Mermithidae

Copyright # 2009 John Wiley & Sons, Ltd.

Leuctra sp. Nemoura sp. Protonemura sp. Perla sp. Dinocras sp. Isoperla sp. Baetis sp. Ephemerella sp. Ecdyonurus sp. Epeorus sp. Rhithrogena sp.

CD

AS

U

X X X X

X X X X

X X X X X X X X

X X

X X X X X X X

X X X X X

X

X

X X X

X X X X

X X X X X X X X X X X X X

X X X X X X X X X X X X X X X X X X X X

X

River. Res. Applic. 25: 639–659 (2009) DOI: 10.1002/rra