Placing Riffle Formations to Restore Stream Functions in a Wet Meadow

Home , Channel pattern, Riffle, Stream restoration

PROJECT REPORT Placing Riffle Formations to Restore Stream Functions in a Wet Meadow

by Alvin L. Medina and Jonathan W. Long

any mountain meadows in the White Mountain Apache Tribe to develop Msouthwestern United States have techniques that are cost-effecti ve and Natural materials and been eroded and dewatered by channel emulate natural processes for restoring downcutting—an action that threatens riparian wetlands in the White Mountains trout fisheries, induces changes in vegeta- of Arizona. We are pleased to report that tion, and can even cause perennially flow- we have been able to build low-cost, in- knowledge of natural ing streams to go dry (Heede 1986, stream riffle formations, using natural Rosgen 1996). Restorationists seeking to materials and processes, to restore restore channel stability and fish habitat degraded streambanks and enhance trout in these riparian systems need to under- habitat and meadow vegetation. stream processes are stand underlying geomorphic processes (Kondolf 2000), especially the role of pool-riffle sequences in maintaini ng Study Site dynamic equilibrium of stream channels Pacheta Creek is a tributary of the Black used to develop a (Dolling 1968, Yang 1971, Heede 1986). River in the southeastern corner of the While many stream rehabilitation White Mountain Apache Reservation in efforts focus on modifying riffle or pool east-central Arizona (Figure 1). A peren- features (FISRWG 1998), systematic eval- nial stream, it has a typical low flow of 1.1 cost-effective means uation of most techniques has been lim- ft3/sec (0.03 m 3/sec) and an estimated ited (Kauffman and others 1997). Rock bankfull flow of 35.3 ft 3/sec (1 m3/s), both structures have been used to restore of which are comparable to other moun- degraded riparian areas by halting further tain streams (DeBano and Heede 1987). of restoring a montane incision (DeBano and Heede 1987) and Although high-inten sity storms occur by reconnecting incised channels to their occasionally during the summer monsoon, former floodplains (Rosgen 1997). Other the stream’s peak flows occur principally approaches have favored plugging the from winter snowmelt across its 2.8-mi 2 wet meadow stream. degraded channel and redirecting flows (7-km2) drainage area. At an elevation of into a newly constructed one (Rosgen 8,938 ft (2,725 m), the stream flows from 1997, Jemison and Neary 2000). Artificial a narrow canyon into a broad (820-3,280 riffles have also been used to enhance ft {250-1,000 m}) wet meadow, known as aquatic habitat, particularly in Europe, by Pacheta Cienega. This meadow reach providing substrates for spawning harbors a robust population of non-native (Brookes 1987, Leopold 1994). brook trout ( Salvelinus fontinalis ) (Rinne In this article, we describe a joint 2000), and is located at the transition effort by the Rocky Mountain Research from steeply sloping silicic rocks to flat- Station (RMRS) of the U.S. Department lying basalt flows (Merrill 1974). This of Agriculture Forest Service and the geologic transition coincides with a shift

120 ECOLOGICAL RESTORATION 22:2 JUNE 2004 in soil type from a gravelly loam to a dense clay loam (USDA SCS and USDI BIA 1981). The susceptibility of this reach to degradation reflects the tenuous equilibrium between the gravel bedload derived from the coarse-textured silicic volcanics and the fine-textured soils of the meadow.

A ssessing the D egradation In 1995, we observed that the channel had begun to downcut 590 feet (180 m) downstream from a culverted road crossing. Most of the gravels along the bed in this reach had washed away and the streambanks were sloughing into the stream. After 558 feet (170 m) of instability, the channel leveled out (less than 0.5 percent slope) into a flat, marshy area where the eroded sediments were depositing. The streambank s were dominated by native graminoids, including blister Figure 1. July 1997. Pacheta Creek at mid-point of treated reach, prior to restoration treat- sedge (Carex vesicaria ), Nebraska sedge ment. Note low level of water relative to banks, undermining of banks, and animal ramp in (C. nebrascensis ), silvery sedge ( C. canes- the left foreground. Photo by Jonathan Long cens), woolly sedge ( C. lanuginosa ), Baltic rush (Juncus balticus ), tufted hairgrass this site to exclude livestock . T ribal nel, and spreading high flows into the (Deschampsia caespitosa ), meadow barley hydrologists directed a project to enlarge floodplain. It is also possible to dissipate (Hordeum brachyanthe rum ), and alpine the capacity of the culvert crossing to stream energy by increasing the meander- timothy (Phleum alpinum ). However, Ken- facilitate passage of bedload. However, ing of the existing channel or by carving a tucky bluegrass ( Poa pratensis ), an exotic drought conditions in 1996 constrained new, more sinuous channel. However, in perennial and an undesirable species for the growth of stabilizing vegetation and observing reference meadow sites, we streambank protection (Medina 1996), encouraged elk to feed and walk along the found that channel sinuosity varies consid- had encroached on drier areas. unusually shallow stream. Where the erably, often reflecting changes in soil com- We attributed channel downcutting channel was becoming incised and unar- position. Rather than laterally eroding or and widening to direct and indirect effects mored, an additional 4 to 8 inches (10 to excavating the meadow’s highly developed of use by elk and cattle (Figure 1). Animal 20 cm) of downcutting occurred by 1997. soils to form a more sinuous channel, we trampling had formed “ramps” along some decided to use the riffle formation method streambank sections used as crossings, left because it uses commonly available rock hoof imprints in the stream bed, and cre- materials to reestablish vertical stability ated slump deposits in the channel—all of R iffle Form ation Technique within the existing channel pattern. which produce turbulence and concentrate and Im plem entation After comparing morphological data stream flows (Trimble and Mendel 1995). Geomorphologists recognize that incision from two other meadow sites, we designed These direct effects induced erosion of of stream channels can be countered by the treatment for Pacheta Creek. We gravel substrates that armored the bed. The dissipating the erosive power of the stream. quickly recognized that we would have to erosion scoured pools, widened the chan- We use the term “riffle formation” to bring in riffle-forming materials because, nel, lowered the water below the over- describe composites of rock materials that although the bedload of Pacheta Creek hanging banks, and exposed plant roots. are 1) sorted, with finer gravels placed was dominated by fine gravels, this nat- These geomorphic changes caused the above larger rocks that form the bottom ural source was insufficient to replace the meadow to become drier, which, in turn, layer at the downstream end; 2) packed coarse gravel substrates that had protected favored the expansion of shallow-rooted down, with a dip to center flows in the the channel bed from erosion. graminoids, such as Kentucky bluegrass. channel; and 3) reinforced with sedge To emulate natural formations, we Starting in late 1995, the Tribe transplants (Figure 2B). Placement of riffle studied stable reference sites to determine addressed the immediate cause of de- formations serve to dissipate energy by the location, length, height, and spacing of gradation by modifying livestock rota- increasing the undulation of the channel riffle formations in relation to stream gra- tions across the landscape and by fencing bed, increasing the roughness of the chan- dient, soil characteristics, and substrates.

ECOLOGICAL RESTORATION 22:2 JUNE 2004 121 We also examined the incised streambanks of Pacheta Creek to locate gravel lenses that revealed the locations of former riffles. We marked starting and ending points for new formations at remnant riffle locations and above and below animal crossings (Figure 2A). This procedure produced an overall spacing between riffles of about four to seven bankfull widths, which is consistent with natural riffle sequences (Leopold and others 1964, Keller and Melhorn 1978). Placement of the formations took one day of intensive labor. A Tribal enterprise delivered 16 tons (14.5 metric tons) of large gravels and small cobbles (1-5 inches {25- 125 mm}) to the site. A work crew consist- ing of T ribal staff, RMRS scientists, and Figure 2. Plan view (A), cross-section (B), and profile (C) of typical reaches treated with riffle participants in the Tribe’s summer youth formations. environmental program transported the rock materials to each riffle site using wheel- riffles that became submerged as new ones the deviations from the average slope to barrows and a small trailer. Workers placed were placed, until each riffle was at the measure bed undulation (Madej 2001). To the rocks at 25 riffles spaced along 558 feet desired height relative to the preceding objectively identify pools and riffles, we (170 m) of the degraded reach. Each forma- and succeeding ones. calculated the differences in elevation tion averaged 7.9 feet (2.4 m) in length. between points along the profile and desig- Some workers raked and stomped the rocks nated changes in bedform wherever a dif- into the bed and under the banks to keep A nalysis of ference was greater than 0.75 times the flows centered through the riffles. standard deviation of the differences Meanwhile, others cut plugs of sedges from Longitudinal Profiles (O’Neill and Abrahams 1984). wet areas in the meadow and placed them We profiled the stream channel at among gravels along the sides of the riffles. Pacheta Cienega using a laser level in July The purpose of transplanting was to stabilize 1995 (when downcutting was first diag- R esults and D iscussion the riffles with a living fabric and to revege- nosed), in July 1997 (prior to riffle place- We found that the increased height of the tate bare areas along the streambanks. ment), in October 1997 (after initial riffle riffles reduced the average distance from In a stable meadow ecosystem, bank- placement), in July 1998, and in July the streambed to the bank in the most full designates the level at which a stream 2000. We recorded bed elevation, thalweg degraded reach by one-third, from 25 to can access its floodplain. We measured water depth (the depth at the lowest point 18 inches (63 cm to 45 cm) (Mann- the height of bankfull in the stable section across the stream), and mean water depth Whitney U test, p < 0.001), which was a of Pacheta Creek and in comparab le in these surveys. From 1995 to 2001, we primary objective of the treatment. The streams to be about 8 inches (20 cm) periodically sampled channel substrates riffle formations also increased average above base flows. Using these data, we using pebble counts (Bevenger and King channel bed undulation, as measured in placed the riffle formations so that the 1995) along 40-m reaches in both the sta- terms of absolute deviation s from the water level rose to within 8 inches of the ble and the downcutting sections. average slope of the channel (Figure 3). top of the banks (Figure 2C). We used graphical analysis techniques The potential to increase the amplitude of We had also observed that water to evaluate changes in channel morphol- bedforms is limited by the stream’s ten- flowing over riffles in our reference ogy during the monitoring period. We dency to smooth out riffles that are too streams was relatively quiet. This auditory entered the longitudinal profile data into a high. However, the created formations clue helped workers build the riffles, since database, and then we used linear interpo- appear to be within the tolerance of the riffles that were built too high produced lation between measured points to com- channel, since bed undulation has not gurgling noises from the rapid flow of pare changes in bed elevation and water changed appreciab ly since they were water over the formations. Workers depth through time. We estimated changes placed (Figure 3). We also used the differ- repacked and raked the rocks toward the in bed material volume by multiplying the encing technique to determine that the banks until the gurgling sounds subsided, average difference in channel bed eleva- enhancement of pools and riffles did not indicating reduced velocities. Through an tion by the length and average width of the produce much change in the percentage iterative process, we added rocks to raise reach. We calculated the absolute value of of pools (Table 1).

122 ECOLOGICAL RESTORATION 22:2 JUNE 2004 the gravels slid into the pools immediately ) m

( downstream. Later that year, we aug-

p mented some riffles that had sunk below

o * l

s 0.4 – their prescribed height and, in 1999, we

e placed large, angular cobbles on the b

n * downstream end of formation s. With a

e 0.3 – * these changes, the bed regained nearly all m of the material lost through the treated m o

r reach from its low in July 1997 (Table 1). f 0.2 – n Minimal changes in the longitudinal pro- o i t file occurred between 1998 and 2000, a i v

e indicating that the modifications to the

d 0.1 – channel bed had become stable. e t u

l Although streams generally rearrange o s gravels placed to form riffles (Brookes b 0.0 – A 1997), initial riffle shifting in this project 1995 1997 (Pre) 1997 (Post) 1998 2000 may have been exacerbated by our use of Year of survey rocks that were relatively homogenous in shape and size. As Leopold (1994) pointed Figure 3. Channel bed undulation, as shown by boxplots of absolute deviations from the aver- out, and we now confirm, substrate hetero- age bed slope (m). Median undulation (central bars) increased after riffle treatment in 1997 geneity helps to maintain a pool-riffle pat- and remained stable through 2000. tern. The combination of adding larger , angular rocks as well as the stream’s incor- Cobbles poration of fine gravels from its natural bed- Large gravels load has increased the heterogeneity of Above Small gravels Treated substrates and reduced the amount of clay treated reach Fines reach hardpan. As shown in Figure 4, coarse sub- 100 – strates have increased in the treated reach from 54 percent in 1995 to 68 percent in 90 – 2001 (Chi-square test, p = 0.03). Further- 80 – more, changes in streambed composition have extended to the reach immediately 70 – upstream of the formations, where we mea- 60 – sured an increase in coarse substrates from 73 percent in 1995 to 88 percent in 2001 50 – % (Chi-square test, p = 0.006) (Figure 4). 40 – Capturing fine gravels in the bedload that would have been exported from the 30 – reach under pre-treatm ent conditions 20 – provides two important benefits. First, the fine gravels have filled in spaces between 10 – the larger substrates, reinforcing the sta- 0 – bility of the formations. Second, the fine Pre- Post- Pre- Post- gravels lie in the size class (0.16-0.64 treatment treatment treatment treatment inches {4-16 mm}) that is optimal for trout Figure 4. Zigzag pebble counts of the bed material pre- and post-treatment, both above and spawning (Rinne 2000). Although gravel within the treated reach, show the distribution of different size materials. The cobbles and placement can be susceptible to infilling large gravels of the riffle formations supplanted the fine hardpan in the treated reach, while with fine sediments (Rosgen 1996), the small and large gravel deposited above the treated reach. bedload at this site is relatively coarse, and aquatic plants have been able to col- onize the fine particles that are trapped on While about 283 ft 3 (8 m3) of bed replaced a significant amount of those the channel margins. material had been lost between July 1995 materials. During the spring runoff in Another important effect of the riffle and July 1997 due to downcutting (Table 1998, the heights of the riffles decreased formations was to increase the average 1), the installation of riffle formations while their lengths increased as some of thalweg water depth (Table 2). The depths

ECOLOGICAL RESTORATION 22:2 JUNE 2004 123 Table 1. Changes in channel morphology measured through longitudinal profile sur- stream itself suggested the location and veys of the entire treated reach from 1995 to 2001. heights for individ ual riffles. As the stream responded to the treatment by Change in bed material Net change in bed Percent pool reshaping some of the formations, we 3 3 volume (m ) from material volume (m ) (determined by bed added larger substrates to step down the Survey Date previous survey since 1995 form differencing) steeper gradient sections. 1995 — 0 54 The advantages of this technique for July 1997 -8.0 -8.0 63 wet meadow restoration are its low cost O ctober 1997 +2.8 -5.2 57 ($500 for the rock and a few days of 1998 +3.2 -2.0 50 2001 +1.4 -0.6 55 donated labor), its use of naturally occur- O verall Change -0.6 +1 ring materials, and its lack of reliance on heavy equipment for installation. While the use of heavy equipment would be Table 2. Means and standard deviations of thalweg depth (cm) of pools, riffles, and more cost-effective in situations requiring overall reach before and after treatment. greater amounts of rock, the ability of manual laborers to properly sort and pack Survey Date Pool Depth Riffle Depth Overall Depth the riffle formation materials and inte- M ean S.D. M ean S.D. M ean S.D. grate the sedge transplants into the for- July 1997 26.8 12.0 10.0 5.2 19.7 11.0 mation should not be underestimated. O ctober 1997 36.6 13.1 17.3 10.8 28.2 15.5 Finally, although it is a structural 1998 30.5 10.9 16.9 8.8 23.8 12.0 intervention, the riffle formation tech- Change 97-98 3.7 6.9 4.0 nique promotes stream evolution through % Change 97-98 14.0 69.0 20.0 natural processes. Although a single large structure could also prevent downcutting, multiple small structures are more effec- of both pools and riffles increased because also helped to reduce animal damage to the tive because they re-create the natural the formations raise the overall water level channel, since elk cross over the more resis- geomorphology of the channel (Heede and are shaped to center flows mid-chan- tant riffles. Furthermore, the treatment has 1986, Rosgen 1997). Furthermore, using nel and away from the streambanks. As the improved the aesthetics of the meadow by multiple small structures allows a wider formations raised water levels in riffles and replacing the pockmarked hardpan with an margin for error in placement and allows pools, they also expanded in-stream habi- attractive mixture of variegated substrates. the stream to rearrange the formations as tat available to trout. Within the past three it reestablishes geomorphic equilibrium. years (1999-2002), we, as well as anglers The formations can easily be modified by probing the site, have observed numerous C onclusions adding or removing substrates of different trout swimming in the deepened pools. Five years after the initial treatment, the rif- sizes. Therefore, we recommend reexam- The formations also improved trout habi- fle formations have proven to be a sound ining the formations following high runoff tat by increasing the availability of wetted approach for restoring high-value pool-riffle events to ensure that they are functioning gravel substrates occupied by macro-inver- systems where natural bedloads are insuffi- as desired and to allow for such adjust- tebrates. This reflects other findings that cient to replace riffle materials lost to ments if needed. For these reasons, the rif- artificial riffles can greatly increase benthic streambed erosion. In contrast to methods fle formation technique provides an organisms (Gore and others 1998). that require soil disturbance, this technique effective, adaptive approach to treating In terms of vegetation, the higher avoids the risk of increased sedimentation wet meadow degradation. water levels along the channel have downstream. However, this treatment Building upon our experience at Pach- allowed the sedge transplants to thrive and should be applied only in conjunction with eta Cienega, we have subsequently treated revegetate former animal crossings and treatment of the causes of degradation, sites with moderate gradients (about 2 per- eroded banks (Figure 5). The addition of which may include improperly designed cent) by forming riffle steps composed of sedge transplants to the riffle formations has road crossings and overuse by animals. substrates ranging from small boulders to increased channel roughness and helped to While gravel and rock formations fine gravels. These formations perform sim- bind the formations to the bank. The have been used to restore degraded ilar geomorphic functions as log steps heightened riffles disperse flows onto the streams, this approach is distinctiv e (Debano and Heede 1987), but with less banks and into the meadow during spring because of its adaptive nature and the risk of failure and need for maintenance. floods. In addition to dissipating stream integration of plants and heterogeneous energy, the riffle formations have re- substrates to form the riffles. While the ACKNOWLEDGMENTS armored much of the bed, rendering it less placement of the formations fit with The authors wish to thank the White Mountain susceptible to erosion. The treatment has empirically determine d patterns, the Apache T ribe and its members who imple-

124 ECOLOGICAL RESTORATION 22:2 JUNE 2004 Leopold, L.B., G.M. Wolman and J.P. Miller. 1964. Fluvial processes in geomorphology. San Francisco: W.H. Freeman. Madej, M.A. 2001. Development of channel organization and roughness following sedi- ment pulses in single-thread, gravel bed rivers. Water Resources Research 37(8): 2259-2272. Medina, A.L. 1996. Native aquatic plants and ecological condition of southwestern wetlands and riparian areas. Pages 329-335 in Proceedings of the symposium on desired future conditions for southwestern riparian ecosystems: Bringing interests and con- cerns together. U.S. Department of Agri- culture Forest Service, Rocky Mountain Forest and Range Experiment Station General Technical Report 272. Merrill, R.K. 1974. The late Cenozoic geology of the White Mountains, Apache County, Arizona. Ph.D. dissertation, Arizona State University. O’Neill, M.P . and A.D. Abrahams. 1984. Objective identification of pools and riffles. Water Resources Research 20:921-926. Figure 5. July 2001. Pacheta Creek at mid-point of treated reach, after restoration treatment. Rinne, J.N. 2000. Effects of substrate composi- Note height of water relative to banks and growth of sedge transplants across formerly bare tion on Apache trout fry emergence. Journal ground. Photo by Alvin Medina of Freshwater Ecology 16(3):355-365. Rosgen, D.L. 1996. Applied fluvial geomorphol- mented this restoration project, permitted the Gore, J.A., D.J. Crawford and D.S. Addison. ogy. Pagosa Springs, Colorado: Wildland publication of these findings, and continue to 1998. An analysis of artificial riffles and Hydrology. demonstrate leadership in the ecological re- enhancement of benthic community diver- __. 1997. A geomorphological approach to storation of their ancestral homeland. sity by physical habitat simulation (PHAB- restoration of incised rivers. Pages 12-22 in SIM) and direct observati on. Regulated S.S.Y. Wang, E.J. Langendoen and F .D. Rivers: Research and Management 14:69-77. Shields, Jr. (eds.), Proceedings of the con- REFERENCES Heede, B.H. 1986. Designing for dynamic ference on management of landscapes dis- Bevenger, G.S. and R.M. King. 1995. A pebble equilibrium in streams. Paper No. 86004. turbed by channel incision, May 19-22, count procedure for assessing watershed Water Resources Bulletin 22(3). 1997. Oxford, Mississippi: The Center for cumulative effects. U.S. Department of Jemison, R. and D.G. Neary. 2000. Stream Computational Hydroscience and Engin- Agriculture Forest Service, Rocky Moun- channel designs for riparian and wet eering, University of Mississippi. tain Forest and Range Experiment Station, meadow rangelands in the southwestern Trimble, S.W . and A.C. Mendel. 1995. The Research Paper RM-RP-319. United States. Pages 305-306 in U.S. cow as a geomorphic agent: A critical Brookes, A. 1987. Restoring the sinuousity of Department of Agriculture Forest Service, review. Geomorphology 13:233-253. artificially straightened stream channels. Proceedings of land stewardship in the 21st U.S. Department of Agriculture Soil Con- Environmental Geology and Water Science century: The contributions of watershed servation Survey and U.S. Department of 10(1):33-41. management, RMRS-P-13. the Interior Bureau of Indian Affairs DeBano, L.F. and B.H. Heede. 1987. Enhance- Kauffman, J.B., R.L. Beschta, N. Otting and D. (USDA SCS and USDI BIA). 1981. Soil ment of riparian ecosystems with channel Lytjen. 1997. An ecological perspective of Survey of Fort Apache Indian Reservation, structures. Water Resources Bulletin riparian and stream restoration in the west- Arizona. U.S. Government Printing Office 23(3):463-470. ern United States. Fisheries 22(5):12-24. 235-991/102. Dolling, R.K. 1968. Occurrence of pools and Keller, E.A. and W .N. Melhorn. 1978. Yang, C.T. 1971. Formation of riffles and pools. riffles: An element in the quasi-equilibrium Rhythmic spacing and origin of pools and Water Resources Research 7:1567-1574. state of river channels. Ontario Geography riffles. Geological Society of America Bulletin 2:3-11. 89:723-730. Federal Interagenc y Stream Restorati on Kondolf, G.M. 2000. Some suggested guide- Working Group (FISRWG). 1998. Stream lines for geomorphic aspects of anadromous Alvin L. Medina and Jonathan W. Long are research corridor restoration: Principles, processes, salmonid habitat restorat ion proposal s. ecologists at the Rocky Mountain Research Station, and practices . Washington, D.C.: The Restoration Ecology 8(1):48-56. 2500 S. Pine Knoll Dr., Flagstaff, AZ 86001, Federal Interagency Stream Restoration Leopold, L.B. 1994. A view of the river . 928/556-2180 and 928/556-2181, Fax: 928/556- Working Group. GPO Item No. 0120-A; Cambridge, Massachusetts: Harvard Uni- 2130, [email protected] and [email protected]. SuDocs No. A 57.6/2:EN 3/PT.653. versity Press.

ECOLOGICAL RESTORATION 22:2 JUNE 2004 125