LOWER COLUMBIA RIVER NATURAL AREA INVENTORY. John A. Christy and Judy A. Putera Oregon Natural Heritage Program

Transcription

1 LOWER COLUMBIA RIVER NATURAL AREA INVENTORY 1992 John A. Christy and Judy A. Putera Oregon Natural Heritage Program Report to The Nature Conservancy, Washington Filed Office, Seattle 3 February 1993 INTRODUCTION The purpose of this study was to identify remnant habitat approximating conditions prior to Euroamerican settlement, along the floodplain of the lower Columbia River. About 143 river miles were examined in both Oregon and Washington, between Bonneville Dam and the mouth of the river, as well as lower portions of Chinook River, Elochoman River, Grays River, Lewis River, Youngs River and the Willamette River. Our objective was to produce a list of sites, ranked by the relative importance of their elements, for use in setting priorities for protection. We also wanted to examine the historical record, to document the composition of natural communities at the time of settlement, and their subsequent disturbance. Previous work Reports of presettlement vegetation along the lower Columbia River are few and fragmentary. Most mention of plants or plant communities are made in conjunction with human activities such as exploration or commerce, and are exceedingly vague. Much of our knowledge had to be inferred from what little there is in the historical record, and from what remains along the river today. The journals of early botanists such as Douglas (1914) and Scouler (1905) are not particularly helpful in 1

2 reconstructing presettlement vegetation. Their comments were general, and localities are difficult to identify. Howell ( ), Piper and Beattie (1915) and Gorman (1926) provided a few clues for some herbaceous elements of the floodplain flora around Portland, as do a few herbarium specimens at Oregon State University and the University of Oregon. Survey notes from the General Land Office, made between 1851 and 1865, are the only reliable source for information about original riparian vegetation, although by that time the region had already been settled by Euroamericans for 25 years. Surveyors were required to list major trees and shrubs, and general descriptions provide some information on the structure of bottomland vegetation, floodplain lakes, riparian forest and the incidence of annual floods. They were not required to describe herbaceous plants, and we know least about this important part of the original flora. Of all the native vegetation along the river, the herb layer has suffered the greatest impacts from grazing, exotic weeds, and flood control, and is the most problematic to reconstruct. Frenkel and Eilers (1976) sampled vegetation in a freshwater intertidal marsh near Cathlamet Bay, the earliest description of this community known to us. Tabor (1976a, 1976b) conducted a habitat inventory along the Columbia and Snake Rivers. The report, based on field work conducted between 1973 and 1975 by teams from three universities, remains the most comprehensive document available on the abundance and distribution of habitat types along the river. Vegetation was mapped on a coarse scale from air photos, using species dominance, stand structure and landform. The mapping was supported by data from vegetation transects, 48 of which were located below Bonneville Dam. No attempt was made to describe or classify plant communities beyond what was needed for mapping. Because the study was intended to document existing habitat along the river, many of the areas sampled were dominated by exotic species. Morrison (1973) and Wiberg and Greene (1981) described Blackwater Island Research Natural Area, in the Ridgefield National Wildlife Refuge. Hinschberger (1978) surveyed the entire study area during an inventory of small mammal habitat, and included descriptions of the major plant communities along the river. Thomas (1980, 1984) described vegetation sampled in and near Cathlamet Bay, and reported several plant species 2

3 new to the Pacific Northwest. Cornelius (1985), Gehring (1990) and Scherer (1991) reported on specific problems associated with the rare plant Rorippa columbiae, but all work was restricted to Pierce Island. Vanderschaaf (1986) described Tenasillahe Island Research Natural Area, on the Julia Butler Hansen National Wildlife Refuge. Hibbs (1987) described chages on islands between the Willamette River and Bonneville Dam. Maxwell (1991) described a number of plant communities and new state records from the Washington side of the river. The recent draft environmental impact statement for the proposed land exchange at Tongue Point (U.S. Fish and Wildlife Service et al. 1992) contains much information for the lower river in the vicinity of Cathlamet Bay. Many early reports, and more recent environmental impact statements prepared by the Army Corps of Engineers no doubt contain a wealth of information for sites along the river, but we did not have time to consult these. METHODS Sources Aerial photography dating back to 1929 is housed at the U.S. Army Corps of Engineers, Portland District. Although early photo coverage is not inclusive, the collection is an invaluable reference for tracking changes in shorelines and for some types of vegetation. Miscellaneous photographs of the river are also housed at the Oregon Historical Society. Primary sources of information for rare plant and animal species, and rare plant communities were the Oregon Natural Heritage Program (1988, 1991), the Washington Natural Heritage Program (1990, 1991), Marshall et al. (1992) and the Washington Department of Wildlife. Historical information was obtained from the Oregon Historical Society (Portland), the Washington Historical Society (Tacoma), the Washington State Library (Olympia), the National Archives (Seattle), the U.S. Army Corps of Engineers (Portland), the Oregon state office of the Bureau of Land Management (Portland), the Multnomah County Library (Portland) 3

4 and James River Corporation (Vancouver). Herbarium collections were consulted at Oregon State University and the University of Oregon. Selection of sites We selected sites that were free from obvious surface disturbance, as shown on recent aerial photography housed at the Washington Department of Natural Resources and the Oregon Division of State Lands. Urban and industrialized areas were not sampled. Areas were rejected if they contained recognizable improvements such as dikes, buildings and roads, or had recently been grazed, logged or covered with dredge spoils. A few areas with such alterations were sampled for comparative purposes, or because specific elements were known to be present. We investigated a number of sites after checking records housed at the Oregon and Washington Natural Heritage Programs. Some of the information on these sites was 15 years old, and needed to be updated. Several other sites were investigated specifically because they were currently being offered for sale. In the field, most sites were accessed from the river, using a 15 foot rowboat with a 25 hp motor. Sites were rejected if they had a large amount of exotic species, or recent disturbance not evident in aerial photos. We completed species lists for most sites visited, and recorded plant cover data in plots of 10 m radius at sites judged to have the best representation of natural vegetation. The data forms were deposited at the Oregon and Washington Natural Heritage Programs. Ranking of sites and plant community element occurrences The relative importance of sites inventoried (Appendix 1) was ranked according to the criteria in Table 1. For plant community elements, global and state ranks, and element occurrence ranks (Appendix 2) were assigned according to standards established by The Nature Conservancy (1988). 4

5 Nomenclature Plant names used here are those of Hitchcock and Cronquist (1973) and Cronquist et al. (1977). Although a number of nomenclatural changes have been made recently by some authors, most are not included here because they would be unfamiliar to most users. Place names are those used on USGS topographic maps. RESULTS Our field season extended from 17 June to 21 October In 44 field days, we investigated and ranked 92 sites between Bonneville Dam and the mouth of the river (Appendix 1). High-ranked sites Not surprisingly, two factors diminished our chances of finding areas free from the effects of Euroamerican settlement. First, the lower Columbia River has suffered impacts from settlement and commerce for about 170 years, 120 years before the construction of Bonneville Dam. Second, the entire floodplain has been altered by 100 years of channel manipulation and 50 years of flood control. Virtually all features of the floodplain have been affected to a greater or lesser degree. Table 2 lists 17 sites that we consider to be the best remaining unprotected examples of conditions approximating those prior to Euroamerican settlement. Animals We observed 17 species of rare, threatened, endangered or sensitive animal species (Table 3). Previously documented element occurrences of animals are listed in Appendix 1. The great egret, sandhill crane, brown pelican, greater yellowlegs and western grebe are either summer to fall migrants or wintering species. All other taxa listed in Table 3 are known to breed within the study area. No amphibians and few reptiles were seen during the 5

6 course of the survey. With the exception of one active heron rookery on Price Island, all bird sighting were of foraging or roosting animals, primarily due to the late date of the survey period. Breeding colonies of great blue herons occur on Hunting, Ryan, Whites, Fisher, Cottonwood, Reed, and Pierce Islands, and at Vancouver Lake. Herons were observed feeding in shallow island inlets and near mudflats of the river shoreline throughout the length of the study area. A large concentration of 22 birds was observed feeding at Campbell Lake in late summer. 1. Green-backed heron. A state-monitored species in Washington, one heron was observed on Hunting Islands and two were seen feeding in mudflats on Sandy Island. 2. Great egret. Observed at three locations in the area of Ridgefield NWR. A large concentration of 21 birds was seen at Campbell Lake in late September. 3. Sandhill crane. Large fall concentrations of cranes were observed near Ridgefield at Canvasback and Campbell Lakes. 4. Greater yellowlegs. Seen foraging at Lord Island, Poysky Slough, and at several locations on the Washington side of the river. 5. Brown pelican. Several groups were observed in July at the mouth of the river on West Sand and East Sand Islands. 6. Caspian tern. Rice Island, created by dredge spoils, currently is the only known breeding colony of the Caspian tern in the study area. They were known to breed on East Sand Island in the mid 1980's. We observed terns foraging along the entire length of the study area, while larger concentrations of resting birds were observed on the mudflats at East Sand Island and Fisher Island. 7. Purple martin. Observed at Julia Butler Hansen NWR, Hunting and Ryan Islands, Ridgefield NWR and 6

7 Franz Lake NWR. Reported previously from both sides of the river, nesting in old pilings and nest boxes. 8. Turkey vulture. A state-monitored species in Washington, vultures occur in the area from March through late September. A roosting individual was observed at the Robert W. Little Preserve on Puget Island, on Fisher Island and near the mouth of the Lewis River. Soaring birds were observed upriver from mile 24. A colonial roost was reported in 1981 on the Julia Butler Hansen NWR in the vicinity of river mile 35. In late September, we saw approximately 15 vultures congregated near Wauna at river mile 43, apparently staging prior to migration. 9. Bald eagle. Eagle nests and winter populations are well documented along the lower Columbia River (Garrett et al. 1988). We observed individuals flying or roosting in Cathlamet Bay, Grays Bay, and on Karlson, Hunting and Lords Islands, Rinearson Slough, Lewis River and Pierce Island. Active nests, although not observed during this survey, have been documented for several of the areas we surveyed (Appendix 1). 10. Osprey. Tracked only in Washington, we saw ospreys at Price, Hunting, Skamania, Pierce and Fisher Islands, the mouth of the Lewis River, and Beacon Rock State Park. In Oregon, several osprey were observed at Gary, Flag, and Chatham Islands. Nests have been documented previously at several of the sites surveyed. 11. Vaux's swift. A single sighting near the river's edge at Beacon Rock State Park. 12. Peregrine falcon. Seen once, perched at West Sand Island. 13. Western grebe. Seen once near the north end of Sauvie Island. 14. Columbian white-tailed deer. Seen on Hunting Islands and common in diked pastures at Julia Butler 7

8 Hansen NWR. 15. Harbor seal. Seen at Hunting Islands. 16. Painted turtle. A species of concern in Oregon, four individuals were seen basking on logs at Millionaires Lake on Sauvie Island. They were also frequent in roadside ditches on the Ridgefield NWR, outside of our study area. We saw tracks of unidentified turtles on Sand Island, at river mile 112, where they had crossed the beach and scaled a 20-foot sandy cutbank. Plants We observed 7 species of rare, threatened, endangered or sensitive plant species (Table 4). Although many exotic species exist along the river, four pose special management problems and will eventually require control. 1. Reed canary grass (Phalaris arundinacea). Ubiquitous along the lower Columbia River, and second only to diking and clearing in the wetland acreage it has degraded or destroyed. 2. Yellow iris (Iris pseudacorus). Probably the second most serious pest on the river, rapidly expanding into the interior of otherwise undisturbed wetlands by means of its ubiquitous floating seeds. 3. Purple loosestrife (Lythrum salicaria). Occurs throughout the length of the study area, in small to moderate numbers but with regular frequency. We saw extensive infestations at five sites in Oregon, between Crims Island and Wallace Island, and between Lord Island and Walker Island. Releases of biocontrol insects should be initiated here. All sightings will be reported to state weed control authorities in Oregon and Washington. 4. False indigo (Amorpha fruticosa). Invading the lower Columbia River and known as far downriver as 8

9 Cathlamet (Glad and Halse 1993). This species may eventually dominate better-drained sites above the intertidal zone. All of these taxa should be listed as noxious weeds by both Oregon and Washington. It is unfortunate that all are readily available from the grass seed trade and nurseries. Plant Communities Cover data were recorded in 105 vegetation plots, and were used to identify 43 plant communities, which we consider to be elements, along the lower Columbia River (Table 5). Many of these communities are restricted to specific segments along the river. Global and state ranks are given in Appendix 2 for all communities occurring in Oregon, and a few are recommended for those in Washington. Ranking and size of plant community element occurrences are also given in Appendix 2. We considered seven plant communities to be rare regionally or globally. 1. Artemisia lindleyana-deschampsia cespitosa (Columbia River mugwort-tufted hairgrass cobble community). Should be ranked G1S1. This community is limited to river miles 140 to 145, just below Bonneville Dam, where the substrate is occasionallyflooded cobbles and gravels mixed with silts. The only known occurrences are both on the Washington side of the river, and are probably relicts of a once more widespread Columbia River Gorge community, now inundated by the water behind Bonneville Dam (Cornelius 1985). 2. Carex aperta (Columbia sedge community). Should be ranked G1S1. This species once formed "extensive meadows on overflow bottomlands in the valley of the Columbia and its tributaries...largely cut for hay and regarded by farmers as the best forage sedge" and was "common about Columbia Slough etc." (Gorman 1926). Piper and Beattie (1915) said it was "the common hay sedge of the Columbia 9

10 River bottoms." It probably extended from Longview to Skamania. The only known remaining example covers about two acres near Vancouver Lake. Elsewhere, it has been completely displaced by reed canary grass. The sedge itself is not rare, and can be found mixed with canary grass in many areas, but it is never plentiful. 3. Elymus mollis (American dunegrass community). Ranked G1S1 in Oregon. This species was once the dominant native grass on coastal foredunes. Although it was planted sparingly in dune stabilization programs beginning in 1916, it has been largely displaced throughout its range by the more aggressive European beachgrass (Ammophila arenaria), the species most commonly used for stabilization (Arnst 1942; McLaughlin and Brown 1942). The single population that we saw on East Sand Island may have been planted, and was not a high-quality occurrence. The island has been augmented by accretion and dredge spoil deposition since engineering began at the mouth of the river in Festuca rubra (Red fescue dune community). Ranked G1S1 in Oregon. Red fescue was a common sodforming grass of open, stabilized sand dunes on the outer coast. It was the dominant grass on the Clatsop Plains at the time of settlement, but overgrazing and trailing by livestock destroyed the sod and caused new dunes to form. Like American dunegrass, it was planted occasionally for dune stabilization, but has been displaced largely by the more aggressive European beachgrass (Arnst 1942; McLaughlin and Brown 1942). Its occurrence on West Sand Island is the primary reason for proposed designation of a Research Natural Area by the Army Corps of Engineers. Presence of active blowouts, and a relatively early stage of succession within the proposed RNA were probably caused by overgrazing when the island was used as a base for horse seining in Baker Bay. 5. Picea sitchensis/cornus stolonifera (Sitka spruce swamp). Ranked G3S2 in Oregon. These stands typically have Cornus stolonifera-salix sitchensis shrub swamp in their interior, with spruce trees 10

11 more frequent on natural levees formed along tidal sloughs. This wetland type originally extended from southeast Alaska to about Tillamook Bay, Oregon. Although about 200 acres remain in three stands known as far south as the Nehalem River, those on the lower Columbia River are the largest and southernmost occurrences left on the Pacific coast. Stands on the river originally extended from the Chinook River to Westport, covering an estimated 12,500 acres in Washington and 6500 acres in Oregon. About 1500 acres (23%) remain on the Oregon shore, and about 700 (6%) on the Washington shore. These figures include interior stands of shrub swamp, excluded in Tabor's (1976a) estimate of acres of spruce swamp for both states. 6. Fraxinus latifolia/urtica dioica (Oregon ash/nettle woodland). Ranked G3S2 in Oregon. This community forms seasonally wet stands with few or no cottonwoods, and is often laced with sloughs and vernal pools. It was probably widespread on bottomlands between Longview and Skamania, and also occurred in the Willamette Valley. Grazing and reed canary grass have degraded these stands to such an extent that it is difficult to reconstruct what the original herb layer might have been. No highquality occurrences were seen during the survey. 7. Salix lasiandra/urtica dioica (Pacific willow/nettle riparian woodland). Ranked G3S2 in Oregon. This community is wetter and is an earlier seral type than the Oregon ash/nettle woodland. It also has been degraded by grazing and reed canary grass. No high-quality occurrences were seen during the survey. Reinterpretation of river segments Tabor (1976a), the Oregon Natural Heritage Program (1988) and the Washington Natural Heritage Program (1991) divided the Columbia River below Bonneville Dam into three segments, based on coastal, coast range and interior valley physiography. Our observations of vegetation, hydrology and other physical features on the floodplain suggest that four segments characterize the area better than those delineated previously. 11

12 1. Brackish tidelands. River miles 1 to 8, from Clatsop Spit to Chinook River and Hammond. Despite the intrusion of marine salts to river mile 18 or 20, distinct salt marsh communities extend only to about river mile 8. All of these are low-salinity marshes, distinguished by the lack of typical high-salinity marsh species such as Salicornia virginica and Jaumea carnosa (Thomas 1980, 1984). Common communities are the estuarine associations of Scirpus americanus and Carex lynbyei, and forb-rich higher marshes typified by Aster subspicatus, Deschampsia cespitosa and Potentilla pacifica. The average tidal fluctuation is about eight feet. 2. Freshwater tidelands. River miles 8 to 65, Chinook River and Hammond to Longview. The floodplain along this segment of the river is of low relief, typically flooded or subirrigated at high tide, and permeated by conspicuous, dendritic and meandering tidal streams and sloughs. Cut banks, when present, show mucky soils with high organic content. Flooding associated with winter storms and high tides often drive floating logs and other debris into the interior of riparian swamps. Historically, these lands overflowed during spring floods, but usually were not subject to high energy, destructive flows. Thousand of acres of bottomland have been diked and cleared for agriculture, and the tidal streams cut off with tide gates. The wettest portions currently are dominated by freshwater associations of Carex lyngbyei and Scirpus americanus, the former extending upriver to Crims island, the latter to Cottonwood Island, just upstream from Longview. Slightly higher sites are dominated by shrub swamps of Cornus stolonifera, Salix lasiandra and Salix sitchensis, with stands of Fraxinus latifolia and Populus trichocarpa occurring on natural levees along the tidal streams. Picea sitchensis/cornus stolonifera swamps historically ranged from the Chinook River to Westport. The Chinook River floodplain also contains two freshwater shrub-swamp communities (Pyrus fusca-salix hookeriana/carex obnupta and Salix hookeriana-spiraea douglasii) typical of outer coastal habitats, and not seen elsewhere on the river. 12

13 3. Overflow plains. Longview to Skamania, river miles 65 to 140. Elevations in this segment of the floodplain are generally higher, and shorelines upriver from Portland typically have sandy cutbanks. Although hydric soils are common, the moisture regime is seasonal rather than perennial, and soils are better drained. The dendritic, meandering tidal streams of the previous segment are replaced by linear, shallow troughs and low ridges that are parallel to the river. These are relics of the extensive reworking of sediments by high-energy floods that occurred prior to flood control. Broad, shallow overflow lakes and ponds are common. These originally were recharged annually by the June flood, and many dried up by late summer. Recharge of most of these is now limited to precipitation. As in the previous segment, thousands of acres of bottomland have been diked and the sloughs gated. Weirs were installed on many of the overflow lakes to retain water for livestock and duck hunting. The wettest sites currently are dominated by extensive stands of Eleocharis palustris in shallows along the river, Salix fluviatilis on sandy banks and bars, and Salix lasiandra on wet flats along channels and around overflow lakes. The Fraxinus latifolia/urtica dioica association occurs on slightly higher sites protected by natural levees. The understory has been degraded extensively by grazing and invasion of reed canary grass. Higher banks and the tops of natural levees are dominated by associations of Fraxinus latifolia and Populus trichocarpa, with Cornus stolonifera, Symphoricarpos albus and Urtica dioica. Historically, several plant communities occurred in this segment that are now extremely rare or extirpated, including Carex aperta marsh, Deschampsia cespitosa prairie, and oak savanna with an understory probably dominated by Festuca rubra and Danthonia californica. Sand deposition between Government Island and Skamania Island formed extensive tidal flats, and east winds from the Columbia River Gorge blew sand into towering dunes on the Oregon shore in the vicinity of Rooster Rock State Park. Although the dunes on the mainland have been stabilized by vegetation, those on Sand Island are still active, and are the last such features on the lower Columbia River. 4. Columbia River Gorge. Skamania to Bonneville Dam, river miles 140 to 145. This short stretch of 13

14 river is characterized by a coarse substrate of cobble and gravels, presumably derived from the Table Mountain landslide that formed the former falls and rapids known as The Cascades. There are few wetlands on the floodplain, and the rocky shores have a sparse flora. Tidal influence reaches zero at Warrendale, just below Beacon Rock. The rare Artemisia lindleyana-deschampsia cespitosa plant community is restricted to this segment. These five miles are all that remain of what was a much more extensive reach of river, now inundated by the waters behind Bonneville Dam. IMPACTS OF EUROAMERICAN SETTLEMENT Alteration of flood regimes The Columbia River drains an area of 259,000 square miles. Prior to the construction of dams on the Columbia and its tributaries, the lower river flooded under two distinct seasonal regimes, one initiated by winter rain west of the Cascades, the other by spring snowmelt east of the Cascades. While winter floods were usually more local in distribution, spring floods affected the entire reach of river below what is now Bonneville Dam. Flood heights gradually diminished downstream, and below river mile 40 the broad estuary and strong tidal influence dissipated its effects. Floodwaters of 20 to 30 feet at Vancouver would rise to only two to five feet in the estuary (U.S. Army Corps of Engineers 1948, 1988). 1. Winter floods. The winter "rain floods" result from runoff generated by seasonal rain storms in and west of the Cascade Range, an area comprising only about 3% of the Columbia River watershed. Some 25 dams in the Cowlitz, Lewis and Willamette River basins have diminished major winter flooding on these streams. Rain floods generally crest and abate within one or two days, and originally occurred primarily on the tributaries or at their confluences with the Columbia River. Such storms caused the Columbia to flood an average of once every five years. Notable winter floods on the 14

15 Columbia occurred in 1881, 1909, 1917, 1933, 1941, 1946, 1964 and The largest of these, in 1964, was 29.5 feet above mean low water at Vancouver (U.S. Army Corps of Engineers 1948, 1988). Winter flooding also occurs between the mouth of the river and Puget Island, when high winter tides coincide with high winds and runoff from major storms, to raise water levels four to five feet above predicted high winter tides of nine or 10 feet. Some of these events can drive floating debris far into the interior of stands of riparian vegetation. The 1933 rain flood combined with winter high tides to reach a stage in the estuary equaling that of the 1894 flood. 2. Spring floods. About 93% of the Columbia River watershed is east of the Cascade Range, extending all the way to the Rocky Mountains. This region has a cold continental climate, and the primary source of high water is from snowmelt, occurring between April and August. Before the dams were built, spring floods were often higher and lasted much longer than winter floods. Most occurred each year in May or June. The average "June flood" regularly inundated 170,000 acres of bottomland along the lower Columbia River for periods up to 60 days. Major floods inundated 250,000 to 300,000 acres. The spring floods were the primary force influencing landforms and vegetation on the river bottoms. The average annual flood reached a stage of 21.5 ft. above mean low water at Vancouver, a had a flow of about 600,000 cfs. These average flows were punctuated periodically by huge flows that had tremendous impact on riparian habitats. Spring floods greater than 600,000 cfs occurred 46 times between 1858 and Floods between 800,000 and 900,000 cfs occurred in 1859, 1866, 1871, 1882, 1887 and Floods exceeded 900,000 cfs in 1862, 1876, 1880, 1894 and The largest of these, in 1894, reached 1,254,000 cfs, twice as large as the average spring flood. Floodwaters reached 43 feet above mean low water at Beacon Rock and 36.2 feet at Vancouver, diminishing to about 12 feet at the mouth of the river. A spring flood in 1849 was probably of the same magnitude (U.S. Army Corps of Engineers 1948, 1988). 15

16 Aerial photographs showing flooded bottomland are housed at the Oregon Historical Society, and most depict areas between Longview and Crown Point. Over enormous areas of bottomland, only the tops of ash, Pacific willow and cottonwood trees were visible at highest flood stages, the lakes and sloughs all being inundated. Historical accounts of floods in rural areas along the lower Columbia River are hard to find. Surveyors for the General Land Office noted in the 1850's that "the Columbia bottoms...are all overflowed by the river in high water", "subject to annual inundation varying from one to fifteen feet deep" and "subject to overflow from the high waters of the Columbia River which occurs annually...in the first of May and continuing until the middle of July, during which time the farmers may sail over their farms in boats." Bottomlands at Rooster Rock State Park flooded from four to 20 feet. On Puget Island, "water stood on practically all the land for several months each year" (Butler 1953). Sauvie Island "was inundated almost every year by the floods and back waters of the two rivers" (Cleaver 1989). Diking, more than 200 hydroelectric and storage dams, irrigation diversions and a long-term decline in precipitation have diminished the occurrence of floods along the lower Columbia River. Comparison of early and contemporary photographs shows some striking changes along the river, due primarily to flood control, and to a lesser extent, channel controls such as pile dikes. Most obvious is a change in shorelines on both islands and the mainland, caused by accretion of sediment. Many islands between Portland and Bonneville Dam had wide aprons of sand and gravel flats around their upstream ends, and cutbanks where flood waters scoured away sediments and vegetation. Undermined vegetation dumped into the river and piled up at the upper ends of islands downstream. Some of these features are still active at Reed, Sand and Skamania Islands. Most islands between Puget Island and Bonneville Dam have increased in size, and many interior sloughs have silted in. The Salix fluviatilis-salix lasiandra community has populated former sand and gravel flats, expanding the size of islands and shorelines, to the detriment of Rorippa columbiae and the Artemisia lindleyana-deschampsia cespitosa community at the upper end of the study area (Cornelius 1985; Scherer 1991). Shallow overflow lakes and ponds have diminished in size because of similar siltation and 16

17 encroachment of upland vegetation. Diking and drainage for agriculture The history of diking and drainage on the lower Columbia River reflects the history of increasing use of the floodplain for purposes incompatible with the annual floods. Surveyors for the General Land Office noted that most of the bottomlands were already claimed at the time of survey, as was true for much of the area between Longview and Portland, while much of the uplands were still unclaimed. The waterfront was the best location for commercial traffic, boats and barges were the only means of transportation, and the bottomlands had the best soils for farming if they could be drained. Government surveyors considered Sauvie Island "very valuable for pasturage [but]...too wet except on the highest places for agriculture." On Puget Island, "high tides made certain parts of the island uninhabitable" (Butler 1953), and "no one lived on the island year round as the water depth was a problem at times...in 1880 folks began planting a garden and pasturing a cow in the wild green vegetation" (Jones 1973). The 1894 flood "covered the roofs of the houses and every living person and animal had to go to the mainland" (Butler 1953). At Stella, "each spring the freshet brought high water to anything along its banks...the buildings along the river were built on stilts to allow for this occurrence...in June the water was so high, the steamboats couldn't land...willow Grove had to deal with high water every spring and all the homes were built on stilts...every year the farmers had to move their cattle to higher ground" (Stella Historical Society 1989). At Skamokawa, the three "valley bottoms were completely covered with water...the Stanard family in Middle Valley eventually abandoned their farm, due to the constant severe flooding of the freshet, which destroyed their farm almost annually" (Martin 1985). The first dikes on the lower Columbia River were constructed in 1899 by the Warrenton Diking District (U.S. Army Corps of Engineers 1988). By the 1890's, diking was underway in both Middle Valley and on Brooks Slough behind Skamokawa (Martin 1985). Klatskanie had dikes by Portions of Tenasillahe Island, Brownsmead and bottomland between Coal Creek and Kelso were ditched and diked by 1914 (Butler 1953; Stella Historical Society 1989). By 1917,

18 acres had been diked on Puget Island, but half remained undiked as late as Karlson Island was diked by local interests prior to 1936 (Jones 1973). Some 2500 to 3000 acres of spruce and cottonwood swamp on the mainland portion of what is now Julia Butler Hansen NWR persisted as late as 1929, although construction of dikes had begun. Formation or reorganization of diking districts accelerated between 1935 and 1940 to comply with provisions of the 1936 Flood Control Act. The act authorized funding for 65 projects by the Army Corps of Engineers, including construction or rehabilitation of dikes, riverbank protection, ditching and pumping facilities. The Corps helped construct 42 of the projects, 11 were built by local interests, and six were built as relief projects during the Depression. By 1948, 99,000 acres, or 58% of the floodplain, had been diked (U.S. Army Corps of Engineers 1948). The 1950 Flood Control Act authorized 32 diking projects, only 13 of which were completed (U.S. Army Corps of Engineers 1988). We did not have time to research the acreage of floodplain diked, nor the vegetation types affected. Soil surveys for Oregon (Green 1983; Smythe 1986; Smith and Shipman 1988) indicate that almost 49,500 acres have been diked along the Columbia River. Soil surveys from Washington do not include data on diked lands. Channel alterations for navigation Dredging and construction of a multitude of pile dikes has been done by the Army Corps of Engineers and port districts on the lower Columbia River. The primary purpose is to maintain flows and depth for a primary shipping channel for the entire length of the river. Freighter traffic is limited to the reach between Astoria and Portland, and barge traffic extends all the way upriver to Lewiston, Idaho. Early Columbia River pile dikes were at widely scattered locations and not under a comprehensive plan. In 1878 a 6,200-foot longitudinal pile dike was constructed at St. Helens (Hickson 1961). Jetties at the Columbia Bar at the mouth of the river were built between 1885 and 1917 (McLaughlin and Brown 1942, Hickson 1961). Other dikes were constructed by the Port of Portland at Martin Island, Walker Island and Snag Islands. In the river channel between 18

19 Vancouver and Cathlamet, 150 spur dikes and 100 contraction dikes were built between 1916 to 1930, and 50 more by 1960 (Hickson 1961). The most obvious result of pile dikes and dredge spoil disposal has been the accretion of sediments around wing dams, formation of new islands, and an increase in the size of existing islands. A large amount of new habitat has formed as a result, and some sites (e.g., Miller Sands, Rice Island, Jim Crow Sands) have become some of the most valuable wildlife habitat on the lower river. Much new habitat is wetland, but many weeds also thrive on higher elevations. The most extensive stands of purple loosestrife on the lower river occur on dredge spoil islands near Wallace Island. Beginning in the 1860's, the Army Corps of Engineers removed thousands of submerged snags from channels in the Willamette River, and cut streamside forests when they were thought to be a hazard to navigation (Sedell and Luchessa 1981, Sedell and Froggatt 1984, Sedell 1986). Presumably, this activity also occurred on the lower Columbia River. Removal of snags and streamside trees greatly altered channel morphology, and greatly reduced the input of organic material that was an important source of food and habitat for river fauna. Fishing Although we did not include fish in our inventory, we were impressed by the widespread remains of a voracious fishing industry that went into decline in the 1920's. The gill net fishery of today is but a shadow of an industry that included horse seining, fish traps, fishwheels, saltworks and canneries. Many shallow areas between Puget Island and Astoria had horse seining operations dating back to the 1860's (Butler 1953; Jones 1973; McClelland 1974; Martin 1985). Some of the sand bars had horse barns, canneries and cookhouses built on pilings, to avoid high tides. West Sand Island had four or five canning or seining companies operating there, and horses were grazed on the island. This activity peaked in 1926 and was banned in Washington in In the 1890's, packing houses at Skamokawa took in as 19

20 many as 40 tons of sturgeon in a single day. One company, located on Price Island opposite Nassa Point, was the largest of four firms on the Columbia River that processed sturgeon (Martin 1985). Fish traps and fishwheels appeared on the river in In 1900, more than 500 of these extended from Fort Canby to what is now Bonneville Dam. They were outlawed by Oregon in 1926 and by Washington in 1934 (Donaldson and Cramer 1971). The first cannery on the Columbia River was built at Eagle Cliff in By 1877 there were 29, and by 1883 there were 39 below Oak Point (Appelo 1969; McClelland 1974). Canneries were also located at Rooster Rock and Warrendale (Donaldson & Cramer 1971). Thousands of rotting pilings are all that remain of this industry on the lower Columbia. Since 1910, runs of anadromous fish on the Columbia River have declined 85 to 90% (Northwest Power Planning Council 1987; Pacific Coast Joint Venture 1992). Although most of this decline resulted from construction of dams on the Columbia River, overfishing early in the century certainly exacerbated the problem. Grazing and farming We did not research the history of grazing or farming along the river, but their impacts are evident on all but the wettest sites. It is well known that the Hudson Bay Company had dairy and beef herds on Sauvie Island and the floodplain east of Fort Vancouver as early as the 1820's. Beginning in 1850, hay was grown on Government Island for horses quartered at Vancouver Barracks. General Land Office survey maps of the early 1850's show farm fields on the floodplain at Sauvie Island and on the south shore east of Portland. These practices extended up and down the river in the following decades. Thousands of acres of floodplain were developed for dairy farming as soon as dikes were completed. Aerial photography from 1929 onwards shows that nearly every large island upstream from Puget Island had people living on it. Prior to the widespread construction of dikes, houses were built on stilts, and people and livestock had to move to higher ground during the annual spring floods. Most of the undiked islands no longer have people living on them, 20

21 and a declining number are still grazed by cattle. Notable exceptions are Martin, Deer, Sauvie and Government Islands, where grazing is still severe. Grazing is likewise severe elsewhere on the floodplain, especially on mainland areas, including most of the National Wildlife Refuges upstream from Longview. Cattle had only been removed recently from Reed Island State Park, Rooster Rock State Parks, and Franz Lake NWR. The worst legacy left by cattle grazing on the floodplain of the lower Columbia River are the thousands of acres contaminated by reed canary grass. Promoted as a hardy wetland forage species by the U.S. Department of Agriculture, the Soil Conservation Service and county extension agents (e.g., Schoth 1938; Hafenrichter et al. 1968; Hannaway and McGuire 1981), it was planted widely between 1940 and 1960, and is still available today. It spread aggressively into unplanted areas, and as far as natural areas are concerned, it remains an unmanageable catastrophe (Apfelbaum and Sams 1987). Fate of the original riparian forests The primary tree species on the floodplain of the lower Columbia River are cottonwood, ash, Pacific willow and Sitka spruce. We saw no old-growth stands of these in the study area, although we saw individual trees that were unquestionably old. The history of exploitation of the river's riparian forests and swamps by Euroamericans is very poorly documented. Virtually all stands were logged at least once, the wood used for lumber, fuel and paper pulp. Removal of thousands of submerged snags and streamside trees to improve navigation would have extensively altered the structure and habitat value of riparian forests along the river. Sedell and Froggatt (1984) estimated that between 1868 and 1950, the Willamette River had an average of one snag pulled for every five feet of river run. By 1985, losses of riparian forest had reduced the average to one snag every 1000 feet. Presumably, the lower Columbia River would have had similar losses. Prior to the widespread use of fuel oil in the 1890's, there was a huge demand on the Columbia River for firewood to fuel steamboats, locomotives and heat homes. Steamboats were the primary means of transportation along the river between 21

22 1851 and the 1920's, and each burned between 10 and 30 cords of wood each day (Sedell and Froggatt 1984). We have not seen figures for the lower Columbia River, but steamboats on the Missouri River were reputed to have burned 300,000 cords of cottonwood during the same period, leaving riparian stands "denuded of usable trees by the turn of the century, a period of 30 to 40 years" (Evans 1989). Arsdol (1964) quoted that "wood cutting for steamboats was probably the first great occupation of [Skamania] county residents... stacks of wood covering large areas could be seen along the shores, waiting for transportation." Downriver, "there were many wood docks along the Columbia River because river boats used cord wood for fuel" (Stella Historical Society 1989). Early photographs at the Oregon Historical Society show cordwood stacked on docks and decks of steamships, but it is not evident what species were used. "Cordwood was about as plentiful as air. But nobody wrote about air -- why write about firewood, or even record statistics about it?" (Reynolds 1942). Appelo (1975) noted that tideland spruce from Grays Bay was sold for cordwood in Astoria in 1883, but evidence suggests that most of the cordwood trade on the river involved conifers logged from hills above the floodplain. Arsdol (1964) quoted that "wood was cut on hills inland...and brought to the landing to be loaded on steamboats for fuel." The Stella Historical Society (1989) described a five-mile wood flume built between 1885 and 1910 for transporting cordwood. Above Bonneville Dam, "pioneers cut cordwood on the mountain slopes...and hauled it to the river banks where it was loaded onto scows...to be used as fuel" (Donaldson and Cramer 1971). Although lower in fuel value, coniferous wood would have been easier to cut and split than hardwood. Conifers comprised 80% of all the fuelwood burned in the North Pacific region, and cutting for this purpose peaked in Oregon in 1918, and in Washington in 1908 (Reynolds 1942). Dairy farming on the Columbia River was the impetus for diking, clearing and draining thousands of acres of riparian swamps and forests. This activity peaked between 1900 and In contrast, dairy farming was not as common along the Willamette River. Clearing of riparian forests there did not accelerate until after 1940, when crop diversification, irrigation and flood control made the floodplain attractive for agriculture (Towle 1982). 22

23 1. Cottonwood. Ubiquitous on the lower Columbia River, cottonwood occurs on both perennially wet soils and drier soils. Pure stands are common throughout the length of our study area, usually on slightly higher elevations such as old floodplain terraces or natural levees. Several stands had individuals with two or three trunks clustered together, most likely originating as stump sprouts formed after logging. Stands logged within the last ten years at Walker Island, Goble and on the Lewis River near Mud Lake, showed vigorous sprouting from cut stumps. Trees we saw ranged from 10 to 48 inches in diameter, and most were between 18 and 36 inches. We saw only a few evidently old-growth cottonwoods of large girth on the lower Columbia River, at Tenasillahe Island, Hunting Islands, Burke Island and Reed Island. These trees are characterized by diameters of eight or nine feet, deeply furrowed bark and broken tops. In 1825, Douglas described cottonwoods ranging from six to 12 feet in diameter. It was "the chief tree on [the Columbia's] numerous islands...all the low banks of the river are covered with it" (Douglas 1914). Cottonwood snags pulled from the Willamette River between 1868 and 1875 measured 90 to 165 feet long and five to nine feet in diameter (Sedell and Luchessa 1981, Sedell and Frogatt 1984). Although used to fuel steamboats east of the Cascades (Evans 1989), the primary use of riparian cottonwoods west of the Cascades seems to have been for lumber and paper pulp. Up to 7 million board feet per year were used for barrel staves, wooden ware, crates, trunks and drawer bottoms (Sargent 1896, Nash 1904, Jepson 1910, Peattie 1953). Demand "caused the destruction of most of the old trees" by 1896 (Sargent 1896). In 1885, a mill in Camas used "straw and some cottonwood, with a small percentage of rag" to make paper (Parsons 1983). A pulp mill at Warrendale employed sandstone grinding wheels to abrade bolts of cottonwood to make paper pulp (Donaldson 1974). 23

24 On both the Columbia and Willamette rivers, logging for paper pulp reached major proportions by the turn of the century (Sedell and Froggatt 1984). "But a few years ago this timber was called worthless, nowadays the steamboats tow great rafts to the paper mills of Oregon City every year" (Nash 1904). Wood chips from upland logging gradually replaced native cottonwood as a source of paper pulp between 1900 and 1920, although use of hybrid cottonwoods for this purpose has increased markedly in the last decade. 2. Oregon ash. Ash is almost as ubiquitous as cottonwood along the lower Columbia River, and is most common on the overflow plains between Longview and Skamania. Although usually intermixed with cottonwood on terraces and natural levees, pure and often extensive stands are frequent on slightly lower elevations between riverside levees and interior overflow lakes and ponds. As noted previously, all the stands we saw had been degraded by grazing and invasion of reed canary grass. Most trees we saw on the lower Columbia River, both mixed with cottonwood and in monotypic stands, ranged from 8 to 48 inches in diameter, and most were between 11 and 21 inches. Ash trees with similar diameters in the Willamette Valley ranged from 59 to 72 years old (Frenkel & Heinitz 1987). Many stands therefore appear to have originated between 1910 and 1930, and reproduction continues at the present time. In contrast, a few stands on Sauvie Island, the Ridgefield NWR and at Burlington Bottoms, have old-growth trees with diameters of five to six feet. Many of them occur at the edges of tidal creeks draining interior overflow lakes, and are absent farther back from the streams. They are always hollow, impossible to date with an increment borer, and are surely of presettlement age. Peattie (1953) reported trees over two hundred years old. Trees of this size seem to be most common on the overflow plains topography between Deer Island and Portland. Adjacent monotypic stands of ash were always the younger age class described previously. 24

25 Although ash trees were used for wagon and carriage frames, tool handles, oars, barrel staves, furniture and interior finish of houses, its primary use was for firewood (Sargent 1894, Jepson 1910, Peattie 1953). Reynolds (1942) provided statistics on oak and alder, but not for Oregon ash. Presumably it was used primarily for domestic heating and not for commercial steamship contracts. 3. Pacific willow. Although all willows were classified as shrubs during this inventory, Pacific willow often attains tree height throughout most of its range. It is ubiquitous between Grays Bay and Bonneville Dam, but reaches its best development as a community in the overflow plains segment between Longview and Skamania. It occupies a topographic position between the Fraxinus latifolia/urtica dioica or the Fraxinus latifolia-populus trichocarpa/cornus stolonifera/urtica dioica communities and what was probably the Carex aperta community, now completely replaced by reed canary grass. The understory is now dominated by reed canary grass, and we found no stands with an intact understory. Trees we saw ranged from 8 to 30 inches in diameter, and most were between 13 and 20 inches. Although absolute dates are unknown, photography housed at the Army Corps of Engineers and the Oregon Historical Society indicates that trees with diameters of 24 to 30 inches are 50 to 60 years old. This is about the maximum diameter seen along the river. Pacific willow is a favored food of beaver. The community seems to be maintained by vigorous resprouting following blowdown or heavy cropping by beaver. Although well suited to periodic inundation, prolonged flooding has killed stands at Rooster Rock State Park and at Smith and Bybee Lakes. The Pacific willow community remains widespread around overflow lakes and ponds, where historically it probably reached its greatest development. Apparently there were no commercial uses of the species in our area (Peattie 1953). Clearing of 25