Thesis (M.S.)--Oregon State University, 2003. / Typescript (photocopy). Includes bibliographical references (leaves 63-67). Also available online.

Thesis (M.S.)--Oregon State University, 2000. / Typescript (photocopy). Includes bibliographical references (leaves 62-65). Also available on the World Wide Web.

A well-accepted idea in geomorphology is that landforms control the type and distribution of biological habitat. However, the linkages between geomorphology and ecology remain poorly understood. In rivers, the geomorphic template controls the hydraulic environment, partly shaping the river ecosystem. But what processes shape the geomorphic template? Here, I examine how two hillslope processes dominant in the Oregon Coast Range, debris flows and deep-seated landslides, affect valley floor width and channel slope, key components of the geomorphic template in riverine ecosystems. I then investigate how patterns in potential salmon habitat differ between streams dominated by deep-seated landslides and streams dominated by debris flows. I show that terrain influenced by deep-seated landslides exhibits (1) valley widths that are more variable throughout the network but less locally variable, (2) more variable channel slopes, and (3) more potential salmon habitat as well as significantly more connectivity between habitat types.

The West Tidewater earthflow, one of the largest in Oregon's history, occurred in December of 1994. The earthflow is located approximately 15 km north of Jewel, Oregon near the summit ofthe Northern Oregon Coast Range Mountains. The earthflow is 900 m long and 250 m wide, giving it a surface area of 9 ha, or 22 acres. Volume is 3.5 million m3. The earthflow occurred in low strength, well-bedded, tuffaceous, carbonaceous, micaceous, clay-rich mudstone, and very fine-grained, feldspathic, clay-rich siltstone of the lower Miocene age Northrup Creek Formation. The soil clay fractions contain up to 90% smectite with indications ofhalloysite. This earthflow is a reactivation ofa 650-year-old landslide (C-14 dating of uncovered buried trees). The failure mode is examined using a Janbu slope analysis and includes double wedge failure near the headscarp. High soil pore water pressure is one of the major causes of this slope failure. Rainfall levels for October, November, and December of 1994 were twice the previous five-year average. Present day groundwater level within the basin is less than one meter below ground surface. The earthflow is partially controlled by two faults of regional extent that dissect the basin near the headscarp in NW-SE and NE-SW directions. The Inceptisol soils in the basin remain moist below 20 cm year around. Soil in the basin may have been further weakened due to loss of root strength following timber harvest on the site in 1991. Soil liquid limits range from 42% to 95%, with PI values ranging from 2% to 77%. Soil clay content ranges between 18% and 30%. Direct shear tests on the mudstone and siltstone bedrock in both drained and undrained conditions produced internal friction angles of 14-18°, with cohesion values of 4 - 8 kPa. Back calculation of study area soil strength using the modified Bishop method results in a residual friction angle of 20.7°. The failure mode ofthe earthflow is from the headscarp downward and is modeled using Janbu methods. The study includes a detailed topographic map and a failure analysis of the earthflow basin.

LiDAR (Light Detection and Ranging) elevation data were collected in the Panther Creek Watershed, Yamhill County, Oregon in September and December, 2007, March, 2009 and March, 2010. LiDAR derived images from the March, 2009 dataset were used to map pre-historic, historic, and active landslides. Each mapped landslide was characterized as to type of movement, head scarp height, slope, failure depth, relative age, and direction. A total of 153 landslides were mapped and 81% were field checked in the study area. The majority of the landslide deposits (127 landslides) appear to have had movement in the past 150 years. Failures occur on slopes with a mean estimated pre-failure slope of 27° ± 8°. Depth to failure surfaces for shallow-seated landslides ranged from 0.75 m to 4.3 m, with an average of 2.9 m ± 0.8 m, and depth to failure surfaces for deep-seated landslides ranged from 5 m to 75m, with an average of 18 m ± 14 m. Earth flows are the most common slope process with 110 failures, comprising nearly three quarters (71%) of all mapped deposits. Elevation changes from two of the successive LiDAR data sets (December, 2007 and March, 2009) were examined to locate active landslides that occurred between the collections of the LiDAR imagery. The LiDAR-derived DEMs were subtracted from each other resulting in a differential dataset to examine changes in ground elevation. Areas with significant elevation changes were identified as potentially active landslides. Twenty-six landslides are considered active based upon differential LiDAR and field observations. Different models are used to estimate landslide susceptibility based upon landslide failure depth. Shallow-seated landslides are defined in this study as having a failure depth equal to less than 4.6 m (15 ft). Results of the shallow-seated susceptibility map show that the high susceptibility zone covers 35% and the moderate susceptibility zone covers 49% of the study area. Due to the high number of deep-seated landslides (58 landslides), a deep-seated susceptibility map was also created. Results of the deep-seated susceptibility map show that the high susceptibility zone covers 38% of the study area and the moderate susceptibility zone covers 43%. The results of this study include a detailed landslide inventory including pre-historic, historic, and active landslides and a set of susceptibility maps identifying areas of potential future landslides.

Debris flows, common disturbances in many mountainous areas, initially scour or bury stream habitats; however, debris flows deliver vast amounts of wood, boulders, and gravel that may ultimately form complex stream habitat to potentially support a diverse salmonid assemblage. The materials deposited by debris flows would otherwise be inaccessible to streams, and thus deposits may play an important role in creating and maintaining complex salmonid habitat over time. Despite the potential of deposits for increasing habitat complexity, most fish studies have focused on the destructive effects of debris flows and short-term recovery and re-colonization in scour zones. Debris-flows that occurred during the record-setting winter storms of 1996 in western Oregon, USA, provide an opportunity to study intermediate-term effects of debris-flow deposits on abundances and habitat for juvenile salmonids. In this setting, I surveyed salmonid abundance and habitat in three Oregon Coast Range streams that contained several debris-flow deposits from the 1996 storms. I explained fish abundance using hierarchical models, accounting for heterogeneous detection probabilities with repeated counts from multiple-pass snorkeling. The "best" hierarchical model of detection probability and abundance was selected (QAIC) from pool and snorkel-pass characteristics separately for juvenile coho salmon (Oncorhynchus kisutch), age 0+ trout, and age 1+ trout (Oncorhynchus spp.) in each stream. Adding distance to the nearest 1996 debris-flow deposit (DDF) produced a significant drop-in-deviance for four of nine "best" models, including at least one in each stream and for each species/age-class. In these four models, salmonid abundance decreased with increasing distance from deposit. As a potential explanation, several pool habitat characteristics were correlated (Spearman's rank) with DDF. Results varied across streams, but generally, percent of substrate as bedrock was lower and boulder density and percent substrate as gravel were higher closer to deposits. Although repeat counts are increasingly used in hierarchical modeling of heterogeneous detection probabilities and abundance for other wildlife species, studies of fish often rely on uncalibrated, single-pass snorkel counts. When exploring the value of repeat counts, I found that juvenile salmonid abundance decreased with increasing distance from debris-flow deposits in more multiple-pass hierarchical models that accounted for heterogeneous detection probabilities than for single-pass models that did not. Thus, modeling heterogeneous detection probabilities with repeated snorkel counts may be beneficial in other situations, addressing limitations of uncalibrated indices without relying on methods such as electrofishing, which may be difficult or impossible for remote study areas, longer surveys, or sensitive species. My findings suggest that debris-flow deposits may influence salmonid abundances after 15 years, and support management of debris flow-prone hillslopes and low-order channels to deliver elements of stream habitat complexity. / Graduation date: 2013

Pathogens, such as Escherichia coli and fecal coliforms, are causing the majority of water quality impairments in U.S., making up ~87% of this grouping's violations. Predicting and characterizing source, transport processes, and microbial survival rates is extremely challenging, due to the dynamic nature of each of these components. This research built upon current analytical methods that are used as exploratory tools to predict pathogen indicator counts across regional scales. Using a series of non-parametric methodologies, with spatially explicit predictors, 6657 samples from non-estuarine lotic streams were analyzed to make generalized predictions of regional water quality. 532 frequently sampled sites in the Oregon Coast Range Ecoregion, were parsed down to 93 pathogen sampling sites in effect to control for spatial and temporal biases. This generalized model was able to provide credible results in assessing regional water quality, using spatial techniques, and applying them to infrequently or unmonitored catchments. This model's 56.5% explanation of variation, was comparable to other researchers' regional assessments. This research confirmed linkages to land uses related to anthropogenic activities such as animal operations and agriculture, and general riparian conditions.

Calcium (Ca) is an essential macronutrient that is increasingly recognized as a biogeochemical factor that influences ecosystem structure and function. Progress in understanding the sustainability of ecosystem Ca supply has been hampered by a lack of information on the various forms and pools of Ca in forest ecosystems. In particular, few studies have investigated the role of Ca-oxalate (Ca-ox), a ubiquitous and sparingly soluble biomineral formed by plants and fungi, on Ca cycling. I investigated Ca-ox pools in two young Douglas-fir forests in the Oregon Coast Range, and found that Ca-ox comprised 4 to 18% of total ecosystem Ca in high- and low-Ca sites, respectively, with roughly even distribution in vegetation, detritus and mineral soil to 1 m depth. The proportion of ecosystem Ca existing as Ca-ox varied by ecosystem compartment but was highest in needle litterfall, foliage and branches. Calcium-ox could be a large amount of Ca in mineral soil; across nine sites comprising a local soil Ca gradient, we found as much as 20% of available Ca in 0 - 10 cm depth mineral soil occurs as Ca-ox. Ca-ox was the dominant form of Ca returned from plants to soil, but disappeared as rapidly as bulk Ca from decomposing litter, suggesting an important pathway for Ca recycling. In mineral soil, Ca-ox was a larger portion of total available Ca in the low-Ca site, which had lower Ca-ox concentrations overall, suggesting that Ca-ox has limited potential to buffer against Ca depletion in forests where Ca is in shortest supply. I investigated foliar chemistry as a method for diagnosis of nutrient deficiencies in high and low-Ca sites where Ca varied inversely with soil nitrogen (N), and which had received fertilization with urea (for nitrogen, N), lime, and calcium chloride three years prior. Foliar vector diagrams suggested N limitation at the low-N site and N sufficiency at the high-N site, but did not suggest Ca deficiency at either site after urea, lime and Ca-chloride fertilization. The high-Ca site displayed 20-60 times higher concentrations of foliar Ca-oxalate than the low-Ca site, although this was unaffected by fertilization. Soil nitrification responded to both N and lime fertilization at both sites, suggesting that fertilization with N may stimulate nitrification that could accelerate soil Ca loss. I also investigated how Ca-ox may influence cation tracers such as Ca and strontium (Sr) ratios (i.e., Ca/Sr) and Ca-isotopes (⁴⁴Ca/⁴⁰Ca), which are used to identify sources and pathways of Ca cycling in ecosystem studies. Laboratory synthesis of Ca-ox crystals exhibited preference for Ca over Sr, and for ⁴⁰Ca over ⁴⁴Ca. In the field, discrimination between Ca and Sr was detected in bulk plant tissues due to Ca-ox accumulation, suggesting that Ca-ox accumulation related to tree Ca supply status could influence interpretations of Ca/Sr as a tracer of Ca cycling. I also found that standard methods of soil exchangeable Ca extraction could dissolve Ca-ox crystals and potentially contribute an additional 52% to standard measurements of exchangeable-Ca pools in low-Ca sites, thus complicating long-standing interpretations of available soil Ca pools and dynamics in many studies. Results of this work show overall that Ca-ox is found in large quantities in plants, detritus, and mineral soil in forest ecosystems, and is a more dynamic component of ecosystem Ca cycling than previously recognized. / Graduation date: 2012

Sediment deposit ages inferred from radiocarbon dating of stream bank material were used to estimate residence times of valley-floor deposits in headwater valleys of the Oregon Coast Range, USA. Inherited ages of radiocarbon-dated material, i.e., time between carbon fixation in wood and its incorporation in a sediment deposit, can result in over-estimation of the ages of those deposits and, hence, the residence times of sediment within those units. Calibrated radiocarbon dates of 126 charcoal pieces sampled from Knowles Creek were used to estimate the distribution of inherited ages in fourteen depositional units representing three deposit types: fluvial fines, fluvial gravels, and debris flows. Within a depositional unit, the inherited age distribution of a piece of charcoal was estimated by convolving its calibrated age distribution with that of the piece of charcoal with the smallest weighted-mean calibrated age (i.e., an approximation of a unit's date of deposition) within that unit. All inherited age distributions for a particular deposit type were then added and normalized to provide a probability distribution of inherited ages for that deposit type. Probability distributions of inherited ages average 688, 1506, and 666 yr for fluvial fines, fluvial gravels, and debris flow units, respectively. Curves were fit to inherited age distributions for each deposit type. These curve fits were then convolved with deposit age distributions (i.e., equal to calibrated age distributions of woody material sampled from stream banks) of samples from Bear Creek (Lancaster and Casebeer, 2007) to correct these deposit ages for inherited age. This convolution gives a corrected deposit age. In cases in which means of corrected deposit age distributions for an upper unit were older than those of a lower unit within a stratigraphic column, the upper sample’s corrected deposit age distribution was set to that of the youngest lower in the stratigraphic section. Convolution shifted individual deposit age distributions towards zero and increased their standard deviation by an average of 365%. However, convolution decreased the standard deviations of normalized probability distribution functions of deposit ages inferred from many samples from 1340 to 1197 yr, and from 471 to 416 yr for lower and upper reaches, respectively, of the Bear Creek valley in the Oregon Coast Range. Convolution decreased estimates of mean deposit ages from 1296 to 1051 yr, and from 308 to 245 yr for lower and upper reaches, respectively, of the Bear Creek. Estimates of percentages of basin denudation passing through each reach's deposit ("trapping efficiency") increased from 11.6% to 14.4%, and from 25.4% to 31.9% for lower and upper Bear Creek, respectively. However, basic shapes of residence time distributions and, thus, inferences regarding removal of sediment from the reaches did not change after deposit dates were corrected. Sediment residence times in the lower Bear Creek valley are exponentially distributed, which implies that all sediment has a uniform probability of evacuation from deposits, whereas the power-law-distributed residence times in upper Bear imply preferential evacuation of younger deposits and preservation of older deposits. Much of the sediment transported onto valley floors via debris flows is deposited, and then is evacuated over longer times. Volumes and residence times of stored sediment in these deposits at the transition from debris flow to fluvial evacuation, and their associated width of valley floors, vary throughout a network. Export volumes and frequencies from tributaries are controls on deposit volumes and may control valley widening of mainstem valley floors. In addition, closely spaced tributaries may exert composite effects on valley floor landforms. It is hypothesized that the volumes of sediment stored at confluences increases with contributing watershed area of tributaries to the point where tributary slopes are low enough to cause most debris flows to be deposited within tributary valleys instead of in the mainstem valley. In four ~1 km reaches with contributing watershed areas of 0.3 to 5.0 km², field surveys provided measures of width of valley floors and volume of deposits, and radiocarbon dating of charcoal provided residence times of sediment in these deposits. Mean residence times of reaches vary between 1.1 and 2.5 kyr. Exponential distributions fit to residence times within two of the reaches imply evacuation of sediment independent of deposit ages. Power-law fits to residence times of the other two reaches imply age-dependent evacuation of deposits. Distribution shapes of residence times, and their means, do not vary systematically with contributing watershed area of mainstems. Mean width of mainstem valley floors increases with contributing watershed areas of both mainstems and their respective tributaries. Volumes of sediment stored on the valley floor increase with contributing areas of mainstems, and these volumes at tributary junctions peaked at tributary contributing areas of ~0.1 km². Percentage of basin denudation entering storage decreases with contributing area of mainstem. This decrease may be due to increasing percentages of sediment supply via fluvial transport for larger watersheds, and much, if not most, of this supply routes through the system quickly. / Graduation date: 2012

Landscapes respond to external perturbations over a variety of timescales, including million-year tectonic forcing, millennial to decadal climate fluctuations, and minutes-long high intensity storms or large magnitude earthquakes. In mountainous regions, understanding the role of landslides in driving the hillslope response to these perturbations is paramount for understanding landscape evolution over geologic timescales and hazards over human timescales. Here I analyze the landslide-driven hillslope response over millennial to decadal timescales using a variety of tools and techniques (e.g. tephrochronology, lidar and air photo analysis, field and subsurface investigations, and seismic refraction) in the Waipaoa Basin (New Zealand) and Oregon Coast Range (USA). For the Waipaoa study catchment, pervasive landslides have been sculpting >99% of the hillslopes in response to >50 m of fluvial incision following the shift to a warmer, wetter climate after the Last Glacial Maximum (LGM) (~18 ka). Then, starting in the late 1800s, European settlement resulted in deforestation and conversion of >90% of the landscape to pastureland—spurring a rapid increase in landslide-driven erosion. To quantify the landscape response, I first reconstruct LGM and younger paleosurfaces using tephrochronology and lidar-derived surface roughness to estimate the volume, timing, and distribution of hillslope destabilization. From these reconstructions, I calculate the post-LGM catchment-averaged erosion rate (1.6 mm/yr) and determine that the timing of the initial hillslope adjustment was rapid and occurred by ~10 ka. Second, I quantify the rate and volume of historic hillslope degradation using a 1956-2010 sequence of aerial photographs, lidar, and field reconnaissance to map the spatial extent of active landslides, create a ‘turf index’ based on the extent and style of pastoral ground disruption, correlate that with downslope velocity, and calculate the average annual sediment flux. From the sediment flux, I calculate an erosion rate over the past ~50 years (~20 mm/yr) that is 10x greater than post-LGM. Lastly, in Western Oregon, I confirm that seismic refraction can determine the size (e.g. depth) and failure style of landslides in western Oregon—data needed to incorporate these poorly studied landslides into future landscape evolution or hazard models. This dissertation includes both previously published and unpublished co-authored material.