Graduation date: 1967

These two studies investigated maternal engagement and retention in a voluntary, home-visiting program. The program screened families at the birth of their first child for risk characteristics associated with poor child and family outcomes. Higher risk families were offered regular home visits and support for up to 5 years. In the first study, a two-level hierarchical general linear model (HGLM) was used to examine the impact of poor community health and maternal isolation on mother's active engagement in the program, following initial enrollment. Data came from 4,057 mothers with firstborn infants, who enrolled in the Oregon Healthy Start (OHS) home-visiting program from 1995 through 1998. At the time of this study OHS was operating in 15 Oregon counties. Results showed that living in a county characterized by poor community health, or maternal isolation from supportive family and friendship networks, significantly reduced the likelihood of mothers actively engaging in home visits after enrollment. Hispanic mothers were significantly more likely to engage in services than mothers of other ethnicities. In the second study, a three-level hierarchical general linear model (HGLM) was used to investigate the impact of community violence, home visitor attributes, and maternal attributes on maternal retention in the OHS home-visiting program. Data came from 1,093 mothers with firstborn infants, served by 71 home visitors, residing in 12 communities being served by the OHS program prior to February 2000. Results showed that mothers who lived in communities that displayed higher levels of community violence were less likely to remain in the OHS program. Mothers were more likely to remain in the program when served by home visitors who received more active supervision. Mothers who were older were more likely to remain in the OHS program than were younger mothers. Hispanic mothers were more likely to remain in the OHS program than were mothers of other ethnic groups. Each study presents the implications for programmatic application. / Graduation date: 2002

The oceanic phenomenon of upwelling along the Oregon coast is examined. Upwelling in both the open ocean and coastal regions is discussed. An idealized model is used, envisaging the ocean off Oregon to consist of homogeneous surface and deep layers separated by a pycnocline. The equations of motion are solved to yield the vertical velocity at the base of the surface layer. A comparison is made between the model and results inferred from hydrographic data. In the open ocean region qualitative agreement is observed between the wind stress curl and the depth of the surface layer. Geostrophic meridional transports relative to the 1000 decibar surface were computed and found to be of the order of the uncertainty. In the coastal upwelling region surface layer zonal transports were computed from the meridional component of the mean wind stress and compared with values inferred from oceanographic data. Coastal upwelling along the Oregon coast is clearly associated with the northerly (longshore) component of the wind stress. / Graduation date: 1964

In Oregon, "Incompetent to Stand Trial (1ST) Patients" were observed to be increasing in number, remaining in the hospital longer, and costing more to treat. A study was designed to investigate variables that could be used to predict their length of stay at Oregon State Hospital. Data for thirteen independent variables (gender, age, having an Axis I psychosis level diagnosis, having an Axis I substance-related diagnosis, having an Axis II personality disorder diagnosis, evidence of involuntary medications, being on atypical medications at discharge, number of seclusion and restraint events, number of felony charges, number of misdemeanor charges, and number of inter-ward transfers) and one dependent variable (length of stay) were analyzed for 198 1ST patients discharged from Oregon State Hospital between January, 1999 and December, 2001. Bivariate correlations for all variables, and length of stay (LOS) means for all levels of each variable were examined and discussed. A standard multiple regression analysis was performed. The regression model accounted for 36.5% (32.7% adjusted) of the variability in (log) LOS. R for regression was found to be significantly different from zero. Five variables were found to be significant contributors to explaining the variability in (log) LOS: (square root) number of inter-ward transfers (16%), gender (5.8%), evidence of involuntary medications (5.2%), (square root) number of felony charges (2.8%), and (square root) number of seclusion and restraint events (1.6%). Despite accounting for more variability in LOS than several previous studies with psychiatric patients, 67.3% of the variability was unaccounted for by the regression model. Unstandardized regression coefficients for untransformed variables were interpreted, revealing that gender, number of inter-ward transfers, and evidence of involuntary medications significantly predicted the largest increases in LOS. Recommendations were made for further research related to LOS of 1ST patients. / Graduation date: 2003

While the general link between adolescent maltreatment and adolescent substance abuse is well established, questions remain regarding factors that may explain variability in the strength of this link. This study examines whether the link between maltreatment and substance abuse varies across three categories of maltreatment: physical, sexual and emotional. It also examines gender differences and timing of abuse (past or current). This study also examines the level of parental monitoring and what effect parental monitoring has on maltreated adolescents and their use of substances. Survey data were collected from six public schools in a rural Southwestern Oregon County on experience of maltreatment (emotional, physical and sexual) and use of substances (tobacco, alcohol and illicit drugs), gender differences and level of parental monitoring. Data were analyzed using t-tests and regression models. Results indicated that youth who currently experienced emotional, physical or sexual abuse were significantly more likely to use tobacco, alcohol, and illicit drugs than were youth who had experienced abuse in the past. They also indicated a significant interaction between gender and physical abuse, sexual abuse and emotional abuse (past and present combined) in predicting illicit drug use with males using illicit drugs at a significantly higher rate than females. As predicted, a significant interaction was found between parental monitoring and maltreatment/no maltreatment in predicting drug use. Maltreated youth with low levels of parental monitoring were significantly more likely to have high levels of substance use than were maltreated youth with high levels of parental monitoring. These results give rise to the need for further research into this subject to aid counselors in helping youth, especially male youth, in substance abuse treatment. Results of this study suggest a need for substance abuse treatment counselors to educate the parents of the treatment clients to closely monitor their adolescent as a possible means of reducing substance abuse. / Graduation date: 2002

A first-order approximation is made of the frequency and magnitude of bedload transport downstream of the Pelton-Round Butte Dam Complex, lower Deschutes River, Oregon with implications for salmonid spawning habitat. Field measurements of channel hydraulics, geometry, and particle size were combined with one-dimensional hydraulic and bedload transport models to determine the streamflow needed to move the channel bed material and to predict how much material would move once bedload transport was initiated. Inflows and outflows from the Dam Complex, and the predicted bedload carried by each, were compared to evaluate the effect of the Dam Complex on downstream discharge and bedload transport. Predicted critical discharges were equaled or exceeded very infrequently during the 72-year streamflow record. In addition, discharges capable of transporting bed material exceeded threshold conditions by very little. This has resulted in low predicted rates and amounts of bedload transport over the period of record. The Pelton-Round Butte Dam Complex was found to slightly increase the magnitude of flood peaks and associated predicted bedload transport for post-dam transport events exceeding threshold transport conditions. An exception to this pattern was the effect of the Dam Complex on the flood of December 1964, the second largest flood on record. This large flood occurred under unique circumstances -- while the reservoir of Round Butte Dam was filling for the first time. Since none of the transport events that occurred after the reservoir finished filling were affected by the Dam Complex in the same way as the 1964 event, the influence of the Complex on the 1964 flood is considered unique. The two largest floods on record (1964 and 1996), despite their relatively large sizes with respect to all other recorded flood events, did not cause major morphological changes to the lower Deschutes River. This observation, along with other independently derived information, supports the conclusion that the lower Deschutes is a very stable river. / Graduation date: 1998

Species associations of fifteen major commercial groundfish species in the northeastern Pacific ocean and their spatial and temporal characteristics were studied using Oregon bottom trawl logbook data, 1987 to 1993. Screening procedures were used to remove questionable data from the original logbook files, which resulted in the exclusion of information from 46% of the total available tows. Two multivariate methods, detrended correspondence analysis (DCA) and Ward's method of hierarchical cluster analysis were used to derive the association patterns of species and species groups. A general linear model that was developed for the primary DCA axis suggested that the species associations are strongly correlated with depth, but minimally correlated with the other environmental variables that were examined (latitude, season, and year). The weak correlations between DCA axis 1 and the temporal variables indicate that species associations in the study region are fairly persistent over time. The same multivariate techniques were used to examine possible sampling effects due to changes in the participating trawl vessels that contributed logbook information. Depth and latitudinal distributions of species occurrence in the logbook were similar to distributions derived from National Marine Fishery Service triennial bottom trawl survey. However, the analysis also showed that the depth coverage by the survey is not broad enough to accurately characterize associations among species that are currently subject to commercial fishing activity. / Graduation date: 1998

Few studies, and none in Oregon, have examined the presence and change of water quality parameters over time in popular natural swimming areas. This information is necessary to better understand water quality and risk of illness from either fecal contamination or cross-infection from other swimmers. The purpose of this study was to quantitatively measure bacterial and selected physical and chemical parameters, and collect background information for changes to the current state water quality criteria. Five natural swimming areas in Linn, Benton and Polk counties were chosen and sampled biweekly for physical, chemical, and bacterial parameters over a nine week period from June 28 to August 31, 1996. The results showed differences in bacterial levels over the sampling period which often varied by degrees of magnitude between sites. For example, the range in Escherichia coli levels was between 0 and 1000 colonies/100mL sample for two sites on the same sampling day. Similarly, the range in fecal coliform levels was between 5 and 500 colonies/100mL sample. The daily colony counts at each site exceeded the state standards at least 10% of the time for E. coli and 21% of the time for fecal coliform. At the most popular swimming site, Montieth Park. the fecal coliform regulatory levels were exceeded 79% of the time and E. coli levels were exceeded 42% of the time. This may be due to turbidity, high bather load, or a broken sewer line. The 30 day log mean of these values shows consistently elevated fecal coliform problems only at Montieth Park. For the other sites, the log means did not exceed the state and federal regulatory limits for fecal coliform or E. coll. This raises questions about which estimates should be used to assess public health risk. None of the other parameters in this study were correlated with bacterial counts, so it appears that none of these factors is solely responsible for elevated bacterial levels. Further testing should be done at Montieth Park to determine the cause of the elevated fecal coliform levels. Any follow-up studies should test several different indicator organisms in addition to E. coli for comparison and assessment of their relationship to public health risk. / Graduation date: 1997

Measuring habitat selection by bats is complicated by their intricate life cycle. Scaling habitat measurements to reflect a hierarchal habitat selection process can help to define habitat associations of bats. I assessed day roost habitat of female long-legged myotis at four scales: the roost structure, micro-habitat surrounding the roost, the stand level, and landscape level. I radio-tracked 16 female long-legged myotis (Myotis volans) for an average of eight days per bat, July through August of 1993, 1994, and 1995 in two drainages in the central Oregon Cascades (Quentin Creek and Lookout Creek) to locate day roost structures. Forty-one day roost structures were identified, of which 1 was a rock face, 4 were green trees, and 36 were snags. The average height of all roost structures was 40 m (SE=2.5). The average dbh for all snags and trees used as day roosts was 100 cm (SE=6.1). Large snags including partially live, hollow western redcedar trees (Thuja plicata) averaging 97 cm dbh (SE=6.6) and 38 m (SE=2.8) high were the most commonly used roost structures. Individuals radio-marked at the same night roost did not use one common day roost. Individual bats were found roosting in one roost for several days, or using multiple day roosts within discrete roost areas. The area which encompassed one night roost and all known day roosts covered 3,258 ha in the Quentin Creek drainage and 6,391 ha in Lookout Creek. I compared physical characteristics and habitat within 20 meters of 33 roost snags with 66 randomly selected snags. The odds that a snag is used as a day roost is associated with roost height; given height, the odds of use is associated with the height of the stand within 20 meters of the snag. There is some indication that the presence of an open canopy around the snag, and the percentage of bark on the snag also could be factors that influence the selection of snags as day roosts. The frequency of occurrence of roost structures within young and late seral stands did not differ from what was expected to occur by chance in these two stand conditions. Roosts did not occur in stands with a harvest history vs. stands without a harvest history disproportionate to availability. I compared the distance to class I (largest) through class IV (smallest) streams between 34 day roosts and 102 randomly selected points. Day roosts were located closer to streams than randomly selected locations in both Lookout and Quentin Creek drainages with 1 exception (Lookout class III). In two cases day roosts were significantly closer to streams than randomly selected locations. Day roosts tended to be closer to streams where night roosts were located than did randomly selected points, regardless of stream class. / Graduation date: 1997

This thesis includes four separate studies. The first two studies assessed edge effects in a 700-year-old forest. After 20 years of exposure, epiphyte assemblages on the clearcut edge were similar to those of the forest interior, but there were some differences in vertical distribution patterns. Several species restricted to the upper canopy of the forest interior occurred farther down in the crowns of trees on the clearcut edge. Many species were associated with thick moss mats. Thalli of two cyanolichen species were reciprocally transplanted among four tree crowns. Lobaria oregana grew less on the clearcut edge than in the forest interior. Populations of Pseudocyphellaria rainierensis had acclimated to the edge environment. The third study evaluated whether these two species require old growth and/or thick, underlying moss mats to achieve normal rates of growth and mortality. Thalli were transplanted into tree crowns in 13 forest stands representing 4 age classes: old growth, mature, young, and recent clearcut. Wooden racks were used instead of trees in clearcuts. Half of the cyanolichen thalli were transplanted onto thick moss mats, half onto bare bark. Both species grew at least as well in the younger forests as they did in old growth, but growth rates were significantly lower in clearcuts. Mortality rates were very low in young, mature, and old-growth forests but high in clearcuts. P. rainierensis grew significantly better on moss than bare bark. The fourth study evaluated the long-term potential of live tree retention for cyanolichen conservation. Lichen litterfall was sampled in a natural, multiple-age stand containing remnant trees and regenerating forest. Two lichen species (L. oregana and Sphaerophorus globosus) were strongly associated with remnant trees. Biomass of both species was highest near remnant trees and was significantly higher within groves of remnant trees than at the edges of these groves or near isolated trees. Cyanolichen populations appear to have persisted on remnant trees since before the last fire. They are slowly recolonizing the regenerating forest. Retention of live trees, including hardwoods, combined with longer rotation periods, has great potential to maintain cyanolichens in managed forests. / Graduation date: 1996