Bianca Eskelson

Data science, the principles that support extracting knowledge from data, has become increasingly important in natural resources management. This thesis applied both machine learning (innovative modelling) and statistics (conventional modelling) to address two ecological questions.In Chapter 2, Automatic bird sound detection: logistic regression based acoustic occupancy model, logistic models and convolutional neural networks were applied to predict bird presence/absence in audio recordings, in order to improve efficiency in analyzing large audio datasets. The acoustic recordings came from a bird sound detection challenge organized by the Institute of Electrical and Electronics Engineers (IEEE) and covered bird songs and calls in a wide range of environments along with the presence of noise. Based on leave-one-out cross-validation, the final logistic model resulted in an overall accuracy of 75% with a false negative rate of 16%. Compared with a convolutional neural network (CNN) model using the same dataset, the logistic model was about seven times faster in terms of processing time. This bird sound detection model using sound frequency percentiles in a logistic model opens up promising approaches to aid in automatic, accurate, and efficient analysis of large audio datasets for monitoring wildlife communities.In Chapter 3, Previous fire severity enhances reburn severity: a case study in interior British Columbia, Canada, an ordinal logistic model was applied to investigate how previous fires influenced the reburn severity in interior British Columbia, Canada, in order to determine the driver of reburn severity. Previous fires affect rates of fuel consumption and accumulation, thus influencing the probability and severity of subsequent fires. In this study, forest stand structural change due to the first fire (in 2009 or 2010), such as changes in basal area and trees per hectare, were used to model the severity of the reburn in 2017. The ordinal model indicated a positive relationship between fire severities in the Riske Creek area. Specifically, fires in the Riske Creek area might not be able to limit the probability or severity of a reburn after seven or eight years.

View record

Thinning treatments are an important management tool, as they help reduce competition and promote tree growth by increasing available resources in a stand (e.g., light, water, etc.). Thinning leads to differences in stand composition and structure, and this variation has been linked to forest productivity. Previous research has found that tree growth can be improved in some mixed species stands, if trees do not directly compete for the same resources. Reduced competition and improved productivity has been found in mixed species stands of Douglas-fir and western hemlock. Many studies on thinning effects have found that thinning improves average tree size and growth, but this does not provide insight into whether small or large trees benefit most from thinning. Using data from 22 pure and mixed Douglas-fir and western hemlock stands that were part of long-term thinning experiments, I analyzed how thinning (0%, 20% and 35% basal area removed) affects stand-structure dynamics and basal area growth in pure and mixed species stands over time. To understand how thinning affects size inequality—expressed by the Gini coefficient—and growth dominance over time, a linear mixed effects models was fit that included thinning and years since thinning as explanatory variables. Results found that size-inequality did not change over time and growth dominance was reduced in mixed species stands, indicating that mixed species stands may be more productive and all trees have improved growth efficiency. An individual tree analysis was performed to understand thinning and competition effects on tree basal area growth. The results indicate that basal area growth was highest in the largest trees. Results also show that inter-specific competition increases basal area growth of western hemlock trees. Both analyses found that mixed species stands resulted in improved basal area growth, likely through reduced competition. Forest managers may look to planting mixed species stands to improve forest productivity.

View record

Forest fires are a common disturbance agent throughout the Pacific Northwest (PNW) and affect stand structure, age, species composition, and carbon storage of the productive PNW forests. Fire regimes in the PNW are predicted to shift towards more frequent and severe fires with climate change, which has important implications for carbon storage in the region. This study examines how fire severity (defined by remote sensing) impacts forest surface carbon pools (duff, litter, and downed woody materials). These carbon pools store a high proportion of stand carbon, and they have short- and long-term impacts on ecosystem function and fire behaviour.I examined one atypically large and severe fire in coastal British Columbia to obtain baseline measurements of post-fire forest floor fuel carbon (duff, litter, woody materials of all sizes) in the region. I found that there were no differences in total surface carbon between burned and unburned plots, but there were less duff and litter fuels in burned plots. This study provides baseline data for studies of post-fire forest floor carbon dynamics in the Boulder Creek region.Data from the United States Forest Inventory and Analysis program were utilized to estimate regional wildfire consumption factors for forest surface carbon pools (duff, litter, fine woody materials) in Oregon and Washington that are representative of the current fire regime. While forest surface pools were consumed in the fire, there were no significant differences in consumption between fire severity classes. 30 – 40% of carbon in each pool were left behind, even after high-severity fire. This research provides both a case study and a regional study on the effects of wildfire on carbon in forest surface pools. Both types of studies provide information that is beneficial for the study of post-fire carbon, giving insights into landscape level impacts or single extreme events.

View record

Effective management of invasive plants preserves biodiversity values, reduces economic costs, and minimizes negative impacts on human well-being. Prevention, the most cost-effective approach to invasive plant management, focuses on predicting species occurrences in high-risk areas and fosters the public’s awareness of invasive plants. This study aims to contribute to the aforementioned prevention foci by (1) investigating the relationships between invasive plants occurrences and socio-economic, greenspace, topographic, and land use variables, and (2) assessing the public’s levels of knowledge, risk perception, and support for invasive plant management in Metro Vancouver, British Columbia, Canada.I utilized invasive plants inventory, land use, topographic, and socio-economic data to identify key drivers of species occurrences. The chances of invasive plants occurrences were higher in wealthier neighbourhoods. The relationships between species occurrences and the tested explanatory variables were different across municipalities. Greenspace type was a surrogate for median household income, gardening expenses, and greenspace area. The results can inform managers of key drivers of invasive plants occurrences in Metro Vancouver, which can ultimately aid in species occurrence prediction efforts. An online survey in Metro Vancouver assessed the public’s levels of knowledge, risk perception, and support for management activities. I found that the public’s perception of invasive plants was ecologically oriented and positively correlated with age and income. The public highly supported community events or the planting of native species. Overall, the public’s risk perception assessment provides managers with insights on which aspects of invasive plants are well-known and which management activities are preferred by the public.

View record