Maintaining the resilience of ecological systems in an era of global change is a priority for management and conservation. In California, forests are currently threatened by a suite of disturbances that include altered fire regimes, legacy effects from timber harvesting, a warming and drying climate, chronic air pollution, and uncharacteristically severe attacks by insects and pathogens. Managing to preserve the characteristic structure and function of California forests under novel disturbance regimes requires a clear understanding of these forests' historical conditions as well as an understanding of the drivers of change in these forests. A major challenge of managing for resilience is the lack of quantifiable metrics to assess changes in a system's resilience over time. This dissertation uses a multi-timescale approach that quantifies changes in the structure and composition of California mixed-conifer forests since European settlement and suggests a framework for measuring and monitoring forest resilience. This work can be used to guide conservation and restoration activities with the goal of maintaining the characteristic structure and function of forests under changing disturbance regimes. In Chapter 1, I explore the demographic responses that have led to a reordering of species dominance in Sierra Nevada mixed-conifer forests. California mixed-conifer forests have been subjected to a century of fire suppression, resulting in a shift in the structure and composition of these forests over time. Historically, a high-frequency, low-severity fire regime maintained structurally heterogeneous forests where dominance was shared among several conifer species. With the removal of fire from this system, forest density increased, as did the prevalence of shade-tolerant fir species at the expense of pines. Previous work suggests that species-specific differences in demography have contributed to a shift away from a heterogeneous, resilient forest to a monodominant forest that is more susceptible to catastrophic loss from fire, drought, or invasive pests or pathogens. However, these conclusions are typically derived from extrapolations from short-term data. I use a 57-year inventory record from an old-growth mixedconifer stand in the Plumas National Forest, CA, where fires have been excluded since the early 20th century. Using a Bayesian hierarchical modeling approach, I measure species-specific rates of mortality, recruitment, and growth over this 57-year period. I also correlated climate trends with demographic data to determine whether climate may be a driver of shifts in species composition. I found that basal area, density, and aboveground carbon have increased linearly over the 57-year period in spite of increasing temperatures, which I expected might have negatively affected growth. The recruitment and growth rates of Pseudotsuga menziesii (Douglas-fir) and Abies concolor (white fir) were significantly higher than the community-level means, while the recruitment and growth rates of Pinus lambertiana (sugar pine) and Pinus ponderosa (ponderosa pine) were significantly lower than the community-level means. Mortality rates were similar among species. These results indicate that differences in species-specific growth and recruitment rates are the main drivers of a shift towards a low-diversity forest system and may potentially lead to the loss of pines from mixed-conifer forests. These results also quantify the strong effect that fire has on the regulation of forest biomass and density in this system. In Chapter 2, I address the need for accurate understandings of historical forest conditions to be used as guides when implementing management and restoration plans. Because historical Sierra- Nevada mixed conifer forests were considered to be resilient to disturbance due to their heterogeneous structure and function, historical conditions are often considered to be the target state for restoration. However, multiple methods for estimating historical forest conditions are available and these methods sometimes give conflicting results regarding the density of forests prior to European settlement. The General Land Office (GLO) surveys of the late 19th and early 20th centuries provide data on forest structure across a broad geographic range of the western US. Distance-based plotless density estimators (PDE) have been used previously to estimate density from the GLO data but this approach is limited due to errors that arise when trees are not randomly distributed. Recently, an area-based method was developed in order overcome this limitation of distance-based PDEs. The area-based method relies on estimating the speciesspecific Voronoi area of individual trees based on regression equations derived in contemporary stands. This method predicts historical densities that are 2-5 times higher than previous estimates, and the method has not been independently vetted. I applied three distance-based PDEs (Cottam, Pollard, and Morisita) and two area-based PDEs (Delincé and mean harmonic Voronoi density (MHVD)) in six mixed-conifer and pine-dominated stands in California, US and Baja California Norte, Mexico. These stands ranged in density from 784-159 trees ha-1. I found that the least biased estimate of tree density in every stand was obtained with the Morisita estimator and the most biased was obtained with the MHVD estimator. Estimates of tree density derived from the MHVD estimator were 1-4 times larger than the true densities. While the concept of area-based estimators is theoretically sound, as demonstrated by the accuracy of the Delincé estimates, the Delincé approach cannot be used with GLO data and the extension of the approach to the MHVD estimator is flawed. The inaccuracy of the MHVD method was attributed to two causes: (1) the use of a crown scaling factor that does not correct for the number of trees sampled and (2) the persistent underestimate of the true VA due to a weak relationship between tree size and VA. The results of this study suggest that estimates of historical conditions derived from applying the MHVD method to GLO data are likely to overestimate density and that tree size is not an accurate predictor of tree area in these open-canopy forests. I suggest caution in using density estimates derived from the MHVD method to inform restoration and management in Sierra Nevada mixed-conifer forests, and recommend the Morisita estimator as the least biased of the distance-based estimators. In Chapter 3, I address the concept of resilience as it relates to forest ecology and management and outline a framework that can be used to determine quantifiable metrics of resilience. Resilience is an aggregate property of ecological systems that maintains the structure, function, and composition of the system when faced with a disturbance. The main challenge inherent in using resilience to inform management and conservation is the multitude of definitions and concepts that have been developed to describe the resilience of ecological systems. The framework I develop for operationalizing resilience builds on the theoretical concept of resilience but provides explicit metrics for measurement. In this framework, resilience is composed of two properties: resistance to disturbance and recovery from disturbance. I outline four dimensions of resistance and recovery that can be used to measure and monitor resilience, including heterogeneity, complexity, quality, and reserves. I dispense with the concept of strictly-defined alternate stable states and instead focus resilience goals on target states, which are determined by ecological, economic, recreational, or aesthetic considerations. I also conduct a literature review of papers which measure forest resilience to assess measurements and analyses that can be used to quantify the four dimensions of resilience in the context of resistance and recovery. The results of this review indicate that studies of resilience can effectively make use of simple methods for quantification and analysis and that the most compelling studies address both components of resilience (resistance to and recovery from disturbance) and multiple dimensions of resilience. I then apply metrics to quantify the dimensions of resilience in three case study systems: the Sierra Nevada mixed-conifer forest of California, the eastern hemlock forest of the northeastern US, and the northern hardwood forest of the northeastern US. I found that this resilience framework is limited by the fact that no single, absolute measure of resilience can be derived. However, the framework is useful for defining baseline resilience measures and establishing protocols for measuring relative changes in forest resilience over time.