UC Berkeley

Disturbance ecology is central to the understanding and management of Sierra Nevada mixed conifer forests (MCF). Three studies relying on field data and hierarchical statistical regression models illuminate relationships between pattern and process in this important forest type. In the first chapter, a suite of hierarchical spatial statistical models using Gaussian process spatial random effects is proposed to quantify fine-scale spatial patterns in the fuel load (biomass per unit area) of several wildland fuel components (duff, litter, fine woody debris, coarse woody debris, understory vegetation, trees, and saplings). A sampling protocol that generates spatially explicit fuel load information at a fine scale (sub-meter to tens of meters) is described and implemented in a Sierra Nevada mixed conifer forest affected by extensive mortality in the 2012-2016 drought. The statistical models are described, validated, and applied to test whether Sierra Nevada mixed conifer forests experiencing varying levels of drought mortality exhibit different fine-scale spatial patterns of wildland fuels. Model validation reveals varying performance in three tasks: 1) Making pointwise predictions of training or validation data, 2) reproducing the distribution of fine-scale (sub-meter to meter) fuel load observations, and 3) reproducing the distribution of coarse-scale (sub-hectare to hectare) mean fuel loads of the various fuel components. Models for the depth of duff, depth of litter, count of fine woody debris particles, and the size of coarse woody debris particles generally perform well in all three tasks and parameter estimates are well informed by the data. However, models for infrequent events such as the meter-scale presence of coarse woody debris, trees, or saplings do not perform well in terms of making pointwise predictions or learning from the data. There are mixed results from the models for the size of fine woody debris particles and the presence of understory vegetation. In general, forests experiencing different levels of drought mortality do not exhibit different fine-scale spatial patterns, with two exceptions. First, in the litter depths model the Gaussian process magnitude, controlling the relative strength of the spatial pattern, is greater on the low mortality plots than on the high mortality plots. Second, the Gaussian process length scale parameter for understory vegetation presence, controlling the distance at which spatial autocorrelation occurs, is higher in high mortality plots than in medium mortality plots. The sampling protocol and statistical analysis described in this chapter enable quantitative description and reproduction of the fine-scale spatial patterns of fuel loads, a prerequisite to predicting how fires will behave in fuel beds with varying fine-scale spatial properties. These models also facilitate study of the relationships between pattern and process by illuminating how parameters describing fine-scale spatial pattern vary in different contexts. In the second study, I apply similar hierarchical spatial models with Gaussian process spatial random effects to describe the spatial pattern of litter, duff, and fine woody debris both before and after three replicate prescribed fires in a Sierra Nevada mixed conifer forest. The analysis reveals that prescribed fire alters not only mean fuel loads, but also the fine-scale spatial pattern of biomass. Prescribed fire increased the relative strength of the fine-scale spatial pattern for litter and duff, 1-hour fine woody debris, and 10-hour fine woody debris. The burns decreased the length scale of the spatial pattern (the distance over which autocorrelation occurs) for litter and duff, increased it for 1-hour fuels, and did not change it for 10-hour fuels. Finally, the Gaussian process noise parameter describing very fine-scale autocorrelation increased for 1-hour fuels as a result of the prescribed burns. Changes to the fine-scale spatial pattern of litter, duff, and fine woody debris are likely to impact the behavior of future fires and the ecological function of these forests. As such, information about the effects of prescribed fire on the fine-scale spatial pattern of fuel loads is important to have a complete understanding of this crucial management practice. Finally, for my third chapter I assess how numerous stressors shape the vital rates (survival, growth, and fecundity) of sugar pine across the vast majority of its range. Sugar pine (Pinus lambertiana) is the largest Pinus species, an important timber species, and a component of several dry conifer forest types of western North America, in particular the extensive Sierra Nevada mixed conifer forest. The species faces several challenges in the Anthropocene, including a disrupted fire regime, an invasive pathogen, forest structure changes, and drought with ensuing bark beetle epidemics. Managers are concerned about the conservation outlook for sugar pine, but it is unclear where and how to best invest conservation resources. Using data from the US Forest Service’s Forest Inventory and Analysis program, I synthesize the vital rate functions by constructing an integral projection model which predicts the effects of various stressors on the asymptotic population growth rate. The asymptotic population growth rate is near or slightly below one even under undisturbed conditions, and the actual abundance (in terms of both stem density and basal area) slightly declined over the duration of the study (2001-2019). The analysis reveals that wildfire, white pine blister rust, and forest density are key drivers of the demographic rates of sugar pine across its range. Drought and site dryness had lesser, but still meaningful, effects. Fire has strong negative effects on survival, resulting in a strongly negative population trajectory on burned sites. Conversely, lower than average forest density (neighborhood basal area) results in a positive population growth rate via beneficial effects on individual growth. These results highlight the value of fire hazard mitigation, particularly where it also reduces forest density, in the conservation of this important species.