Accurate representation of surface energy partitioning is crucial for studying land surface processes and the climatic influence of land cover and land use change using coupled climate-land surface models. A critical question for these models, especially for newly coupled ones, is whether they can adequately distinguish differences in surface energy partitioning among different vegetation types. In the first chapter, I evaluated three years (2004-2006) of surface energy partitioning and surface climate over four dominant vegetation types (cropland, grassland, needleleaf evergreen forest, broadleaf deciduous forest) across the United States in a recently coupled regional climate model (WRF3-CLM3.5) by comparing model output to observations (AmeriFlux, CERES, and PRISM data) and to standard WRF output. I found that WRF3-CLM3.5 can capture the seasonal pattern in energy partitioning for needleleaf evergreen forest, but needs improvements in cropland, grassland and broadleaf deciduous forest.

To extend the capability of the regional climate model in studying the interaction of climate and agriculture, in the second chapter, I coupled a version of the Community Land Model that includes crop growth and management (CLM4crop) into the Weather Research and Forecasting model (WRF) and evaluated against multiple observations. The evaluation showed that although the model with dynamic crops overestimated LAI and growing season length, interannual variability in LAI was improved relative to a model with prescribed crop LAI and growth period, which has no environmental sensitivity. Improvements in climate variables were limited by an overall model dry bias. However, with addition of an irrigation scheme, soil moisture and energy fluxes were largely improved at irrigated sites. With this improved model, I further investigated whether the dynamic crop growth influenced the irrigation effects on climate. With prescribed crop LAI and growth, irrigation effects on climate were under-predicted in moderately irrigated regions. Moreover, relative to the dynamic crop growth version, the prescribed crop growth model underestimated irrigation water use and simulated much higher soil evaporation.

The third chapter is an application of the coupled model in studying the irrigation effects on heat waves. A potential decline in irrigation due to groundwater depletion would not only directly affect agriculture, but also could potentially alter surface climate. In this study I investigated how irrigation affects heat wave frequency, duration, and intensity using fifteen heat wave indices and a regional climate model. Across all indices, irrigation reduced heat wave frequency and duration, but increased intensity. Irrigation effects on heat waves are statistically significant over irrigated cropland and but not significant for non-irrigated regions. The magnitude of effect varies by index and is more sensitive to the choice of temperature metric than to the choice of temperature threshold. Regions experiencing strong groundwater depletion, such as the southern high plains, may suffer more and longer heat waves with reduced irrigation.

Overall, my research confirmed the dynamic crop growth model and irrigation are important in studying the agriculture and climate interaction. The research on irrigation effects, as well as on weather and climate prediction, should include dynamic crop growth and realistic irrigation schemes to better capture land surface effects in agricultural regions.

Future, warmer temperatures are predicted to increase alpine productivity, but few studies have addressed the role of water in constraining such responses. We tested the hypothesis that, in the absence of additional water during the growing season, warming may not increase community-level productivity by warming plots March-November and providing supplemental water during the snow-free growing season in an alpine plant community at Niwot Ridge, Colorado. We measured productivity responses to treatments at three levels of biological organization: community-, life form-, and species-levels in 2010-2012. Heating advanced snowmelt 9.4 ± 0.14 days (x ̅ ± sd) and subsequently decreased cumulative soil temperatures and increased cumulative soil moisture. Warming alone did not alter community-level productivity during any years of the experiment but warming with watering increased community-level productivity by 20% during the first year of treatments. Forb productivity increased with both warming and watering in all years of the experiment, while cushion productivity only increased with watering treatments after the first year of treatments. Graminoid productivity was insensitive to treatments and year, and while succulent productivity varied idiosyncratically by year and treatment, there were no overriding changes between treatments and controls. Responses at the species-level did not always follow the responses of their respective life form group in all or even a single year of the experiment, nor did they reveal compensatory responses that could fully explain productivity at higher levels of biological organization. For example, when the forb life form group responded to treatments in 2010, A. fendleri was the only species to exhibit a response. In 2011, both A. fendleri and G. rossii responded to treatments while, in 2012, G. rossii was the only species to respond to treatments. In our experiment, warming appears to indirectly increase soil moisture by advancing snowmelt, allowing species to break dormancy earlier and take advantage of early season moisture from a prolonged snowmelt. Our results also suggest that community-level responses mask life form group responses, as well as individual species responses within the community under future climate changes. Further, interannual climate variability strongly influences productivity at all levels of organization and is essential to account for when studying how climate change will influence alpine species, life forms, and communities.

Surviving and thriving requires minimizing energy expenditure while navigating (sometimes harrowing) environments in search of nourishment, mates, and shelter from predators and the elements. Associative learning is one of the tools available to organisms as they work to solve this problem, allowing them to efficiently use scents, sounds, and visuospatial features to predict the likelihood and location of rewards. We combine in vivo electrophysiology, complex behavioral paradigms, and statistical modeling techniques to explore the activity of the CA1 subregion of the hippocampus, which integrates a range of converging inputs that are thought to provide the sensorimotor, visuospatial, and affective information required to associate properties of the environment with inherently valuable stimuli. To organize these streams of information, the CA1 subregion must attenuate activity irrelevant to the organisms’ immediate goals, while simultaneously allowing select neuronal populations to respond to the appropriate inputs. This thesis consequently characterizes the activity patterns of inhibitory interneurons during associative memory processing. Interneurons work cooperatively with constellations of inputs to coordinate select subpopulations of pyramidal cells. In Chapter 2, I show that inhibitory interneurons selectively modulate their firing rate responses to odorants and spatial locations during associative memory processing, suggesting that interneurons can in fact be recruited by distinct, behaviorally relevant inputs. Chapter 3 documents the development of a statistical modeling strategy for identifying rapid shifts in interneuron spike timing relative to oscillatory synaptic currents. This tool facilitated the discovery of interneurons whose spike timing reliably evolved over the course of associative memory processing. Lastly, Chapter 4 investigates CA1 interneuron spike timing in response to distinct combinations of olfactory and visuospatial inputs. We apply the modeling approach described in Chapter 3 to characterize spike-phase relationships with respect to theta (4-12 Hz), low gamma (35-55 Hz), and high gamma (65-90 Hz) local field potential (LFP) oscillations. We show that entrainment into higher frequencies is predicted by theta phase relationships, and we find that interneuron spike-phase relationships can vary according to distinct combinations of inputs. These findings coalesce to offer insights on the role of CA1 inhibition in optimally integrating visuospatial, sensorimotor, and affective information to support associative memory.

Climate change, land use, and intensifying natural disturbance regimes are affecting forests globally. Earth system models (ESMs) must predict forest responses to these global changes to simulate future climate, hydrology, and ecosystem dynamics. The Earth system modeling community is increasingly using vegetation demographic models (VDMs) to simulate vegetation dynamics in ESMs. VDMs explicitly represent tree growth, mortality and recruitment, enabling advances in the projection of forest vulnerability and resilience, as well as evaluation with field data. Predicting future forests requires understanding if, where, and how forest communities will be resilient to global change through regeneration processes such as allocation to reproduction, dispersal, and seedling survival. Simulation of regeneration processes has received far less attention than simulation of processes that affect mature tree growth and mortality in spite of its critical role in maintaining forest structure, facilitating turnover in forest composition over space and time, enabling recovery from disturbance, and regulating climate-driven range shifts. The role of forest regeneration processes is particularly important in California where intensifying fire regimes and climate change are reducing conifer seedling recruitment and causing concern that vegetation regime shifts will push conifer forests towards more shrub-dominated ecosystems.

This dissertation advances empirical understanding and modeling infrastructure for predicting how forest regeneration processes will mediate changes in forest distribution, structure, and functional composition (i.e. “forest responses”) in response to global change. It is structured around three papers, two of which were previously published (Chapters 1 and 2). Chapter 1 critically reviews how forest regeneration processes are currently represented in VDMs and makes recommendations for advancing model algorithms and parameterizations. This critical review finds that regeneration processes are not currently represented sufficiently to capture how regeneration will mediate changes in future forest distribution, structure, and functional composition. There is a need to improve parameter values and algorithms for reproductive allocation, dispersal, environmental filtering in the seedling layer, and tree regeneration strategies adapted to wind, fire, and anthropogenic disturbance regimes. These improvements will require synthesis of existing data, specific field data collection protocols, such as long-term forest monitoring plots that identify individuals to species, and novel model algorithms compatible with global scale simulations.

Chapter 2 draws upon the insights and conclusions from Chapter 1 to develop a new modeling scheme, the Tree Recruitment Scheme (TRS), for simulating environmentally sensitive tree recruitment in VDMs. We evaluate the TRS by predicting tree recruitment for four tropical tree functional types under varying meteorology and canopy structure at Barro Colorado Island, Panama. By comparing TRS predictions to those of a current VDM, quantitative observations, and ecological expectations we find that it improves the magnitude and rank order of recruitment rates among four tropical tree plant functional types (PFTs). It also captures recruitment limitations in response to variable understory light, soil moisture, and changing precipitation regimes. These results indicate that adopting this framework will improve VDM capacity to predict PFT-specific tree recruitment in response to climate change, thereby improving predictions of future forest distribution, composition, and function. We implement the TRS in a leading VDM, the Functionally Assembled Terrestrial Ecosystem Simulator (FATES). Technical documentation for this implementation is provided in Appendix A.

Chapter 3 uses remote sensing and machine learning to analyze how changing climate and fire regimes in California are affecting long-term, post-fire conifer forest regeneration.. We quantify decadal-scale variation in post-fire conifer canopy recovery in 44 burned areas throughout California’s Sierra Nevada, Klamath Mountains, and Interior Coast Ranges and use generalized additive models to identify key drivers of recovery outcomes. The work in this chapter shows that post-fire conifer canopy recovery is reduced after higher severity fire, in areas that re-burned within 32 years, and when mean annual precipitation in the first three years after fire falls below 1,000 mm yr-1. This extends the spatio-temporal scope of prior work focused on shorter term post-fire seedling regeneration by providing insight into what drives more persistent changes in forest structure. It also provides a valuable dataset for benchmarking predictions of forest responses to changing fire regimes and supports prior hypotheses that increasingly intense fire regimes will negatively affect post-fire conifer regeneration, likely leading to an increase in the area dominated by shrubs.