Isoprene is the most abundant volatile organic compound (VOC) emitted from vegetation, mainly trees. Because it plays an important role in tropospheric chemistry leading to formation of pollutants and enhancing the lifetime of the greenhouse gas methane, concern about the response of isoprene emissions to the rise in atmospheric CO2 concentration and global climate change has been increasing over the last few years. The consequences of predicted climate change will have complex repercussions on global isoprene emission. The increasing atmospheric CO2 per se will have direct effects on terrestrial vegetation since CO2 is the substrate of photosynthesis. Because photosynthesis is limited by CO2 at current ambient concentrations, an increase in CO2 is expected to increase leaf biomass (i.e. isoprene emitting surface). Predicted warmer climate, extended drought periods, the possible shift in plant species in favour of isoprene emitters and the increase in length of growing season, may cause an increase in global isoprene emissions with profound perturbations of air quality and the global carbon cycle. The aim of this thesis was to investigate the effect of environmental variables such as light, temperature, drought and leaf-to-air vapour pressure deficit (VPD), and the short- and long-term effect of atmospheric [CO2] on isoprene emission from temperate and tropical tree species. Both leaf and whole ecosystem level fluxes were studied. At the leaf scale, a short-term experiment with leaves of potted two-year old trees of Quercus virginiana was carried out, exposing plants to two drying-rewatering cycles. Leaf isoprene emission fell, but the process was considerably less sensitive to water stress than photosynthesis and stomatal conductance. In drought conditions, the large reduction in photosynthesis caused the percentage of fixed carbon lost as isoprene to increase as plants became more stressed, reaching peaks of 50% when photosynthesis was almost zero. Isoprene emissions also showed a strong negative linear relationship with pre-dawn leaf water potential (psi-leaf). In another experiment carried out at the large enclosed facility of Biosphere 2 (B2L, Arizona, USA), studying isoprene emission from leaves of three-year-old plants of Populus deltoides grown at three CO2 atmospheric concentrations (430, 800 and 1200 mu mol mol-1 CO2) in non-stressed conditions, instantaneous increases in atmospheric [CO2] always resulted in a reduction of isoprene emission and a stimulation of photosynthesis. Moreover, in the long-term, the CO2 inhibition effect for isoprene emission became a permanent feature for plants growing under elevated [CO2]. Again, isoprene emission was less responsive to drought than photosynthesis. Both water-stress and high VPD strongly stimulated isoprene emission and depressed photosynthetic rate as a result of stomatal closure and the resulting decreases in intercellular [CO2] (Ci). This also led to a dramatic increase in the proportion of assimilated carbon lost as isoprene. The effect of atmospheric elevated [CO2] and its interaction with high VPD and water stress on ecosystem gross isoprene production (GIP) and net ecosystem exchange of CO2 (NEE) in the Populus deltoides plantations was also studied. Although GIP and NEE showed a similar response to light and temperature, NEE was stimulated by elevated CO2 by 72% and depressed by high VPD, while GIP was inhibited by elevated CO2 by 58% and stimulated by high VPD. Similar to what was observed at leaf level, under water stress conditions GIP was stimulated in the short term and declined only when the stress was severe, whereas NEE started to decrease from the beginning of the experiment. This contrasting response led the percentage of assimilated carbon lost by the ecosystem as isoprene to increase as water stress progressed from 2.5% and 0.6% in well-watered conditions to 60% and 40% for the ambient and the elevated CO2 treatments, respectively. Again, we found water limitation and high VPD off-set the inhibitory effect of elevated CO2, leading to increased isoprene emissions. The effect of a mild water stress on GIP and gross primary production (GPP) was also observed in the model tropical rainforest mesocosm of B2L. Although GPP was reduced by 32% during drought, GIP was not affected and correlated very well with both light and temperature. The percentage of fixed C lost as isoprene tended to increase during drought because of the reduction in GPP. Consumption of isoprene by soil was observed in both systems. The isoprene sink capacity of litter-free soil of the agroforest stands showed no significant response to different CO2 treatments, while isoprene production was strongly depressed by elevated atmospheric [CO2]. In both mesocosms, drought suppressed the sink capacity, but the full sink capacity of dry soil was recovered within a few hours upon rewetting. In summary, elevated CO2 increased biomass production and photosynthesis while depressing isoprene production. However, both drought and VPD may off-set the CO2 effect and lead to enhanced isoprene emission. We conclude that the overall effect of global climate change could be of enhancing global isoprene emissions while depressing the soil sink, and that the soil uptake of atmospheric isoprene is likely to be modest but significant and needs to be taken into account for a comprehensive estimate of the global isoprene budget.

Maritime forests are a critical interface between ocean and terrestrial ecosystems, providing important ecosystem services and functions. Along the U.S. southern Atlantic coast, maritime forests are dominated by <i>Quercus virginiana</i>. Maritime forests and <i>Q. virginiana</i> have been heavily impacted by conversion to agriculture, residential development, and pine stands. Southern pine beetle outbreaks have led to salvage and thinning silvicultural treatments of pine stands which offer an opportunity to restore more complex maritime forests. This research project is comprised of two experiments which allowed me to study the performance of planted <i>Q. virginiana</i> seedlings in response to (1) animal browse, (2) competing vegetation, and (3) varying overstory pine canopies. For both experiments, one-year-old bareroot seedlings were planted as split-plot experimental designs. The first experiment evaluated control of deer browse (fenced and not fenced whole plots) and competing vegetation (0, 1, and 2-yr vegetation control subplots) as independent variables. Overall seedling survival was 60% after two years. There was a significant interaction between deer browse and competing vegetation for seedling height, diameter, crown width, and lateral branches. Seedlings were larger for all response parameters when fenced with vegetation control. Vegetation control significantly improved seedling performance only in fenced plots, indicating a shift in pressure from herbivory to competition when deer were excluded. Foliar nitrogen (N) was significantly greater in fenced plots than non-fenced plots and in 2-yr vegetation control subplots than non-weeded subplots. The second experiment evaluated varying pine overstories (clearcut, heavy thin, light thin, and no thin whole plots) and competing vegetation control (0 and 2-yr vegetation control subplots). Overall seedling survival was 78% after one growing season, with clearcut plots at the greatest survival (83%) and no thin at the lowest (72%). Seedling growth and foliar nitrogen were significantly greater in clearcut plots followed by the heavy thin, light thin, and no thin plots. Vegetation control consistently promoted seedling height, but was only beneficial to diameter and crown width in clearcut/heavy thin plots. <i>Q. virginiana</i> seedlings demonstrated plasticity in their ability to acclimate to the varying microclimates created by silvicultural treatments, as demonstrated by light response curves, stomatal density, and specific leaf area. These results highlight the importance of fencing to remove deer browse, introducing light in the understory, and further improving seedling performance by removing competing vegetation.

Trees provide numerous benefits crucial to urban environments, yet poor growing conditions often prevent trees from reaching their genetic potential for growth, longevity, and ecosystem function. To overcome these limitations, greater understanding of tree growth in the urban environment is needed. The goal of this research project was therefore to characterize a broad suite of soil characteristics associated with urban tree plantings and evaluate their suitability for modeling physical dimensions and growth rates of urban trees. A series of observational studies and experiments was conducted on urban soils inhabited by two tree species (Zelkova serrata (Thunb.) Mikano and Quercus phellos L.) in Washington, DC and one tree species (Quercus virginiana Mill.) in Jacksonville, FL – two major metropolitan areas of the eastern United States with contrasting climate and soils. Characterization of urban soil attributes within cities revealed low variability for some properties (soil texture, pH, and certain plant nutrients with coefficients of variation (CV) below 0.5), but high variability (CV>1.0) for others (nitrate, ammonium, copper, and zinc). This is dependent on the location. These findings suggest that tree planting site evaluations may not require measurements for all soil properties and that representative sampling may be sufficient to accurately characterize most soil properties within a city. Field assessment of urban tree soils also revealed that conventional measures of soil compaction are difficult to obtain due to obstructions by roots and other foreign objects. To address the critical need for efficient and reliable assessment of soil compaction around urban trees, an experiment was conducted to develop bulk density estimation models for four common soil texture classes using soil strength and soil moisture as predictor variables. These models provided medium (0.42) to high (0.85) coefficients of determination when volumetric water content (VWC) was log transformed, demonstrating that measurements of soil texture, strength, and moisture can provide rapid, reliable assessment of soil compaction. Tree growth modeling focused on three response variables: canopy projection (CP), canopy volume (CV), and peak-increment-area age (PIA). To calculate PIA, tree-ring analysis was used to determine the age at which maximal trunk diameter growth occurred between transplanting and time of sampling. Because Q. virginiana has difficult-to-distinguish growth rings, an intensive tree-ring analysis of cores collected from these trees was conducted. The analysis revealed interseries correlation coefficients of up to 0.66, demonstrating that Q. virginiana can be aged with fairly high confidence in an urban setting. Empirical models developed for all three tree species using the suite of soil and site variables explained 25% – 83% of the observed variability in tree physical dimensions and growth rates. Soil pH was found to be a significant predictor variable for the majority of growth models along with nutrients such as Fe, B, Mn, and Zn, which are also associated with soil alkalinity. Models for PIA possessed the highest coefficient of determination, suggesting that measurements of soil conditions can be used confidently to predict the age at which growth rate subsides in these species. CV and CP were not predicted as well by soil-related variables, presumably because above-ground constraints such as pruning and building encroachment can affect canopy size without necessarily affecting growth rate. Certain prediction models for all three species included predictor variables with counterintuitive influences on tree growth (e.g., negative influences of soil depth on Q. phellos and soil volume on Q. virginiana), suggesting that either these urban trees are responding to these variables in a novel manner or that variables unaccounted for in these models (perhaps related to urbanization or high vehicular traffic) are concomitantly influencing tree growth. / Ph. D.

<p dir="ltr">Coastal ecosystems are critically important habitats for the services they provide on a global and local scale. Maritime forests are found within the southern Atlantic coast, and they serve as a boundary between the ocean and land. These forests stabilize coastlines, recharge groundwater, and provide a protective buffer against storm damage. Southern live oak (<i>Quercus virginiana</i>) was historically the dominant canopy species in maritime forests; however, previous land conversions to loblolly pine (<i>Pinus taeda</i>) plantations have shifted the abundance of loblolly pine to become the dominant canopy tree in maritime forests. Loblolly pines are fast growing, and they regenerate vigorously but they are not well adapted to coastal stressor. In recent decades, outbreaks of southern pine beetle (<i>Dendroctonus frontalis</i>) have provided restoration practitioners an opportunity to clear tracts of pine dominated maritime forest to restore live oak to the canopy. This research project is comprised of two experiments studying the performance of planted <i>Q. virginiana</i> seedlings on maritime forest restoration sites in coastal Georgia. The first experiment evaluated planting density (1-meter, 2-meters, 3-meters), mulch (with or without), and fertilizer (with or without). Overall seedling survival was 99% after four years. The application of fertilizer had an initial positive effect on seedling diameter after the first growing season. The application of mulch increased seedling height in the second to fourth growing seasons, diameter in third and fourth, and crown width in the fourth growing season. Planting density had no consistent effect over the first four years, and no biological significance was observed for foliar nutrient content. The second experiment examined eight different groupings of intra- and interspecific competition between <i>Q. virginiana</i> and <i>P. taeda</i> including: oak or pine alone; oak surrounded by oak, pine, or oak/pine; pine surrounded by pine, oak, or pine/oak at 0.5-m spacing between all seedlings. Two years after outplanting, survival did not vary by treatment. Oak centered competition plots were positively impacted by border tree height and diameter in year one and border height positively affected the center tree height in year two. Pine centered competition plots were positively impacted by border tree height in year one and year two. Oak centered competition plots with a mix of oak and pine on the border had significantly lower osmotic potential than other pine centric treatments after two years. Overall, oak centered treatments had lower osmotic potential than pine centered treatments. Ectomycorrhizal (EMF) species composition changed, and relative abundance increased from the initial planting to two years later but there was no variation between treatments and most EMF species were generalists. These results highlight the importance of mulch and fertilizer to reduce transplant shock and how competing seedlings can train seedlings to allocate photosynthate to shoot growth to help promote aboveground growth.</p>