In North America, cottonwood (Populus) and other members of the family Salicaceae are considered to exhibit the classic colonization-competition trade-off. Adaptations that make them highly successful colonists in disturbance-prone floodplain environments appear to reduce their ability to compete for resources in more benign environments. Because cottonwood recruitment dynamics are so tightly coupled to the natural disturbance regime, river regulation has led to widespread decline in seedling establishment along the active channel. However, pioneer trees that require disturbance events for regeneration use a variety of strategies for persisting during periods of relative stability. The use of spatial refugia, while critical for population recovery of many mobile organisms, is generally not considered an important strategy for trees. Episodic channel abandonment in meandering river systems, and the subsequent infill and terrestrialization of the abandoned channel, has been recently highlighted as critical for maintaining the population of a key pioneer riparian tree species, Fremont cottonwood (Populus fremontii). While controls on seedling establishment along the active channel are predominantly abiotic, temporal changes in abandoned channels result in a shift toward a more physically stable and more competitive environment. How a species with such strong colonization traits can establish in abandoned channels, and for how long, are the main questions addressed in my dissertation.

In a controlled community mesocosm experiment (Chapter 1), I used field-informed gradients of substrate texture and herbaceous cover to test interacting effects of soil moisture and interspecific competition on first year cottonwood seedling survival. I found that primary controls on cottonwood seedling establishment switched from abiotic to biotic drivers as a result of the biogeomorphic development of abandoned channels. Like on the active channel, seedlings were strongly moisture limited in conditions immediately following channel abandonment, but competition became a more important determinant of survival in conditions representative of an older abandoned channel. However, I also found that cottonwood seedlings were better competitors than anticipated, and were able to survive in the more physically benign and competitive conditions as well as the more physically stressful conditions to which they are classically adapted. This suggests that abandoned channels provide conditions favorable for cottonwood establishment for a broad window of time. While my focus was on understanding mechanisms controlling seedling establishment within abandoned channels, my results clarify interactions between abiotic and biotic controls that are more broadly applicable within a meandering river corridor. My results also add to evidence that species lie along a competition-colonization continuum, and have implications for incorporating secondary recruitment locations into management and restoration of pioneer riparian forests.

The well-known interspecific trade-off of high-light growth in early colonizing species, versus low-light survival in species that are better resource competitors, leads to the logical conclusion that pioneer trees such as cottonwood are shade intolerant. Empirical evidence also finds seedlings rarely establish in vegetated areas. So how are cottonwoods able to establish in the more densely vegetated (i.e., shadier) environment of abandoned channels? I examined this question using a shade-cloth mesocosm experiment (Chapter 2), in which I considered interactive gradients of moisture and light. In dry ecosystems, sunlight can be considered both a resource and a stress, with shading reducing evaporative demand by maintaining a cooler understory microclimate. I found that shading resulted in reduced vapor pressure deficits and higher soil moistures, and a strong positive effect on first year cottonwood seedling survival in the Mediterranean climate of California. However, seedlings that survived to the end of the experiment showed decreased final biomass and root growth in shade. These results suggest that cottonwoods are much more plastic in their shade tolerance than has been previously assumed. The positive effect of shade on survival was observed regardless of soil moisture availability, whereas the positive effect of sun on growth was much stronger in wetter conditions, suggesting that soil moisture is the dominant limiting resource for seedling growth. I conclude that their high moisture requirement, along with their plasticity in shade tolerance, is what allows cottonwoods to successful establish in abandoned channels. I suggest that lack of understory recruitment along the active channel may have more to do with the fact that vegetated areas are typically higher and drier as a result of biogeomorphic feedbacks.

Based on current theory and previous empirical evidence, it was considered likely that cottonwood establishment would be limited to the period immediately following channel abandonment, when the abandoned channel retains elements of physical dynamism to which the species is well adapted. However, my experimental evidence of better competitive ability and more plastic shade-tolerance, particularly under conditions of high soil moisture, suggested that cottonwood establishment in abandoned channels may be supported for a much longer window of time. On the Sacramento River, California, I used a chronosequence (space-for-time) approach to understand patterns of cottonwood establishment as a function of time since abandonment and biogeomorphic stage (Chapter 3). I examined patterns in overstory community composition and cottonwood diameter size distribution as indicators of past establishment dynamics. I used tree ring analysis to sample the age structure and determine establishment timing. I addressed the main sources of error in the use of tree ring analysis for determining establishment age by cross-dating tree cores, and developing and applying correction-factors for cores potentially missing the earliest years of growth. I also quantified the uncertainty around my correction-factors using a Monte Carlo simulation approach, and propagated the uncertainty through my analyses. My results support a recruitment window that begins at channel abandonment, and consistently lasts ~ 20 years, regardless of site age. Thus, while channel abandonment is episodic, a cohort of trees always successfully establishes, and cohorts can continue to successfully establish for a period of approximately two decades. The duration of the recruitment window extends in time the spatial refuge provided by abandoned channels, and thus helps ensure continued persistence of the cottonwood population.