UC Davis

Many coastal marine invertebrates and fish begin life as planktonic larvae that can be transported several kilometers offshore by coastal currents. To have any hope of surviving to reproduction, larvae must avoid predators during dispersal and return to shore with enough energy for further development. Most larvae fail to do so, prompting Rumrill (1990) to begin his review of larval mortality by stating that coastal marine animal larvae "lead transitory lives of great risk and grave uncertainty." I examine, using stochastic modeling, how these risks and uncertainties are shaped -- and in some cases, overcome -- by the passive and active interactions of larvae with spatially varying features of their environments. Chapter 2 focuses on the consequences of increased larval mortality near the coast compared with offshore. The relative safety of offshore waters is often mentioned in discussions of the evolutionary origins of planktonic larvae, but is omitted by most modeling studies. I show that ignoring this feature may, in some cases, substantially alter predictions of coastal population dynamics and connectivity. Furthermore, oceanographic features that slow nearshore larval movement (such as coastal boundary layers) are double-edged swords, limiting the offshore loss of larvae but also preventing larvae from escaping nearshore hazards.

Chapters 3 and 4 illustrate how larvae improve their chances of success by slowly swimming vertically to exploit differences in current velocity, food abundance, and predation throughout the water column. In Chapter 3, I consider a broad, continuous set of behaviors for larvae dispersing in an idealized environment approximating the two-layer flow typical of upwelling circulation. I show that while some behaviors successfully increase feeding opportunities and alongshore movement or limit the fraction of larvae that are lost offshore, no behaviors I modeled achieve both at once. I speculate that the former class of behaviors is suitable for organisms that spawn many cheap, long-lived larvae that feed during dispersal, while the latter is preferable for organisms with fewer, more expensive larvae that cannot feed. I extend this analysis in Chapter 4 by using dynamic programming to construct behaviors that optimize a metric of success that balances delivery to coastal habitats, predation risk, and energy budgeting. I demonstrate that some behaviors observed in nature are optimal in specific conditions, and that there exist realistic non-optimal behaviors that perform reliably well as conditions are varied. I hypothesize that many behaviors observed in nature result from selection for success in both typical or static conditions as well as extreme or variable ones. More broadly, I emphasize in Chapters 3 and 4 that predictable spatial structure in the environment creates an opportunity for larvae to change their destinies, and that these changes are most evident when mortality and energetics are considered alongside larval delivery to coastal habitats.