Resource managers must often make difficult choices in the face of imperfectly observed and dynamically changing systems (e.g., livestock, fisheries, water, and invasive species). A rich set of techniques exists for identifying optimal choices when that uncertainty is assumed to be understood and irreducible. Standard optimization approaches, however, cannot address situations in which reducible uncertainty applies to either system behavior or environmental states. The adaptive management literature overcomes this limitation with tools for optimal learning, but has been limited to highly simplified models with state and action spaces that are discrete and small. We overcome this problem by using a recently developed extension of the Partially Observable Markov Decision Process (POMDP) framework to allow for learning about a continuous state. We illustrate this methodology by exploring optimal control of bovine tuberculosis in New Zealand cattle. Disease testing—the control variable—serves to identify herds for treatment and provides information on prevalence, which is both imperfectly observed and subject to change due to controllable and uncontrollable factors. We find substantial efficiency losses from both ignoring learning (standard stochastic optimization) and from simplifying system dynamics (to facilitate a typical, simple learning model), though the latter effect dominates in our setting. We also find that under an adaptive management approach, simplifying dynamics can lead to a belief trap in which information gathering ceases, beliefs become increasingly inaccurate, and losses abound.

Background

Theory suggests that individual behavioral responses impact the spread of flu-like illnesses, but this has been difficult to empirically characterize. Social distancing is an important component of behavioral response, though analyses have been limited by a lack of behavioral data. Our objective is to use media data to characterize social distancing behavior in order to empirically inform explanatory and predictive epidemiological models.

Methods

We use data on variation in home television viewing as a proxy for variation in time spent in the home and, by extension, contact. This behavioral proxy is imperfect but appealing since information on a rich and representative sample is collected using consistent techniques across time and most major cities. We study the April-May 2009 outbreak of A/H1N1 in Central Mexico and examine the dynamic behavioral response in aggregate and contrast the observed patterns of various demographic subgroups. We develop and calibrate a dynamic behavioral model of disease transmission informed by the proxy data on daily variation in contact rates and compare it to a standard (non-adaptive) model and a fixed effects model that crudely captures behavior.

Results

We find that after a demonstrable initial behavioral response (consistent with social distancing) at the onset of the outbreak, there was attenuation in the response before the conclusion of the public health intervention. We find substantial differences in the behavioral response across age subgroups and socioeconomic levels. We also find that the dynamic behavioral and fixed effects transmission models better account for variation in new confirmed cases, generate more stable estimates of the baseline rate of transmission over time and predict the number of new cases over a short horizon with substantially less error.

Conclusions

Results suggest that A/H1N1 had an innate transmission potential greater than previously thought but this was masked by behavioral responses. Observed differences in behavioral response across demographic groups indicate a potential benefit from targeting social distancing outreach efforts.

Understanding the socioeconomic drivers of biological invasion informs policy development for curtailing future invasions. While early 20th-century plant trade expansions preceded increased establishments of plant pests in Northern America, increased establishments did not follow accelerating imports later that century. To explore this puzzle, we estimate the historical establishment of plant-feeding Hemiptera in Northern America as a function of historical U.S. imports of live plants from seven world regions. Delays between establishment and discovery are modeled using a previously unused proxy for dynamic discovery effort. By recovering the timing of pest arrivals from their historical discoveries, we disentangle the joint establishment-discovery process. We estimate long delays to discovery, which are partially attributable to the low detectability of less economically important insect species. We estimate that many introduced species remain undiscovered, ranging from around one-fifth for Eurasian regions to two-fifths for Central and South America.