Prior to the arrival of HLB in a region large-scale surveillance programs are usually instigated in order to detect the disease as early as possible.  Early detection is necessary to minimise the impact of the disease and facilitate any containment or eradication interventions.  Large-scale surveillance surveys are however expensive, covering large geographic regions and stretching fiscal and manpower resources.  Available resources must thus be deployed in the most optimal way.  The choice of which locations within a region to survey is a complex problem since there may be hundreds of thousands of possibilities to choose from. Predicting how the epidemic will spread through a heterogonous landscape of citrus plantings and how this relates to where sampling resources should be deployed to find the ‘needle in the haystack’ is challenging and most surveys are consequently sub-optimal.  We bring together state of the art epidemiological modelling and stochastic optimisation techniques to determine the optimal pattern of sampling deployment across a landscape.  We find that the optimal pattern of sampling resources in a region is often counter-intuitive; for example simply targeting the highest risk locations is rarely the optimal course of action.  We show how the optimal pattern depends subtly on epidemiological factors such as the spatial pattern of citrus plantings and vector densities in a region.  We also show how geo-referenced information on likely entry points into a region, e.g. trade and travel hubs, can be incorporated to improve the probability of achieving early detection.

The detection of a new HLB epidemic in a planting often occurs when the disease has already reached high incidence.  This is problematic and once the epidemic has become established it is difficult to recover the economic productivity of the grove.  However, if new epidemics can be detected early enough then more can be done to control the disease.  Early detection requires that surveillance surveys be in place before the disease arrives; that is, a number of trees within a healthy grove should be inspected at regular intervals for symptoms of HLB.  Exactly how many trees should be surveyed and how frequently this should be done is a non-trivial problem and one that has not previously been addressed in plant pathology.  We present a theoretical method that relates the dynamics of an invading epidemic to the dynamics of a monitoring program.  The method determines exactly how an early detection survey should be designed in order to achieve a high probability of detecting an epidemic whilst it is at an early stage.  We compare the theoretical method to a complex simulation model which replicates the spatial and temporal dynamics of HLB in the field.  By running the model thousands of times we can make probabilistic predictions on early detection survey design that can be directly compared with the theoretical method.  We find striking similarities between the simple theoretical and more complicated simulation approach that enables us to make valuable new insights as well as deliver methods for transfer into practice.