Do an invasive organism's dispersal characteristics affect how we should search for it?

Abstract

We investigated how an invading organism's dispersal characteristics affect the efficacy of different surveillance strategies aimed at detecting that organism as it spreads following a new incursion. Specifically, we assessed whether, out of the surveillance strategies tested, the best surveillance strategy for an organism varied depending on the way it disperses. We simulated the spread of invasive organisms with different dispersal characteristics including leptokurtic and non-leptokurtic kernels with different median dispersal distances and degrees of kurtosis. We evaluated surveillance strategies with different sampling arrangements, densities and frequencies. Surveillance outcomes compared included the time to detection, the total spread of the invasion and the likelihood of the invasion reaching new areas. Overall, dispersal characteristics affected the surveillance outcomes, but the grid surveillance arrangement consistently performed best in terms of early detection and reduced spread within and between fields. Additionally, the results suggest that dispersal characteristics may influence spread to new areas and surveillance strategies. Therefore, knowledge on an invasive organism's dispersal characteristics may influence how we search for it and how we manage the invasion to prevent spread to new areas.

Accounting for spatially heterogeneous conditions in local‐scale surveillance strategies: case study of the biosecurity insect pest, grape phylloxera (Daktulosphaira vitifoliae (Fitch))

“…Recent studies have highlighted that temperature has a significant impact on the survival of different phylloxera strains . Additionally, our simulations assume that a pest is detected at a set detection probability; however, we acknowledge that sampling effort, expertise, method and timing will also likely impact the probability of detection and note that the variance of the model predictions will increase and decrease with low and high detection probabilities, respectively . Accounting for property history (e.g.…”

“…40 Additionally, our simulations assume that a pest is detected at a set detection probability; however, we acknowledge that sampling effort, expertise, method and timing will also likely impact the probability of detection 45 and note that the variance of the model predictions will increase and decrease with low and high detection probabilities, respectively. 46 Accounting for property history (e.g. absence of hygiene compliance or shared machinery with other properties) or incorporating visual assessments of plant health by experts may also improve surveillance.…”

Optimising risk-based surveillance for early detection of invasive plant pathogens

“…As well as not being able to consider optimal surveillance in isolation of control strategies, these approaches are generally less able to explicitly model pest introduction and spread in a realistic landscape, which limits their use in precise targeting of surveillance activities. Although some recent studies have considered the spatial characteristics of optimised surveillance in more detail, these have either been applied to simulated landscapes [41][42][43] or have been based on network modelling approaches 44,45 which are less able to capture spatial spread patterns. Although the current study is intended to explore optimal surveillance deployment rather than make concrete suggestions for implementation, further work will apply our method to current ongoing threats, such as the spread of Las in California 8 and X. fastidiosa in Italy 10 .…”

Simulation to investigate site-based monitoring of pest insect species for trade

“…Evaluating the effectiveness of a given surveillance design in the field is often not practical ( Venette et al 2002 , Caton et al 2021 ), and results may not be transferable. Modeling has been used to support area-wide surveillance ( Manoukis et al 2014 , Berec et al 2015 , Caton et al 2021 ), but site-based pest modeling has focused on integrated pest management ( Lux 2014 ) or the detection of new or emerging pests ( Triska and Renton 2018 ), not surveillance of established pests to support market access.…”

“…Simplification of movement means that diffusion does not represent individual trajectories, the center of the population remains the same ( Caton et al 2021 ), and the population continues to disperse ad infinitum so individuals are unable to respond to suitable landscape habitats. Surveillance strategies at smaller spatial scales are therefore commonly simulated by modeling pest movement as a random walk, parameterized to include responses to landscape and other environmental cues ( Lux 2014 , Triska and Renton 2018 ). An additional critical factor is lure attractiveness, which is generally parameterized to decay with distance from trap using a probability model (e.g., Manoukis et al 2014 ).…”