Bayesian Networks to Compare Pest Control Interventions on Commodities Along Agricultural Production Chains

Abstract

The production of an agricultural commodity involves a sequence of processes: planting/growing, harvesting, sorting/grading, postharvest treatment, packing, and exporting. A Bayesian network has been developed to represent the level of potential infestation of an agricultural commodity by a specified pest along an agricultural production chain. It reflects the dependency of this infestation on the predicted level of pest challenge, the anticipated susceptibility of the commodity to the pest, the level of impact from pest control measures as designed, and any variation from that due to uncertainty in measure efficacy. The objective of this Bayesian network is to facilitate agreement between national governments of the exporters and importers on a set of phytosanitary measures to meet specific phytosanitary measure requirements to achieve target levels of protection against regulated pests. The model can be used to compare the performance of different combinations of measures under different scenarios of pest challenge, making use of available measure performance data. A case study is presented using a model developed for a fruit fly pest on dragon fruit in Vietnam; the model parameters and results are illustrative and do not imply a particular level of fruit fly infestation of these exports; rather, they provide the most likely, alternative, or worst-case scenarios of the impact of measures. As a means to facilitate agreement for trade, the model provides a framework to support communication between exporters and importers about any differences in perceptions of the risk reduction achieved by pest control measures deployed during the commodity production chain.

Invoice Financing of Supply Chains with Blockchain technology and Artificial Intelligence

“…closer to the sourcing of the raw products being processed and sold to customers. Using a combination of Bayesian Networks (BNs) [4], [5], which is an approach to Artificial intelligence (AI), and crosschain function calls across private blockchains [6] we can enable financiers to make informed funding decisions under uncertainty.…”

General Purpose Atomic Crosschain Transactions

“…For example, the authenticity of the product's claim of being organic or "Fair Trade" goods, and the place of production of the product, including the entire (or partial) production chain of the goods. The latter is especially important with respect to a systems approach to pest risk management that encompasses more environmentally friendly techniques [14]. The use of Atomic Crosschain Transactions in this scenario will enable the Product Provenance blockchain to request the compliance of a product which is purported to be organic.…”

Risk Analysis Frameworks Used in Biological Control and Introduction of a Novel Bayesian Network Tool

“…Another approach gaining favor in evaluating risk and management solutions for invasive species, and introduction of BCAs, is the use of BN models (Holt et al, 2018;Jamieson et al, 2013Jamieson et al, , 2016Mengersen et al, 2012;Wright et al, 2005). BNs can link physiological and ecological input variables using conditional probabilities to calculate posterior probabilities of specified outcome states, such as degree of potential spread of and injury by invasive species (Marcot, Hoff, Martin, Jewell, & Givens, 2019;Wyman-Grothem, Popoff, Hoff, & Herbst, 2018).…”