Simon McKirdy

The development of inspection protocols and mitigation strategies are critical components of plant and animal biosecurity measures. The inclusion of available detection data can greatly enhance the evidence base for these types of decisions. However, a key step in analysing these data is the choice of an appropriate statistical model. This paper focuses on determining an appropriate model for biosecurity border and post-border detection of non-indigenous species for Barrow Island, Australia under stringent biosecurity controls. This is a flagship biosecurity project that is under close national and international scrutiny. A range of standard models were compared, including standard and zero-inflated Poisson, negative binomial and log-normal alternatives. These models failed to adequately describe the key characteristics of the data, namely an excess of zero and single organism detections, a range of detections between two and a hundred organisms, and a few extreme values, ranging between 250 and 1000 organisms. Alternative models were explored, including: (i) modelling the censored data ignoring the zero and extremely large detections, (ii) a component where detections were modelled ignoring the zero detections and including outliers and finally, (iii) zero-inflated model with the complete data set inclusive of zero detections. The negative binomial based models consistently gave the best outcomes under different degrees of inflation or over-dispersion, but are limited in that they cannot provide information about the mechanisms underlying zero-inflation. A three component log normal mixture model was found to be the best fit as it addressed these issues. This study demonstrates the importance of model choice in analysing biosecurity data, and suggests that mixture models may be more appropriate than more standard distributions. The data set gathered at Barrow Island is, however, unique for biosecurity border and post-border detection biosurveillance. Given this, general inferences about the underlying phenomena made based on the model should be made with caution. Comparable datasets should be obtained and tested to validate the choice of model for this type of data.

Anthracnose in lupins, caused by Colletotrichum lupini, was first detected in commercial crops in Western Australia (WA) in 1996. This incursion of an exotic plant pathogen into Australia constituted a major biosecurity threat to the local lupin industry. The disease caught the industry unawares with the majority of cultivars at that time being susceptible and there were major issues with local plant pathologists and lupin agronomists having very little exposure to diagnosis and management of this disease. In 1997, the first major breakthrough was made when resistance to anthracnose was confirmed in several breeding lines and commercial cultivars of narrow–leafed lupins (Lupinus angustifolius), and landraces of Lupinus albus. These findings led to the release of cultivars with elevated levels of resistance to the disease. Important information on relative yield loss, critical seed infection levels, fungicide seed treatment and geographical risk factors have also been discovered through applied research leading to use of seed testing and registered fungicides for the control of early infection. In addition, a spatio–temporal model was developed to simulate the spread of anthracnose initiated by infected seed and other sources. The modelling has contributed to the formulation of strategies for management of lupin anthracnose. An extension campaign through field days, seminars and regular media exposure promoted the management package developed from the research.

Plant biosecurity is a global issue that continues to grow in importance as the volume of trade between countries and the number of people travelling increases. Australia is free from many of the pests and diseases that affect plant industries and natural environments in other countries. This freedom provides a competitive advantage to Australia as a major agricultural exporter reliant on its international reputation as a producer of ‘clean and green’ agricultural and food products. Australia also places a high value on protecting our unique environment and lifestyle for future generations. Plant biosecurity is essential to protect these values. Plant biosecurity is focussed on those pests (insects and plant pathogens) that are; not currently present in Australia, are present but not in all production regions of Australia and are being actively controlled, or those pests that represent a new threat as their biology has changed. Plant biosecurity can impact on food safety, food security, trade, market access, market development, production costs and, ultimately, the profitability and sustainability of plant industries. Incursions of new pests directly threaten the economic viability of Australia's plant industries, which have an annual farm gate value of over $18 billion and annually contribute over $12 billion to export income. Even the perception that a pest is present in Australian produce can have a rapid and negative impact on Australia's reputation as a producer of safe, quality food products. The Australian lupin industry is threatened by several pests including Sitona spp. and Uromyces lupinicola (lupin rust). Both pests would significantly impact on lupin production in Australia should they be introduced. To minimise the risk of entry and establishment of threats such as a Sitona sp. and lupin rust, research activities must cover the full biosecurity continuum, pre–border, border and post–border. Plant biosecurity is a continuum that draws together many different disciplines. It differs from plant protection in that it is risk management – being strategic for the future needs. This paper provides an overview of plant biosecurity from an Australian perspective with two case studies of serious biosecurity pest threats to the Australian lupin industry. The case studies explore the critical questions that need to be addressed when identifying the threat posed by a pest species. The paper also addresses the need for a high level of biosecurity awareness and reporting throughout the international lupin industry.

Australia's management of bushfires illustrates how we should respond to major incidents. With up to date weather information, GIS, models predicting hotspots, outbreaks and potential control lines on the time scale of hours to days, agencies have an enhanced ability to manage fires. However for other major incursions, such as emergency plant pest outbreaks, our technological ability and support is far less advanced. This project aims at investigating the use of the NASA Terrestrial Observation and Prediction System (TOPS) for management of Emergency Plant Pest (EPP) incursions in Australia. In theory, by combining the daily environmental and climatic parameters (soil moisture, soil type, temperature, light exposure, aspect, etc.) with the host’s biology, one can predict what the photosynthetic rate (in terms of gC/m2/day) or fitness of a crop is. By combining the crop fitness with pest biology and host parameters, predictive climate-based simulations can then lead to estimates of the stages of pest outbreaks and guide the selection of feasible and effective containment or management options. A large computational resource will be required to do these three way interactions for any large scale mapping in a reasonable time. NASA’s supercomputers will initially be used to assist the process of modelling photosynthetic rates or GPP for Victoria and southwest West Australian. Initially one or two defined EPPs, such as the Glassy-winged sharpshooter, will be piloted with the project aiming to produce a more generic template model for other pest species.