You are here: Home >

data mining

The aim of this pilot project was to develop a statistical methodology for analysing the database so that insights into resistance development might be obtained.

Research outcomes:

The Australian phosphine resistance management strategy is underpinned by a national resistance monitoring program that provides continuous updates on the distribution and strength of resistance. This information is collected and stored in the Australian Grain Insect Resistance Database (AGIRD). This data warehouse now contains detailed information on the incidence of resistance in Australia for the past 20 years from thousands of sites monitored over this period.

The existence of AGIRD provides a unique opportunity for an analysis of the contribution of broader scale factors and practices to the occurrence of resistance that cannot be evaluated in local scale projects focussed on tactic development. However, the data in their current form are quite difficult to access and interrogate in a systematic way.

The aim of this pilot project was to develop a statistical methodology for analysing the database so that insights into resistance development might be obtained. We chose a limited set of the data from AGIRD with which to construct and test the methodology. The set chosen was detections of strong resistance to phosphine in all pest species over a 20 year period in Queensland. We analysed risk of resistance associated with factors such as storage type, site type, commodity, insect species, chemical treatment history and spatial-temporal elements.

Despite the limited geographical extent of the data analysis, the project demonstrated that very useful information relevant to improving resistance management strategies can be mined from AGIRD. For example, our analysis provided strong support for the implementation of resistance management tactics. The results demonstrate that the integrated use of alternative chemical treatments and physical methods such as grain cooling will significantly reduce the incidence of resistance.

Finally, this pilot project has demonstrated that the statistical methods developed could be used successfully to interrogate AGIRD. A whole database analysis is now feasible and this would allow much stronger trends and conclusions to be made. It would also allow for similarities and differences to be robustly identified for different regions of Australia and to determine whether the same biological and environmental factors, biosecurity practices and agricultural context are important. Also the modelling could be expanded to assess whether geographical movement of insects contributes to strong resistance.

Research implications:

Despite covering only a limited amount of data held in the AGIRD database, we are able to make a number of conclusions. Over the 20-year period, Strong Resistance (SR) was detected most frequently in the rusty grain beetle, Cryptolestes ferrugineus, (138 samples) followed by the lesser grain borer, Rhyzopertha dominica (46 samples), and much less frequently in the rice weevil, Sitophilus oryzae (9), red flour beetle, Tribolium castaneum (14) and sawtoothed grain beetle, Oryzaephilus surinamensis (10). SR occurred in storages with a history of phosphine use indicating that practices at particular storages are selecting for resistance rather than resistant insects invading from other locations.

Strong resistance in C. ferrugineus occurred most frequently in Central Queensland including the Emerald and Biloela districts (C), Kingaroy/South Burnett (SEB) and the region around Dalby (SEC) in storages not using fenitrothion. SR was not detected in storages using fenithrothion. Frequency of SR in this species was high in central storages and low on farm and at merchant premises, and higher in unsealed storages. As expected SR occurred in insects infesting commodities associated with the central handling system. This analysis revealed that incipient resistance was first detected in C. ferrugineus in 1992 and 1993 from single observations. There were then no detections of fully expressed strong resistance until 2007 and since that time detection rate has plateaued.

SR in R. dominica occurred in most regions but was a little more frequent in Burnett region. Detections were significantly less frequent in aerated and bunker storages. Use of pirimiphos-methyl was shown to be ineffective at controlling resistance in this species. Resistance was most frequent in central storages and on farm with relatively few detections from merchant premises. SR was first detected in this species in 1997 and there was a rapid increase in frequency to 1999 which was followed by a decline to 2004, another peak around 2010 and a recent decline.

Some conclusions could also be made about species with lower frequencies of SR. Strong resistance in O. surinamensis was highly associated with grain merchant premises and unsealed storages. Resistance in S. oryzae occurred in all sectors of the grain industry but was not detected on farm in Queensland. Use of fenitrothion appears to have prevented development of SR in T. castaneum as this resistance occurred only in storages where fenitrothion was not used and never where it was used.

Overall, it was clear from the results that each pest species must be considered individually when developing resistance management tactics as each can react in different ways to various treatments. As we improve our understanding of the biology and ecology of these insects we will be better able to target more effective management strategies.

Acknowledgements:

We thank Ms Hervoika Pavic and Dr Manoj Nayak for their assistance in preparation of the data for analysis.

PROJECT LEADER


Dr Patrick Collins
Project Leader CRC50177: Australian Grain Insect Resistance Database - data mining

p.collins@crcplantbiosecurity.com.au
Phone: 07 3255 4467
Fax: 07 3846 6371

Read More

PROJECT DETAILS

Status
Complete
Institution
Queensland University of Technology
Term
March 2011 - April 2012
Budget
$36,050

PROGRAM DETAILS

LOCATION

CORE CRC PARTICIPANTS