Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Australie → Do locusts seek greener pastures ? : an evaluation of MODIS vegetation indices to predict presence, abundance and impact of the Australian plague locust in southeastern Australia

University of Melbourne (2016)

Do locusts seek greener pastures ? : an evaluation of MODIS vegetation indices to predict presence, abundance and impact of the Australian plague locust in southeastern Australia

Weiss, John Edward Ritchie

Titre : Do locusts seek greener pastures ? : an evaluation of MODIS vegetation indices to predict presence, abundance and impact of the Australian plague locust in southeastern Australia

Auteur : Weiss, John Edward Ritchie

Université de soutenance : University of Melbourne

Grade : Doctor of Philosophy (PhD) 2016

Résumé partiel
The management of pests is complex and difficult. Biosecurity programs aim to reduce the risk of entry, establishment and spread of pests that affect social, agricultural and environmental values. Predicting the dispersal of pest insects or pathogens is critically important in managing and restricting their impact. Insects disperse to resources or hosts that are often distributed patchily and with variable quality. Host selection involves not only choosing the right species of plant, but also selecting an individual plant within that species that is, or will be, suitable for feeding, survival and development. Insects need to detect their host from a distance usually utilising visual or olfactory cues or both. Greens and yellows attract many phytophagous insects, although other wavelengths can also be attractive.
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 then predict the host’s photosynthetic rate (in terms of gC uptake/m2/day). The pest’s presence or abundance often relies upon the host’s plant photosynthetic rate. One approach to modelling or predicting the presence/abundance of a pest is then to use the photosynthetic rate or similar measure of growth of the plant as a surrogate for its suitability to a pest. By then combining this surrogate measure of suitability with a pest’s biology, climate-based simulations could predict pest outbreaks and help identify feasible and effective containment or management options.
NASA’s Terrestrial Observation Prediction System (TOPS) models daily global photosynthetic rates at a spatial resolution of 1 km. In this study, I examined the feasibility of utilising Moderate Resolution Imaging Spectroradiometer (MODIS) vegetation indices (as a potential component of TOPS) to model photosynthetic rates and primary productivity values of vegetation types for south-eastern Australia. I assessed the ability of these vegetation indices to measure the suitability of vegetation to particular pests – specifically the Australian plague locust. Locusts are known to differentiate between yellow and greens, and anecdotal information indicates that green colours and pastures attract the Australian plague locust. Initially, I investigated annual and seasonal variation of MODIS’s vegetation indices across eastern Australia at three different scales (tiles, regions and local areas). I showed that the values of the vegetation indices have high spatial and temporal variability. I determined that any prediction based on the condition of the vegetation has to be at both fine spatial and temporal scales. The locust incursion into Victoria during summer and autumn of 2010 – 2011 provided an ideal opportunity to test the suitability of MODIS vegetation indices as predictors of locust presence and abundance. From December to April, I surveyed 19 locations weekly that had pasture sites of various levels of green, for presence and abundance of two generations of adult locusts. I found strong spatial auto-correlation at two spatial scales (locations and sites), indicating that the distribution of adult locusts was spatially clustered. However, the values of the vegetation indices at the one-kilometer scale did not correlate with the density of adult locusts.
To examine correlates of distribution further and to have a larger data set spread over a greater area and a longer time, I examined 10 years of historical locust survey data for eastern Australia. With the larger set of data, I was able to assess how the vegetation indices averaged over scales of one, ten, and fifty kilometers correlated with adult locust occurrence and abundance. While the MODIS vegetation indices measured at a 50-km spatial scale showed the strongest relationship with the abundance of locusts, locust presence or abundance was not strongly related to any of the MODIS vegetation indices. When utilising the vegetation indices as a predictive tool for the 2008-2009 locust records in NSW, I found spatial accuracy varied widely from month to month and generally underpredicted locust abundance. Finally, I investigated another potential use of MODIS vegetation indices in locust management, the capacity to estimate the level of damage of larval feeding on primary production. Although aerial observers could easily discern the visual impact, neither of my remote sensing approaches were able to identify change in vegetation indices due to larval feeding.

Présentation

Version intégrale (41,96 Mb)

Page publiée le 30 juillet 2017