Rohan Sadler - Research Associate

School of Plant Biology (M090)
University of Western Australia
35 Stirling Highway
Crawley  WA  6009
AUSTRALIA

telephone     +61 8 6488 4257
fax               +61 8 6488 7925
email            rsadler@cyllene.uwa.edu.au

 
B.Sc.Agric., UWA (1993)

Academic Background

My undergraduate training was broad, from plant and animal biology to agricultural economics. Between 1999 and 2000 I completed a Masters of Science in Agriculture (Preliminary) where I concentrated on statistics and geographic information systems. My thesis was entitled “Perennial Veld Grass (Ehrharta calycina): The ecology of a persistent plant invasion in an urban bushland, Kings Park, Perth”. I began studying with the Ecosystems Research Group in March 2002.

Research Interests

• Spatio-temporal dynamics of vegetation systems
• Statistics for spatial processes
• Computational statistics
• Stability and resilience of ecosystems
• Long-term monitoring using remote sensing
• Metrics for landscape ecology

PhD Thesis – Spatio-Temporal Dynamics of Semi-Arid Grasslands: Quantifying Ecosystem Events

Supervisors: Dr Pauline Grierson (UWA, Plant Biology); Prof Martin Hazelton (Massey University, Institute of Inf. Sci. & Tech.)
Funding: Australian Postgraduate Award; Bushfire CRC
Submitted: December 2006

Assessment of the stability and resilience of ecosystems would be enhanced if the impact of a particular sequence of ecosystem events (fire, drought, grazing) on, say, vegetation productivity, could be accurately quantified and/or predicted. Our objective is to develop mathematical and modelling approaches to quantifying the impact of ecosystem events on vegetation change as part of an adaptive and long-term monitoring and management framework. In this manner my research increases our understanding of the long term role of bushfires  in ecological systems, a key step in managing bushfires so as to minimise their economic risk and impact whilst being ecologically sustainable in enhancing biodiversity and ecosystem functioning.

We are applying an ad hoc statistical modelling approach using Monte Carlo simulation to digital images of permanent monitoring plots of tussock grasslands in the semi-arid Pilbara region of north-western Australia. Data has been collected twice a year since 1996, and the data set comprises  1280 images to date. Image data can be characterised by a number of different image metrics. For example, a time series of images is reducible to a multivariate time series, enabling application of statistical computing techniques. Importantly, ecological theory can be used to generate spatio-temporal models of vegetation dynamics whose behaviour is dependent on ecosystem events such as fire, flood or drought. Models can be fitted to the data by applying the image metrics to simulated images generated from the models. In this way, competing ecological models or hypotheses can be tested. Currently we are working on showing:

1) How image metrics facilitating model fitting are selected;
2) How to simulate the spatio-temporal dynamics of vegetation systems using stochastic shapes which mimic shifts in vegetation patterning with ecosystem events such as fire; and,
3) How spatio-temporal models built on expert rules (and hence without an explicit mathematical distribution) can be fitted to data.

The output of validated models can be used for short term predictions of system behaviour, and is pertinent to exploring interactions among multiple ecosystem events. For example, the ecological impact of fire will be reduced dependent on the timing of previous fires systems, system biomass and the recent patterning of rainfall. It is hoped that empirical understanding of the impact of such interactions on vegetation dynamics will permit a rigorous comparison of alternative management scenarios in highly complex and variable systems.

Other Projects

• I have been moonlighting on work with my Bushfire CRC colleague Matthias Boer .  At the moment we are examining the linkage between power law behaviour in weather events and in fire frequency-size distributions. Other ideas are on the table, so watch this space!!
• Lately I have played around with state and transition models and remotely sensed data. Its part of my thesis but slightly on a tangent to the model fitting stuff. 
• I have also been organising statistics based workshops for the Montreal 2005 and Memphis 2006 Ecological Society of America meetings. This years meeting is “Multiple Linear Regression Using R”. 
• My backyard – a grey water system should be completed by the end of the 2006 year with chooks, vegies, fruit trees, frogs, lizards, and a bocce lawn.

Presentations

Sadler, R., Hazelton, M., and Grierson, P. (2005). Fitting ecological knowledge to remotely sensed long-term monitoring data: a framework from semi-arid grasslands of the Pilbara region of north-western Australia.  Proceedings of the 90th Annual Meeting of the Ecological Society of America (Montreal, Canada  7-12 August 2005).

Sadler, R., Hazelton, M., and Grierson, P. (2004). Spatio-temporal dynamics of Pilbara grasslands. Proceedings of the 89th Annual Meeting of the Ecological Society of America (Portland, Oregon  1-6 August 2004). Poster Presentation.

Sadler, R., Hazelton, M., and Grierson, P. (2004). Spatio-temporal dynamics of Pilbara grasslands. Proceedings of the 7th International Rangelands Conference (Durban, RSA, 26 July - 1 August 2003). Poster Presentation.

Book Reviews

Sadler, R. J. (2005) Rule breaking by building a frame for restoration. Conservation Biology 19 (3), 977.

Useful Links

• I am a convert of the statistical language R: it is completely free and does anything that you want it to (disregard the initial learning curve, its worth it!!).
• GRASS: a free GIS that I use with R. Vector attribute functionality and multisession capacity essential for applications on computing clusters is now available with version 6+.
• Get wise with managing your ecological data sets. For more information on ecoinformatics visit KNB
• Ecological Complexity: a journal created by Larry Li at the Ecological Complexity and Modelling Lab @ University of California, Riverside.
•Rangeland science at Jornarda Experimental Range, New Mexico , with special mention to my host of three weeks Jeff Herrick.
•The future of ecological science, the US's NEON project
•A very important person in my life – Deanna Rokich – who works at BGPA