Advancing Early Drought Stress Detection in Eucalyptus Using Digital Phenotyping and Machine Learning

An important part of our research will involve characterising fully differentiated wood accurately, in order to make comparisons with the predictions of our models. To this end we’ll be making use of a range of tools, both for non-destructive evaluation and using samples from felled trees.

At EucXylo we are harnessing the power of on-site sensors and remote sensing systems to monitor and measure tree growth at the stand and tree level (see the IMPACT tab for more information). We make use of a wide range of environmental and growth sensors which are connected to the Internet of Things. We monitor soil and air conditions, tree growth and development and various ecophysiological variables. We make use of terrestrial laser scanning, as well as airborne LiDAR and various types of imagery to gain large-scale perspectives of developing trees. All these techniques renders big data sets, and there is always scope at EucXylo to find innovative ways to interact with such data.

Insights from our research are the basis by which we continually build and improve predictive models at multiple scales. Where our models are weak, or we do not understand how to express or characterise a process in the models, we will perform experiments to answer our question.

We are implementing these models in open-access languages/environments like Python, and harness the power of innovations like GEMS developed through our and the AgroInformatics partnership to bring everything together.

Not everything we are doing, or want to do, is currently possible. At EucXylo we are constantly developing approaches and exploring new frontiers to measure and model things. For example, we have developed new approaches using CT Scanning to visualise the cambium, and we have explored new image analysis techniques to extract information from images. In addition, we are currently exploring using confocal Raman microscopy to quantify lignin and cellulose in the cell walls of eucalypts.

There is fascinating research being done to understand the molecular basis of wood formation. Better understanding of things like what “gene switches” operate to determine cell fate are very interesting at EucXylo.  But beyond that: what mechanisms and processes come into play? Another interesting area of work is to understand the interplay between transcriptomics, proteomics and ultimately metabolomics, and how gene expression ultimately plays out in situ. We are focussing on these questions from different points of view, including consideration the effects of drought

Plenty of research has been undertaken in recent decades into the seasonal dynamics of xylogenesis in various important Northen hemisphere tree species. Analysis of timing of the onset of growth, and the durations of developmental stages, for example, provide valuable insights into how and why wood properties vary as they do over a season. But this kind of research in eucalypts is woefully lacking. At EucXylo, high-precision monitoring equipment in concert with detailed wood and cambial sampling techniques are used to obtain new and expanding datasets to understand how the differentiating xylem develops over time.

All of the xylogenesis responses which can be observed and quantified need to be seen and understood in the context of the whole plant. Several of our research projects are geared to better understand (and ultimately model) a wide range of ecophysiological processes and phenomena and how these feed back to wood property variation in eucalypts.

One area of interest at EucXylo is carbohydrate allocation and understanding how source-sink, or the maintenance of carbohydrate balances by other mechanisms, will influence tree allometrics and flows of sugar to the developing xylem. Another area of interest is plant hydraulics, not least because some Eucalyptus species are the tallest flowering plants in the world. Linking sap flow and stem size dynamics in Eucalyptus provides great insight into water movement and storage and is an important part of understanding xylem development in the team. Another area of focus is the effect of environmental stressors such as drought on xylem development. Several projects (completed and ongoing) in the team have already been based on this topic and the group is well geared towards performing thin-sectioning and wood anatomy analyses.

At EucXylo we are harnessing the power of on-site sensors and remote sensing systems to monitor and measure tree growth at the stand and tree level (see the IMPACT tab for more information). We make use of a wide range of environmental and growth sensors which are connected to the Internet of Things. We monitor soil and air conditions, tree growth and development and various ecophysiological variables. We make use of terrestrial laser scanning, as well as airborne LiDAR and various types of imagery to gain large-scale perspectives of developing trees. All these techniques renders big data sets, and there is always scope at EucXylo to find innovative ways to interact with such data.