APPF and University of Adelaide PhD student, Brooke Bruning, shares experiences from her recent trip to Phenome 2019 and Purdue University, USA.

Phenome Conference 2019  |  6 – 9 February 2019

Earlier this month, I was given the opportunity to travel to the USA to attend Phenome 2019 and present my PhD research on ‘Hyperspectral distribution maps to predict the concentration and spatial distribution of nitrogen and water content in wheat’. The 3rd annual Phenome 2019 conference was held at El Conquistador Resort, Tucson, Arizona from February 6-9, 2019.

The conference began with a field trip out to the Maricopa Agricultural Center, University of Arizona. We toured the research station where we saw a range of field buggies and the TERRA-REF field scanning system and sensors. The TERRA-REF high-throughput phenotyping field-scanning robot is the largest field crop analytics robot in the world and covers a ground area of 372m x 23.8m. It is a 30-tonne steel gantry that autonomously moves along two steel rails while using an array of sensors and cameras to continuously monitor crops below. The instrument box houses a 3D laser scanner, thermal IR, stereo-RGB, hyperspectral and chlorophyll fluorescence cameras as well as environmental sensors.

The TERRA-REF Field Scanner at the Maricopa Agricultural Center, University of Arizona.

The following day was the first day of conference sessions. There were seven single-track sessions across the remaining three days of the conference, plus an additional technology session showcasing the latest innovative tools for data collection, analysis and storage. With numerous speakers presenting their research on topics such as robotics and automated systems, novel proximal and remote sensing technologies, big data and computational phenomics, automated phenotyping at different scales, and computer vision and machine learning, there were talks to keep every attendee interested!

The posters were grouped into four subcategories: computational biology, data management, education and outreach, phenomics enabled biology, and sensors and systems. The sessions had less-defined themes and structures which worked well because it allowed a range of topics and styles (e.g. short talks, keynotes etc.) to be presented during each session. I gave my presentation on the first session of Day 1.  I only had five minutes to entice the crowd to come and view my poster to learn more about my research on hyperspectral imagery for nitrogen and water in wheat. While the talk went well, the real benefit of attending and presenting was the poster sessions. The three poster sessions allowed for detailed conversation and questions with those who were interested, more so than what was possible in a five minute talk.  My poster generated interest from a range of people, from those who had never heard of hyperspectral imaging to those who were working in similar areas.

The biggest benefit of attending Phenome 2019 was seeing and meeting the multi-disciplinary community of plant biologists, breeders, ecologists, engineers, agronomists and computational scientists that the conference attracted. Phenome was a rich and diverse networking environment and also served as a platform to further learn about world-wide plant phenotyping research, tools, technologies, and data analytics. As an additional bonus, on the final day of the conference, I received notification that I had been awarded a travel grant to cover the registration costs of the conference.

Purdue University visit  | 13 – 14 February 2019

While in the US, I also took the opportunity to visit Purdue University; specifically their state-of-the-art phenotyping facilities. I visited three groups who all specialise in different aspects and approaches to plant phenotyping:

  1. The Institute for Plant Sciences’ Controlled Environment Phenotyping Facility (CEPF)
  2. Agricultural and Biological Engineering Plant Phenotyping Lab Greenhouses
  3. Indiana Corn and Soybean Innovation Center (ICSC), Agronomy Center for Research and Education (ACRE)

My first visit was to the Controlled Environment Phenotyping Facility (CEPF) to meet with Dr Yang, the director of Digital Phenomics at Purdue University, and whom I briefly met at Phenome 2019 the previous week. The CEPF consists of a large Conviron growth chamber allowing for precise environmental control. The capacity of the growth chamber is 256 individual pots with a maximum height clearance of approximately 4m, ideal for the large number of corn studies they conduct. The automated conveyors allow for plant randomisation and transport to the precision irrigation and fertigation stations. The conveyors also transport the pots through the RGB, fluorescence and hyperspectral imaging chambers, which allow imaging of plants up to 4m high.

The Controlled Environment Phenotyping Facility (CEPF) at Purdue University. (Above) pots travelling along the conveyors towards the hyperspectral imaging system and (below) the Conviron growth chamber which can hold 256 plants up to 4m tall.

The CEPF was the first phenotyping facility that I had visited other than the one at which I am based (the APPF’s Adelaide node, The Plant Accelerator® at the University of Adelaide’s Waite Campus). Seeing the similarities and differences between the two facilities in their set-up, size and operation procedures was both exciting and beneficial to both parties involved, as was sharing our experiences, success stories and difficulties with our respective high-throughput phenotyping systems.

I also had the chance to visit Dr Jian Jin and the Agricultural and Biological Engineering Group’s Plant Phenotyping Greenhouses. The group has two different phenotyping greenhouses: an automatic high-throughput phenotyping greenhouse and a greenhouse with built-in imaging gantry. The automated phenotyping greenhouse has a capacity of 100 plants and contains an automatic watering station and a hyperspectral imaging chamber. The belt and pot randomisation within the phenotyping greenhouse has been optimised to remove >90% of micro-climate impacts and allows for 24/7 continuous monitoring of plants.

The in-greenhouse gantry system was built as an intermediate step between single-plant phenotyping and aerial phenotyping with the aim of simulating UAV (unmanned aerial vehicle) remote sensing platforms. Pots are placed below the gantry in rows (with similar spacing as would be found in the field) and can be imaged every 30 min with hyperspectral, thermal and 3D cameras. The gantry set-up is being used to collect high-frequency data (multiple acquisitions in a day) to investigate properties that affect remote sensing data but cannot otherwise be investigated with a single data acquisition – effects such as sun angle, shading, acquisition time, changes in plant phenology across the day, plant angle and orientation, etc.

The Greenhouse Facilities of the Plant Phenotyping Lab at Purdue University. (Above) the automatic high throughput phenotyping greenhouse and (below) the in-greenhouse gantry.

The final group at Purdue University that I had the opportunity to visit was the Indiana Corn and Soybean Innovation Center (ICSC). The Site Manager, Jason Adams, showed me around the 25,500-square-foot facility based at the Agronomy Center for Research and Education (ACRE). This research facility not only focusses on plant imaging, but also has technologies for plant processing, seed analysis, threshing and shelling, and data management. However, the main reason for visiting was to see Purdue’s field phenotyping platforms: the PhenoRover and their array of UAVs. The PhenoRover is a crop sprayer which they have modified into a semi-automated mobile, ground-based vehicle that carries an array of sensors capable of measuring numerous plant traits simultaneously. Its main use currently is for measuring biofuel crops for the identification of superior biofuel traits and breeding of next-generation biofuel crops, mainly sorghum. However, the PhenoRover can be adapted (i.e. by changing the sensors, track width, height, speed etc.) to accommodate a range of potential applications with various crops and traits of interest. Their fleet of UAVs, both fix-winged and quadcopters, has been designed and built by students to collect data on plant traits and characteristics. Each UAV has been developed for specific applications and therefore vary in size, battery life, operation and equipment (cameras, GPS, sensors, flight stabilisation, etc.) I found the visit to ICSC particularly beneficial because they operate at a field-scale. Although greenhouse based work can sometimes be scalable and applicable to outdoor environments, I still believe field phenotyping is important for crop improvement.  While greenhouse-based phenotyping may help understand the genetic background of plants, field phenotyping will help understand the environmental differences at the production scale, whether they be biotic and abiotic stress tolerance, enhanced nutrition or yield improvement.

The PhenoRover at the Indiana Corn and Soybean Innovation Center. The PhenoRover is a semi-automated, mobile, ground-based vehicle carrying a range of sensors and cameras (photo courtesy of Purdue Agriculture).

Overall, my trip to the USA exposed me to just how many different groups and disciplines are working in plant phenotyping – plant biologists, ecologists, engineers, agronomists, breeders, aviation specialists and computer scientists. High-throughput facilities, such as the Maricopa TERRA-REF system and the greenhouse and controlled environment facilities at Purdue University, bring together all these skill sets to address the most pressing problems in plant sciences and agriculture. I had the chance to see just how much research is being undertaken world-wide in plant phenotyping and how many facilities are dedicating resources for such research. As well as meeting a range of like-minded students, colleagues and phenotyping professionals from across the globe, my trip to the USA renewed my interest in plant phenotyping and exposed me to potential directions post-PhD.

Brooke Bruning
PhD student
The Plant Accelerator®, The University of Adelaide