We’re delighted to announce that the winner of the 2024 Harper Prize is Karina Guo! The Harper Prize is awarded annually for the best paper published in the journal by an early career researcher.
Karina Guo, William K. Cornwell, Jason G. Bragg (Journal of Ecology, 112, 2183–2197)
Determining the ecological and evolutionary relationships between functional traits and drivers of ecosystem effects and responses, such as climate, is one of the most active areas of research in ecology. Guo et al. find that strengthening of the trait-climate correlation for leaf area of Eucalyptus trees occurs just above the species level, indicative of adaptation at the inter-specific phylogenetic level rather than local adaptation within species. Guo’s paper is remarkable in being one of the first operational machine learning applications in trait ecology, demonstrating the potential to quickly and effectively automate the extraction of trait data from sources such as digitised herbaria. This work is also remarkable in that it emerged from an undergraduate research project, demonstrating the impact that research can have, even at very early career stages. The authors have made the data and code openly available and I have no doubt that this work will stimulate a step-change in the expansion of trait databases as well as tests of the ecological and evolutionary drivers of trait-climate relationships for species worldwide.
– Yvonne Buckley, Senior editor, Journal of Ecology.
I’m Karina Guo, now a doctorate student with the Research Centre for Ecosystem Resilience at the Botanic Gardens of Sydney and at the University of Sydney. Currently, I’ve pivoted away from Machine Learning and have dived deep into working on genetics and evolution in my current doctoral project. Specifically, I am working on Myrtle rust (Austropuccinia psidii) and its interaction with a foundational species in Australia, the Broad-leaved Melaleuca (Melaleuca quinquenervia). I’m working directly with practitioners to better understand how our research can be best applied in the field where restoration is occurring readily.
- Could you give us a bit of background about yourself and how you got into ecology?
As a kid, I was always that one that was “making potions” out of dirt, mud, plants at the playground – much to my parent’s annoyance when they saw the state of my clothes. As I grew older, I don’t think I ever stopped frolicking in the bushes. I’ve always been someone who would love to touch, observe, listen, and – also to the horror of my parents – taste the nature around me. Which made me that hiker that stops every 5 seconds to ogle at something. I couldn’t help it really, there was always something to learn and to love about it all. And ecology is all about that. Taking a minute to stop and listen and link the deep interweaving web of connections between everything. So, I don’t think I ever fell in love with ecology – I think I was always a part of it.
- Can you provide a few sentences that summarise the research in your paper and how it advances the field?
This paper uses machine learning to extract traits of leaves on the numerous species of gum trees (Eucalyptus, Corymbia, Angophora genera). Traditionally scientists measure leaf traits, such as leaf area, tediously by hand. This means that the number of measurements taken can be highly limited by funding and the labour available. By using machine learning to bypass this, I was able to extract leaf traits from all the historical pressed plant specimens stored in herbaria, with some sheets dating back as far as 1839. This meant that I was able to expand our trait datasets by 50-fold! The sheer size of this dataset painted such details that previous datasets lacked. From this I was able to interrogate questions that were previously out of reach. For example, for the first time I was able to ask how these traits varied across climates, within groups of trees of different evolutionary ages. Furthermore, as one of the first operational studies of machine learning in trait ecology, our workflow poses as an exciting prototype applicable to diverse taxa and traits for future ecological applications.
- Have you continued this research and if so, where are you at now with it?
Unfortunately, my research has taken me away from machine learning. But that doesn’t mean I’m not still fascinated by the potential it has! In all my research, the topic of it inevitably comes up. We’re always having conversations about it – bouncing around ideas on how machine learning can be used, either to open up new possibilities, improve our current methods, or streamline our usual tasks.
- What did you enjoy most about conducting this research?
Through this project I highly enjoyed refining our model to its final state. Making the machine learning model was an iterative process, and that meant I was able to in a sense, watch it grow. With each round of model optimisation, whether it was changing hyperparameters, refining the training data, or projecting it onto a testing sheet, I was able to see tangible improvements to the model. It really provided me with a deep sense of pride! We had some hiccups along the way, such as concerns with overfitting or finding a new eucalypt species with the most bizarre linear leaves, but that just added to the fulfillment of it all. And, when that was finished, the deployment of it across all the eucalypt herbarium sheets was truly exciting! I was streaming the model’s updates, watching it process each batch of herbarium sheets. Knowing that this would be the largest trait dataset we’ve ever had kept me on the edge of my seat as I waited impatiently for it to be finished.
- Please could you briefly explain what winning the award means to you?
I was surrounded by amazing people, each with a depth of knowledge that felt endless. So, when I first started this project, I doubted whether I could measure up to its scope. It was a typical case of imposter syndrome, that I admittedly still haven’t shaken off. But receiving an award like this does undeniably help. It’s a tangible reminder to myself that I am capable of this – and beyond! But more importantly, I would like to thank all my supervisors and colleagues that supported me along the way when I doubted myself.
You can find the winning article, and all the shortlisted papers for the 2024 Harper Prize in our virtual issue.
