Predicting the future of our tree species

By Albert Ciceu, Austrian Research Centre for Forests (BFW)

Did you know that due to the rapid development of climate change some species are unable to adapt fast enough to the new climatic conditions and might reduce their distribution or even face possible extinction?

In such cases, species distribution models (SDM) come handy since they can be used to identify areas with a potential biodiversity loss and assist species migration to more suitable areas. SDM uses advanced computer algorithms, georeferenced biodiversity observations and geographic layers of environmental information to build predictive models that can be used to make inferences about the potential distribution range of species in space and time.

In the SUPERB project, we use species distribution modelling to develop a European tree species and seed provenance recommendation system called Seed4Forests. Currently, the SUPERB partners in our demonstration areas count on our species and provenance recommendation list when considering restauration activities such as planting, and we hope that in the future this will be a common practice also for other forest restoration projects.

To obtain such a model, well-developed hard skills are required for data wrangling, data manipulation, algorithm selection and fitting, as well as data visualisation. Especially because these tasks are usually executed via a programming language. In late June this year, our team at Federal Research and Training Centre for Forests, Natural Hazards and Landscape (BFW) in Austria decided that it would be beneficial for some of us to improve and polish some of these skills by attending a species distribution modelling course.

In early July last year, one of my colleagues and I booked early bird tickets to the 8th edition of the Species Distributions Modelling course held in Evora, Portugal, by two renowned and well-established scientists in this field: Miguel Araújo and Babak Naimi. The course was scheduled to start on the 7th of November and last 8 days, during which we would go through all steps involved in building and testing models of species distribution.

The summer passed as fast as you can say ‘Schnitzel’ and here we were catching a flight to Lisbon at 3:30 AM Sunday morning. We had the whole Sunday to wander around amazing and hilly Lisbon were we – despite being sleepy – enjoyed a long walk.

As two tree geeks would normally behave we couldn’t resist the temptation to visit the botanical garden and test our tree identification skills. We both performed well with small exceptions when it came to some Mediterranean tree species and non-native tree species of Europe. Early in the morning we took the bus to Evora, a beautiful, lively university city. We would later learn that Evora’s streets are not ordinary. You can easily hear the sound of Fado music drifting while encountering students dressed as if they are coming from a Quidditch game between the Gryffindors and the Slytherins (the main inspiration of Harry Potter’s uniforms came from Portugal).

But back to our arrival: when we arrived in Evora, the first class was ready to start. We joined the small group of around 12 young scientists who were physically present and 2 more attendees online, eager to find out how and when species distribution models can be useful.

 After a quick round of introductions, I was surprised to find out that although we were drawn by the same interest for this course, our backgrounds were very different. Based on their presentation, I could easily divide the group in 3 main clusters: those interested in birds, plants lovers, and marine life enthusiasts.

The course started with a short introduction of the terminology used in species distribution modelling. Prof. Miguel Araujo was mostly responsible for the theoretical aspects of the topic, describing in detail the timeline from the ‘early days’ of SDMs to the cutting-edge research that is now conducted in the world as well as in his labs in Evora and Madrid. He filled our memory citing and describing methodologies applied in different research papers that tackle aspects of species distribution modelling in time and space. Throughout the course he gave us a good understanding of the theoretical underpinnings of ecological niche models while emphasizing the strengths and limitations of those models in the context of different uses.

Prof. Babak Naimi, on the other hand, was responsible for developing our hard skills. He introduced us to the sdm package, a package developed by himself and Miguel with the main purpose of building species distribution models. Throughout the week he described the main functionalities of the package, while during his intense coding sessions we discussed extensively the use of the package in tacking our research questions. It was indeed the perfect mixture of theory and practice.

The 8th and last day of the course was reserved for presenting our work through short presentations. We had been split in small groups of 3 to 5 people and given the task to design and build a species distribution model. Using some data already available at European scale, my group and I developed 3 SDMs for 3 oak species (Quercus frainettoQuercus cerris and Quercus pubescens) which is of big interest for the SUPERB demo areas. We predicted the current and the future distribution of these species under different climate scenarios. In order to improve the ”projection power”  of our model we also predicted the SDM of a new invasive insect species (Corythuca arcuata) as this species has a huge potential to threaten the stability and limit the distribution of the oak species in Europe.

Soon it was time to fly back to the country of Mozart and the city of Johann Strauss. The course was a treat and we are ready and excited to apply our newly acquired skills in the SUPERB project helping you choose the most suitable species for the forests of our future.