Read more about SBDIs satellite symposium at Oikos
Swedish LifeWatch and Biodiversity Atlas Sweden will soon merge into the joint Swedish Biodiversity Data Infrastructure, a single e-infrastructure in close collaboration with the Living Atlases community. At Oikos satellite symposium, 4 February, data-focused demo sessions of the new infrastructure were held to show new opportunities for data driven research. Here you can read more about what happened at the symposium.
What data are there? How can I use them?
Ecology is evolving fast. Our discipline goes through a rapid transformation process to support global sustainability targets and understand human pressures on our ecosystems. In this process, a number of characteristic trends can be observed. For example, new methods for data collection and analysis, such as citizen science and autonomous observation techniques are quickly developed and adopted. Structured monitoring data, and molecular and genomic information add to the data contents exponentially. At the same time, ecologists increasingly engage in well-organized community networks. Those trends lead to a fast accumulation of available data and enable more inter-disciplinary and scalable ecological research.
A major facilitator in this process is the emerging landscape of biodiversity information systems, which offer services for easier access and publication of relevant data. These infrastructures have the potential to offer fundamentally new ways to conduct ecological science in the future. Two of today’s Swedish biodiversity infrastructures, Biodiversity Atlas Sweden (BAS) and Swedish LifeWatch (SLW), will merge into the joint Swedish Biodiversity Data Infrastructure (SBDI) providing new opportunities.
The symposium will include both presentations as well as data-focused demo sessions. During the afternoon you can experience and together with experts test Swedish biodiversity data infrastructures, and discuss with analysis experts about your research question. You can also get advice on suitable data and analysis approaches, and together with experts explore and dig deeper into specific data sets.
Organisers: Biodiversity Atlas Sweden, Swedish LifeWatch, and Dept. of Forest Resource Management SLU
Morning session - Presentations
- Welcome/Practical information
- Use and re-use of open data in ecological analysis - Swedish research infrastructure for biodiversity data - Anders Telenius (GBIF, NRM), Fredrik Ronquist (NRM) & Holger Dettki (ArtDatabanken, SLU)
- Open systematic forest and landscape data - Anna-Lena Axelsson (SLU)
- Socio-ecological modelling in the Koster National Park - Dawn Field (GU)
- Improving habitat suitability models using opportunistically collected presence, only Citizen Science data - Ute Bradter (ArtDatabanken, SLU)
- Do Citizen Science based models accurately predict species distribution? - Laura Henckel (ArtDatabanken, SLU)
- Advances in the use of opportunistic data for scientific purposes - Alejandro Ruete (Greensway AB & Dept. of Ecology, SLU)
- Using machine learning for identification and conservation of threatened small habitats - Matti Ermold (county board Jönköping)
Afternoon session - Exhibitions/Demonstrations
Short presentations: 13:00-14:00
- Integration of molecular data into Biodiversity Atlas Sweden - Maria Prager (SciLifeLab, BAS)
How can we help you publish, access and use metabarcoding data?
Please complete and/or spread the word about our online survey
- Arenas for co-operation through citizen science ARCS, a portal for citizen science research in Sweden - Mari Jönsson (ArtDatabanken, SLU)
- Short Introductions to Ehibitions/Demonstrations:
- What can you get from the Analysis Portal for biodiversity data? – Debora Arlt (ArtDatabanken, SLU)
- The BioAtlas Data Portal: species occurrences, explore a region – Manash Shah (GBIF, NRM)
- The BioAtlas Data Portal: species occurrences, explore a region, spatial portal – Christian Svinseth
- The BioAtlas Mirroreum: A web-based environment for Reproducible Open Research – Christian Svinseth
- Swedish National Forest Inventory (NFI): Open resources and tailored data “on-demand” – Jonas Dahlgren (SLU)
- The Swedish Forest Soil Inventory - data on site conditions and soil chemistry – Johan Stendahl (SLU)
- Landscape data from NILS: the National Inventory of Landscape in Sweden – Henrik Hedenås (SLU)
- What can you get from the Analysis Portal for biodiversity data? – Debora Arlt (ArtDatabanken, SLU)
- The Swedish National Forest Inventory NFI: Aggregated data, maps and microdata from the Swedish NFI can be used for large-scale studies or modelling. The data is also useful for global studies as similar datasets exist in other countries. We will provide an overview of open data sources and demonstrate a number of web-based analysis tools that you can use to explore our aggregated data. We will inform about the services provided by our research support function and give information on how to proceed if you want to order microdata, tailored datasets or specific analysis. We are also here to discuss your research idea that involves NFI data. - Anna-Lena Axelsson & Jonas Dahlgren (
- National Inventory of Landscape in Sweden (NILS): Systematic data documenting landscape changes and status in Sweden. - Henrik Hedenås (SLU)
- The Swedish Forest Soil Inventory: Statistics and data on site conditions and soil chemistry documenting the status and change in Sweden's forest soils. - Johan Stendahl (SLU)
- Biodiversity data and integrated analysis tools from the Analysis Portal, Swedish LifeWatch: The Analysis Portal is a service from Swedish LifeWatch – a research infrastructure for biodiversity data. Here you can search and filter data, you will find visualization and analysis tools, and you can download data from all data sources connected to SLW. - Debora Arlt (SLU)
- The BioAtlas Data Portal: As part of the Living Atlases Community the Biodiversity Atlas Sweden is a collaborative research infrastructure that aggregates biodiversity data from multiple sources. The BioAtlas portal offers opportunities for innovative, interdisciplinary research on biodiversity and ecosystems. It provides visualization and analysis tools, using maps along with environmental and contextual information to produce distributions, predictive models and charts of the relationships between biodiversity and the environment. You will also be able to import your own records and export maps and data sets.
- Explore occurrence data, explore a region: - Manash Shah (GBIF, NRM)
- Spatial Portal: - Christian Svindseth (GBIF)
- Mirroreum, Dockerisering: Mirroreum - A web-based environment for Reproducible Open Research: Can be used entirely through the web browser, or can run locally, or deployed at a server in the cloud. To support work using R, it provides a platform for working with spatial data using R and the web-variant of the RStudio IDE as well as a Shiny server. This means it comes with a pre-installed set of various assorted packages. Among the many pre-installed packages is ALA4R and livingatlases - R packages that provide access to data sources in the Living Atlases community. Various other packages are pre-installed to support for example ecological niche modelling work. - Christian Svindseth (GBIF)
- Species Observation Explorer: What do species observation data from Artportalen tell us about species occurrence in an area? When there are no observations – is this because the species really isn’t there or because no one has been visiting or reporting from the area? With the Species Observation Explorer you can check it out. - Alejandro Ruete (Greensway AB)
Advances in the use of opportunistic data for scientific purposes. Alejandro Ruete (Greensway AB):
Non-systematically collected, a.k.a. opportunistic, species observations are accumulating at a high rate in biodiversity databases. Occupancy models have arisen as the main tool to reduce effects of limited knowledge about effort in analyses of opportunistic data. Our advances lay mainly in the way big volumes of data are interpreted. Two common but contrasting assumptions with big data are that i) big volumes of data ensure the presence of the expected pattern, and ii) that bad quality data points need to be “found and destroyed”. We show how we need to distill the information contained in big data to answer a particular scientific question, and at the same time, there is no need to remove “bad” data. In fact, every data point has some information to contribute. We will show this with several examples of how our “Bayesian data distillery” works and what you can do with it. We will show how to derive measures of site use within seasons from estimates of daily occupancy, and how these better capture the presence of species making a specific use of a site (e.g. breeding) and help filtering other species (e.g., migrants passing by, single itinerary prospecting individuals). We will show how to rethink the definition of species presence at a site, and how this affect biodiversity measures. Finally, we will tease with what happens at larger scales.
Using machine learning for identification and conservation of threatened small habitats. Matti Ermold, county board Jönköping:
The monitoring and conservation of threatened species and habitats is one of the responsibilities of the Swedish county administrative boards. This encompasses for example the economic compensation of land owners on whose properties those species or habitats can be found. Rich fens are a unique wetland type and one of the few habitats that are part of the Swedish environmental protection agency’s action program for threatened species and habitats (ÅGP). Rich fens are under threat mainly due to drainage, nutrient pollution and lack of management. Several previous attempts to identify those often-small habitats had a low success rate. To improve identification of rich fens I used random forest classification based on several spatial (soil pH, Calcium, soil type, density of headwaters, land use type) and biological variables (density of common rich fen plants). I calibrated the model on field observation of 525 potential rich fens in the county of Örebro of which 220 where later classified as being actual rich fens. Results of the trained model on the test data showed that 80% of the rich fens where labeled correctly.
The aim of this study was to test and later deploy the model and let it predict rich fens from unlabeled wetland data in Jönköping county. This will be followed by field visits which will help to further improve the reliability of the algorithm by updating the labels. On successful identification of previously unknown rich fens the county board aims to contact land owners, inform them about the value of the habitat on their land and hopefully inspire some to be part of the economic compensation program to manage and preserve rich fens.