Inferring ecology from opportunistic citizen science data
Av Debora Arlt
Changes in the environment, such as climate and land use change, affect organisms, their interactions and the functioning of our ecosystems. In order to mitigate and meet such effects we want to understand and predict how organisms react to the changes. For this we normally use data on the occurrence of organisms and their relation to the environment from standardized surveys. Because of the costs involved those designed surveys often cover only a limited area and few points in time. Today many people report opportunistic observations of species through Internet based data portals like the Swedish Species Gateway, with reports covering many localities over large areas and during many days over many years. Can we use these opportunistic reports? Can they contribute to better understand ecological processes on more detailed and larger scales than is possible using data from standardized surveys?
The challenges lay in the unstructured nature of this data, as reports are made opportunistically at the will of the observer. Observers decide when and where to observe and what to report. When designing standardized surveys we aim to control the observation process to be able to extract information about the underlying ecological process from the recorded data. With opportunistic data the observation process is largely unknown, potentially creating unwanted patterns in the data that, if ignored, can lead to incorrect conclusions. Research is progressing to device methods to handle some of this uncertainty. But we need to check when those methods work..
There is certainly much information contained in opportunistic reports and in my research I aim to explore when and how we can extract this information. I use validation studies comparing patterns resulting from opportunistic and reference data from standardized surveys. Some of the results are encouraging showing correspondence in some overall patterns. Nevertheless, much remains to be done to determine why results for specific situations (species, sites, times) can be largely different. We need new studies tailored at investigating how we can identify which records contain the most valuable information about ecological processes. We also need means (metrics, illustrations) that can be easily adopted