Applications to compare gaps in biodiversity data awarded first prize in GBIF challenge

Alejandro Ruete’s entry, Exploring ignorance in space and time, earned the top prize of €20,000 upon its selection by an expert jury who reviewed 16 submissions to this year’s Challenge. As part of his submission, Ruete developed two web applications: The first application, SLWapp, compares the ignorance calculated for seven species groups in Sweden, using Swedish LifeWatch data and the Swedish Species Observations web service provided by Swedish LifeWatch. The other application, GBIFapp, explores ignorance in space and time for amphibians in Europe, using GBIF data.

- What made the winning entry so attractive was its simplicity and scalability. It makes few assumptions, and unlike some other entries it only requires simple occurrence data,” said Roderic Page, a University of Glasgow professor and chair of both the Challenge jury and the GBIF Science Committee. This means the tool can be applied to any geographic region or taxonomic group in the GBIF network.

About the applications

The approach presented for the challenge is meant for managers and users of primary biodiversity data to identify gaps, to assess the spatial and temporal bias inherent to the data, as well as to evaluate the relative gain in knowledge added from new observations. It does so by quantifying and comparing the lack of data (Ignorance) in biodiversity databases, like GBIF.

The aim is to provide ignorance scores and maps that are easily comparable and scalable across dimensions, to report the spatial and temporal distribution of sampling effort (or lack of it). Ignorance maps will serve to properly inform users of the bias inherent to the data and to provide them with tools to properly analyse the data. Simplicity is crucial for web-based implementations on e-infrastructures for biodiversity information. Therefore, this approach expresses ignorance solely based on raw presence-only data (absences are optional but not required) summarized per grid cells or pixels. For the same reason, this approach relies on as few assumptions as possible. The aim is not to include any covariates or correlations and to avoid prediction, estimation and interpolation methods. The core algorithm is then thought to be fast enough to be implemented in web-based tools and Application Programming Interfaces (APIs). However, it can also be used off-line in the researcher's prefered analysis environment. This project is in line with the need identified by Rocchini et al. (2011) and will provide quality control tools for protocols for biodiversity analysis such as the one proposed by Hortal et al. (2007).

The algorithm behind the Ignorance Score is designed for comparison of bias and gaps in primary biodiversity data across taxonomy, time and space:

1.Taxonomy: applies to any species groups, but has also applications at the species level.
2.Time: can be used to aggregate or dissect bias over time.
3.Space: compares per pixel or between summaries of irregular polygons, and can be suited to different resolutions.