Biodiversity data obsolescence and land uses changes

Abstract

BackgroundPrimary biodiversity records (PBR) are essential in many areas of scientific research as they document the biodiversity through time and space. However, concerns about PBR quality and fitness-for-use have grown, especially as derived from taxonomical, geographical and sampling effort biases. Nonetheless, the temporal bias stemming from data ageing has received less attention. We examine the effect of changes in land use in the information currentness, and therefore data obsolescence, in biodiversity databases.MethodsWe created maps of land use changes for three periods (1956–1985, 1985–2000 and 2000–2012) at 5-kilometres resolution. For each cell we calculated the percentage of land use change within each period. We then overlaid distribution data about small mammals, and classified each data as ‘non-obsolete or ‘obsolete,’ depending on both the amount of land use changes in the cell, and whether changes occurred at or after the data sampling’s date.ResultsA total of 14,528 records out of the initial 59,677 turned out to be non-obsolete after taking into account the changes in the land uses in Navarra. These obsolete data existed in 115 of the 156 cells analysed. Furthermore, more than one half of the remaining cells holding non-obsolete records had not been visited at least for the last fifteen years.ConclusionLand use changes challenge the actual information obtainable from biodiversity datasets and therefore its potential uses. With the passage of time, one can expect a steady increase in the availability and use of biological records—but not without them becoming older and likely to be obsolete by land uses changes. Therefore, it becomes necessary to assess records’ obsolescence, as it may jeopardize the knowledge and perception of biodiversity patterns.

Taxonomic bias in biodiversity data and societal preferences

“…We thus anticipate an increasing bias between taxa mostly known from observation-based occurrences and taxa mostly known from specimen-based occurrences. In addition, a lot of records are old and incomplete, and could soon, or already, be obsolete 46 , which risks reinforcing the taxonomic bias against classes with relatively few recent occurrences.…”

Global mapping of potential natural vegetation: an assessment of Machine Learning algorithms for estimating land potential

“…The primary problems of using GBIF for PNV mapping will remain however, as these are primarily due to high clustering of points and under-representation of often inaccessible areas with very high biodiversity (Yesson et al, 2007;Meyer et al, 2016). GBIF records have been shown in the past to give biased results (Escribano et al, 2016), so that spatial prediction methods that account for high spatial clustering, i.e. bias in training point representation in both space and time;…”

Temporal degradation of data limits biodiversity research

“…() and Escribano et al. (), can be used to prioritize species for new sampling campaigns and identify areas where knowledge is primarily from old records.…”

“…Surveys can be planned accounting for those sites that hold the oldest records or by identifying areas that suffered significant land use and/or environmental changes after the last collection. Stropp et al (2016), Escribano et al (2016) taxonomic checking in TAFs are still under development (Nguyen, Soto, Kontonatsios, Batista-Navarro, & Ananiadou, 2017;Vanden-Berghe et al, 2015), manual taxonomic checking can be carried out for small databases or for subsets of data (Zermoglio et al, 2016).…”

“…Thus, without new surveys, a biodiversity database will inevitably become less useful. Approaches such those applied by Stropp et al (2016) and Escribano et al (2016), can be used to prioritize species for new sampling campaigns and identify areas where knowledge is primarily from old records.…”