Checking bioclimatic variables that combine temperature and precipitation data before their use in species distribution models

• Published in: Austral ecology Vol. 47; no. 7; pp. 1506 - 1514
• Main Author: Booth, Trevor H.
• Format: Journal Article
• Language: English
• Published: Richmond Blackwell Publishing Ltd 01.11.2022
• Subjects: Antarctica; Atlas of Living Australia; BIOCLIM; Bioclimatology; Biodiversity; climate; Climate change; Discontinuity; ecological niche models; Ecological niches; Equatorial regions; Geographic information systems; Geographical distribution; habitat suitability models; Hydrologic data; Inspection; niches; Precipitation; Remote sensing; temperature; Variables; Antarctica
• DOI: 10.1111/aec.13234
• ISSN: 1442-9985

The 19 BIOCLIM variables available from the WorldClim databases have become the most widely used set of climatic variables for creating species distribution models (SDMs) and ecological niche models (ENMs). Nevertheless, in recent years, there has been an increasing trend to exclude four of these variables from SDM/ENM analyses. The four values concerned are interactive as they combine both temperature and precipitation assessments. Their exclusion was justified initially due to discontinuities in their interpolated surfaces observed in a study using data from version 1 of WorldClim. The discontinuities were thought to be artefacts of the measures used, and such sudden changes were unlikely to be biologically meaningful. A freely available biodiversity database and an open access geographic information system were used here to check the four interactive variables for discontinuities in data from both versions 1 and 2 of WorldClim. Over most of the world, the four interactive BIOCLIM variables from WorldClim were found to change smoothly. Major and minor discontinuities, however, were noted for all four variables in specific parts of some of the six continents (excluding Antarctica) covered by WorldClim databases. Discontinuities were related to sudden changes in the quarterly periods used to calculate the variables. These discontinuities were often found in equatorial regions where there are only small changes in temperature across the year. Bimodal precipitation distributions may also cause problems. It is recommended that the quick and simple method described here should be used to facilitate visual inspection and numerical checking for possible discontinuities of these variables before SDM/ENM analyses. If discontinuities are found for an interactive variable in the study area, it is recommended that the variable should not be used. Ways in which the four variables could be recalculated when new databases are created are suggested to reduce discontinuity problems.