CR 78:261-270 (2019)  -  DOI: https://doi.org/10.3354/cr01574

ABSTRACT: Ecological studies based on gridded climate data are increasingly common, but the effect of the uncertainty of climate data on ecological estimations is rarely assessed. Here, we assessed the effect of the uncertainty of local and global climate data (LCD and GCD, respectively) on ecological land classifications, using the Holdridge life zones and the country of Argentina as a case study. We quantified the uncertainty in LCD using information from meteorological stations and a simulation approach, which allowed us to create an altered LCD (aLCD), and to quantify the agreement between LCD and aLCD through confidence limits. Then, we quantified the agreement between the life zone maps derived from GCD and LCD—while accounting for LCD uncertainties—and assessed the effects of spatial resolution, environmental resolution, and topographic heterogeneity. We found that the mean agreement between LCD and aLCD for the whole country was about 75%, but there were important variations in the amplitude of the confidence intervals depending on the region. In addition, the mean agreement between ecological maps derived from LCD and GCD ranged between 40 and 83%. However, the LCD-GCD agreement fell inside the confidence limits of the LCD, and decreasing the spatial resolution from 1 to 50 km did not change the results. Decreased thematic resolution improved the agreement in the tropical and antiboreal regions. Overall, our study shows that uncertainties in ecological applications of climate data are higher in environmentally heterogeneous lands, and highlights the need for incorporating climate data uncertainties into ecological studies.

KEY WORDS: Agreement · Climate data · Land classification · Holdridge life zones · WorldClim