Spatiotemporal climate variability in the Andes of northern Peru: Evaluation of gridded datasets to describe cloud forest microclimate and local rainfall

Abstract

Tropical montane cloud forest may be especially sensitive to climate change. However, our ability to understand effects of climate on montane biodiversity remains limited by the resolution of climate data. We compared 5 years of in situ weather data from cloud forests in northern Peru, regional weather stations, and gridded datasets to examine how climatologies reflect (a) forest microclimate buffering and (b) local rainfall in a sparse data region; we also examined spatiotemporal variability and regional trends. Across a 1,700–3,100 m gradient in which temperature did not covary with relative humidity (RH), in situ data showed interactions between climate and land-use. Forest humidity buffered warming-induced evaporative drying across elevations, and inside forest maximum vapour pressure deficit (VPDmax) did not change with elevation, whereas with a 22% reduction in RHmin at stations, VPDmax increased >10-fold from high to low elevations. Cloud forest dried out on sunny days after 3 days without rain, especially during ENSO-related drought concurrent with peak solar insolation. Climatologies were twice as precise for temperature as rainfall. Chelsa captured a 3.9°C reduction in maximum temperatures inside forest (MAE 1.6°C, R2 = 0.95) whereas WorldClim reflected drier lapse rates and higher Tmax outside forest. CHIRPS provided the best fit for monthly rainfall (MAE 23 mm, R2 = 48), capturing regional drought but underestimating rainfall >150 mm·month−1. Consistent with stations, CHIRPS showed strong support for regional increases in wet-season rainfall. Reduced variability and more regular dry seasons were only detected by montane stations, especially south of 6°S, where rainfall seasonality shifted to earlier wet-season peaks and reduced dry-season rainfall as part of a transition from the Northern to Central Andes. Our results show that cloud forests may be partly buffered from warming but are likely to become extremely vulnerable under reduced humidity either through forest loss or drought.