Fine‐scale climate change: modelling spatial variation in biologically meaningful rates of warming

Abstract

The existence of fine-grain climate heterogeneity has prompted suggestions that species may be able to survive future climate change in pockets of suitable microclimate, termed 'microrefugia'. However, evidence for microrefugia is hindered by lack of understanding of how rates of warming vary across a landscape. Here, we present a model that is applied to provide fine-grained, multidecadal estimates of temperature change based on the underlying physical processes that influence microclimate. Weather station and remotely derived environmental data were used to construct physical variables that capture the effects of terrain, sea surface temperatures, altitude and surface albedo on local temperatures, which were then calibrated statistically to derive gridded estimates of temperature. We apply the model to the Lizard Peninsula, United Kingdom, to provide accurate (mean error = 1.21 °C; RMS error = 1.63 °C) hourly estimates of temperature at a resolution of 100 m for the period 1977-2014. We show that rates of warming vary across a landscape primarily due to long-term trends in weather conditions. Total warming varied from 0.87 to 1.16 °C, with the slowest rates of warming evident on north-east-facing slopes. This variation contributed to substantial spatial heterogeneity in trends in bioclimatic variables: for example, the change in the length of the frost-free season varied from +11 to -54 days and the increase in annual growing degree-days from 51 to 267 °C days. Spatial variation in warming was caused primarily by a decrease in daytime cloud cover with a resulting increase in received solar radiation, and secondarily by a decrease in the strength of westerly winds, which has amplified the effects on temperature of solar radiation on west-facing slopes. We emphasize the importance of multidecadal trends in weather conditions in determining spatial variation in rates of warming, suggesting that locations experiencing least warming may not remain consistent under future climate change.

Monthly microclimate models in a managed boreal forest landscape

“…By modelling average extreme temperatures separately over 12 months, we could also study the temporal dynamic of the effects of landscape physiography and vegetation on near-ground temperatures. This temporal variation in effects and importance of microclimatic drivers seems to be common, and has been found when modelling seasonal, daily or hourly temperatures separately (Ashcroft and Gollan, 2013a;George et al, 2015;Maclean et al, 2016).…”

“…However, we expected the influence of topographic shading to be moderate in a lowland landscape, which was probably even further declined due to overlapping effects of vegetation. In open landscape or in complex terrain, net solar radiation can be the dominant driver of spatial variation of high temperatures (Dingman et al, 2013;George et al, 2015;Maclean et al, 2016), but in lowland forests insolation of the ground is a function of both terrain and canopy shading. In northern Sweden, northern and southern slopes were found to have quite similar air temperatures but rather contrasting ground temperatures, perhaps due to wind effects and heat accumulation in the soil (Dahlberg et al, 2014).…”

Large- and small-scale environmental factors drive distributions of cool-adapted plants in karstic microrefugia

“…Rorison et al, 1986;Daly et al, 2010;Maclean et al, 2016). This could allow species to shift their distributions towards the cooler and more stable microclimates of north-facing slopes and lower parts of dolines (B atori et al, 2011).…”

Opposing responses to drought shape spatial population dynamics of declining grassland birds

“…Despite the attribution of climate change effects across broad taxonomic groups, fundamental challenges persist on how to improve our understanding and predictions of species responses to climate change (Ehrlén & Morris, 2015;Urban et al, 2016). Recent work has emphasized the need to move away from static modelling of species distributions in favour of integrating key dynamic ecological processes reflective of population dynamics (e.g., dispersal, migration and demographic indices) and to consider the multiscalar influence of environmental factors (e.g., short-and long-term climate and lag effects) that affect where species occur and persist on landscapes (Franklin, 2010;Guisan & Thuiller, 2005;Maclean, Suggitt, Wilson, Duffy, & Bennie, 2017). …”