Wed, Aug 04, 2021:On Demand
Background/Question/Methods
The temperatures experienced by organisms differ substantially from meteorological readings, and yet are not frequently measured. In order to predict species response to novel or changing environments, we are then forced to predict microclimate conditions using mathematical or statistical models. However, these models may or may not predict aspects of temperature critical for determining survival and population growth rates. In particular, absolute maximum, minimum, and autocorrelation of temperatures can be critical determinants of life history parameters. In most cases, however, microclimate model performance is evaluated using average temperatures. We examine the variance, maximum, minimum and autocorrelation of underbark and river temperatures from southern Ontario, and compare these to both available meteorological data and models of the microclimates.
Results/Conclusions In general, these microclimates have a stronger autocorrelation signal than air temperature, and as a result models with memory better represent this feature of the temperature time series. On the other hand, microclimate maximum and minimum temperatures are not well described by these models, and instead require a tighter connection to external drivers like solar loading or groundwater influence. While it is possible that highly complex models with many details may be able to represent multiple features of microclimate time series such as mean, variance, autocorrelation, maximums and minimums, it also seems likely that the detailed parameterization will limit use. Instead, we may need to select the microclimate model that is optimized for our ecological question.
Results/Conclusions In general, these microclimates have a stronger autocorrelation signal than air temperature, and as a result models with memory better represent this feature of the temperature time series. On the other hand, microclimate maximum and minimum temperatures are not well described by these models, and instead require a tighter connection to external drivers like solar loading or groundwater influence. While it is possible that highly complex models with many details may be able to represent multiple features of microclimate time series such as mean, variance, autocorrelation, maximums and minimums, it also seems likely that the detailed parameterization will limit use. Instead, we may need to select the microclimate model that is optimized for our ecological question.