Functionally relevant climate variables for arid lands: a climatic water deficit approach for modelling desert shrub distributions

Abstract

AbstractAimWe have three goals. (1) To develop a suite of functionally relevant climate variables for modelling vegetation distribution on arid and semi‐arid landscapes of the Great Basin, USA. (2) To compare the predictive power of vegetation distribution models based on mechanistically proximate factors (water deficit variables) and factors that are more mechanistically removed from a plant's use of water (precipitation). (3) To quantify the climate gradients that control shrub distributions in a cold desert environment.LocationThe central basin and range ecoregion of the western USA (36–42°N).MethodsWe used a modified Thornthwaite method to derive monthly water balance variables and to depict them using a water balance climograph. Eighteen variables were calculated from the climograph, representing different components of the seasonal water balance. These were used in boosted regression tree models to derive distribution models for 18 desert shrub species. The water balance approach was compared with an approach that used bioclimatic variables derived from the PRISM (Parameter‐elevation Relationships on Independent Slopes Model) climate data.ResultsThe water balance and bioclimatic approaches yielded models with similar performance in predicting the geographical distribution of most shrub species. Cumulative annual climatic water deficit was consistently the most important water balance variable for predicting shrub type distributions, although predictions were improved by the inclusion of variables that describe the seasonal distribution of water balance such as water supply in the spring, fall actual evapotranspiration, monsoonality and summer decline in actual evapotranspiration.Main conclusionsThe water balance and bioclimatic approaches to species distribution modelling both yielded similar prediction accuracies. However, the water balance approach offers advantages over the bioclimatic approach because it is mechanistically derived to approximate physical processes important for plant growth.