Abstract

VACA, Raúl A.  and  GOLICHER, Duncan J.. Integración del balance hídrico en la modelación de la distribución de especies de árboles mexicanos. Bot. sci [online]. 2016, vol.94, n.1, pp.25-42. ISSN 2007-4476.  https://doi.org/10.17129/botsci.241.

Species distribution models are typically based on relationships between species occurrence and simple climatic variables. An alternative strategy is to use mechanistic or semi-mechanistic models that incorporate variables that are known to have important effects on physiological responses of the organisms. We compared models built using simple climatic variables with models incorporating calculations of monthly soil water balance when predicting the distribution of 258 Mexican tree species associated with five different ecosystems. The species chosen varied in their degree of exposure to periods of drought. Water balance was calculated using a bucket model that integrated the effects of evapotranspiration and rainfall over a year. Models incorporating calculated soil water balance were contrasted with models that used net annual rainfall as a measure of variability in water availability. The effects of mean annual temperature and temperature range were included in both sets of models. More detailed water balance models showed a significantly improved ability to predict the distribution of some, but not all, species. The observed increase in mean AUC when replacing Panual by DHanual was higher for the group of species of xeric scrub (0.15), intermediate for the group of species of deciduous tropical forests (0.09), and significant, but comparatively low for the group of species of tropical moist forests (0.07). Calculated water balance did not improve predictions for species found in montane forests. We conclude that including detailed calculations of soil water balance can improve the species distribution models in areas of both moist and dry warm tropical forests, where temperature remains comparatively constant. However including variability in soil water balance is less important when species distributions are limited by the effect of elevation on temperature. Model predictions could be further refined through the incorporation of soil parameters and geomorphological variability at a local scales.

Keywords : bucket model; generalized additive models; water balance; species distribution models.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )