Site Factors Influencing Tree Mortality During Drought in Texas
MetadataShow full metadata
Climate models predict an increase in the frequency of severe weather events, including prolonged drought conditions coupled with exceptionally high temperatures. These so called "global-change-type" drought events have been linked to numerous forest dieback events worldwide. Texas experienced such an event in 2011, which reportedly killed 6% of all trees in the state. The purpose of my research was to identify site factors that modified local rates of tree mortality. In 2014, I censused 64 plots across the state, of which, 40 were included in the final analysis focusing on Central Texas. Over 6000 trees were included, mostly in the genera Juniperus (n=3487), Quercus (n=1054), Ilex (n=745), Ulmus (n=347) and Diospyros (n=308). Each tree above 10 cm circumference was identified to the species level, its basal circumference was measured and its health status recorded in one of seven categories between dead with no sign of resprouting to <25% crown die-back. For each plot, I also collected site variables describing community composition, elevation, slope, aspect, solar insolation, water storage capacity, soil texture and depth, as well as climate factors, including annual precipitation from 2008 to 2011 and daily temperatures in 2011. I need binary logistic regression in a multivariate model selection analysis for individual species and in some cases, pooled samples of two closely related species, to determine which factors produced the most efficient model for the prediction of crown dieback and tree mortality.
The number of days in which trees were exposed to temperatures over 35 or 38°C in 2011 had significantly positive effects on tree mortality in three out of nine models. Heat exposure had independent effects on crown dieback and on the odds of resprouting. Precipitation in 2011 had significant effects on eight out of nine models, but they were positive for some species and negative for others. Topographic effects (e.g., slope and aspect) were significant for four out of nine models. Collective stand density was not as good a predictor of mortality as species-specific densities. For example, the mortality odds of Juniperus ashei on the Edwards Plateau were more closely related to intraspecific density, suggesting self-limiting. Although J. ashei was by far the most common species, Quercus fusiformis was not negatively affected by J. ashei density, but instead by its own intraspecific density and the density of Diospyros texana.
I conclude that studies of drought-related tree mortality focusing on purely abiotic factors such as climate, topography and soil are incomplete, because they omit the influence of species interactions in multispecies communities. My study exposed some of the complexities associated with linking climate events to vegetation changes, in particular, the effects of landscape variation, vegetation composition and management history.