Predicting toxicity of tall larkspur (Delphinium barbeyi): measurement of the variation in alkaloid concentration among plants and among years.

Tall larkspur (Delphinium barbeyi) is the principal mountain larkspur responsible for the majority of cattle deaths on mountain rangelands in western Colorado and central and southern Utah in the United States. Ten plants in each of two tall larkspur populations in the mountains near Ferron and Salina, Utah, were marked, and single stalks were harvested periodically through the growing season for 4 yr. Toxic alkaloid concentration [alkaloids containing the N-(methylsuccimimido)-anthranilik ester group] was determined by Fourier transform infrared (FTIR) spectroscopy. Individual larkspur plants varied in alkaloid concentrations, especially in early growth (14-38 mg/g). As the concentration declined over the growing season, variation among plants also declined. There were yearly differences in alkaloid concentration among individual plants (P < 0.01) and populations (P < 0.001), even after accounting for differences in phenological growth between years. Variables such as precipitation, temperature, days since snow melt, growing degree days (sum of mean temperature each day from snow melt), and plant height and weight were all considered in a Mallows Cp multiple regression selection procedure to predict alkaloid concentration. The mixed model procedure in SAS adjusted the regression equation for locations and years. Growing degree days was the best single predictor of alkaloid levels: In y = (3.581 - 0.00423 GDD), R2 = 0.85. Internal validation of this equation within individual years and locations from which the equation was developed, produced correlations between observed versus predicted values ranging from r = 0.73 to 0.93. External validations on nine other larkspur populations produced correlations ranging from r = 0.76 to 0.99. This predictive equation can provide a tool for ranchers and land managers to make management decisions of when to graze cattle in larkspur areas.