TESTING A PALEOPRECIPITATION PROXY ACROSS TEXAS FLORAS

Stable carbon isotopes of C3 leaves have been shown to correlate with precipitation, although a universal model continues to be elusive. In this study, we test the model published by Kohn (2011) with modern Texas floras across a gradient in mean annual precipitation ranging from 250 mm to 1500 mm. Leaf litter, bulk leaf matter, and wood were collected from thirteen relatively undisturbed sites and analyzed for δ¹³C, taking care to exclude grasses and C4 dicots. Values averaged across species and sample type ranges from an average of -26.34‰ (VPBD) in the Davis Mountains to an average of -30.06‰ (VPBD) from near Beaumont, Texas. Regression of latitude, altitude, and log₁₀ of mean annual precipitation + 300 mm against δ¹³C averages produces an R² of 0.92 (p=0.0007). Precipitation predicted from the Kohn equation ranged from 272 mm to 2720 mm. However, compared to actual precipitation, the Kohn model produced error generally increasing from west (3% under-prediction) to east (75% over-prediction. Stepwise regression (R² = 0.53, p = 0.005) of individual sample δ¹³C indicates that elevation differences in this sample set are not significant and produces the equation δ¹³C = -33.4773 –8.199log₁₀(MAP +300) + 1.0031 „ Latitude, °„ . Data averaged for each site closely agrees with the Texas model, showing a small improvement over the Kohn (2011) model, and agrees better with actual mean annual precipitation. Based on these results, the Kohn (2011) model (and other similar equations) overestimates MAP in areas where MAP exceeds 500 mm/year. Use of d¹³C_{plant material} as a proxy for MAP is best utilized in arid to sub-humid settings.