We began our analysis of Tamarix distributions and climate by looking at January mean minimum temperatures and growing season length, to test hypotheses that high altitude and northern limits on Tamarix growth and spread reflect temperature limitations during the growing season or damage by cavitation25 at sufficiently low temperatures. However, no relationships between Tamarix occurrence and either temperature parameter were sufficiently consistent to warrant use in projecting future range shifts (Figs. 7.3A, B). However, we did observe that for both January minimum temperature and growing season length, no Tamarix occurred below or above a certain range.

We then examined the relationship between annual precipitation and Tamarix occurrence to test existing hypotheses that Tamarix is limited in its distribution to relatively dry areas. Despite the complex suite of hydrogeomorphic factors that appear to affect Tamarix's success, mean annual precipitation was a strong predictor of Tamarix occurrence. Tamarix occurrence peaked at the lowest mean annual precipitations found in the United States (47 mm/yr) and declined to zero by 1150 mm/yr (Fig. 7.4). An observed rise in occurrence between 350 and 650 mm/yr of precipitation was not explained by overrepresentation of warmer regions in that precipitation range; the rise persisted when the analysis of distribution as a function of precipitation was constrained to sites with growing season length within the range tolerated by Tamarix. Fitting of a logistic model to the data for Tamarix and annual precipitation yielded constant values of C = 1.703 and B = 0.003. Model fit was extremely good, yielding a raw R2 value of 0.954 and a corrected R2 of 0.917. For comparison, a model without precipitation (setting P = 0) yielded a raw R2 value of 0.444.

The finding that annual rainfall strongly predicts Tamarix occurrence is consistent with characteristics of the native habitats in which Tamarix occurs, observed limitations on Tamarix expansion at the western extreme of its range in California, and the working hypotheses of investigators of Tamarix in the western United States. The most recent regional climate model (RegCM) developed for the western United States (Giorgi et al. 1998) predicts a complex

100 150 200 250

growing season length B

Figure 7.3. (A) The relationship between mean minimum January temperature and Tamarix distribution across all levels of annual precipitation. (B) The relationship between Tamarix distribution and growing season length in three different annual precipitation ranges. For both January temperature minima and growing season length, the relationship with Tamarix distribution was inconsistent across precipitation zones.

Figure 7.4. The relationship between annual precipitation totals and the probability of Tamarix occurrence in riparian areas in the United States. A logistic curve fitted through the data yields an adjusted R2 value of 0.917. Model and parameter values are cited in the text.

spatial array of precipitation responses for the area under doubled atmospheric CO2 concentrations, which, across the region infested by Tamarix, yields an approximate net precipitation change of -67 mm/yr (Zavaleta 1998). In the arid Southwest, a decline of that magnitude represents a 15 to 40% decrease in annual precipitation. Because of the difficulty of accounting for seasonal and spatial variation in RegCM model predictions, specific geographic predictions of Tamarix spread are not feasible. However, the strong negative relationship between rainfall and Tamarix distribution found in this study suggests that a decline in precipitation or system drying due to warming-enhanced evapotranspirational losses would allow Tamarix to spread to areas currently too moist for it to invade.These areas would likely include greater portions of California and Oregon and larger areas of eastern Texas, Oklahoma, Kansas, Nebraska, South Dakota, and even Arkansas, where Tamarix has been observed near the Oklahoma border on the Arkansas River (Holla Bend National Wildlife Refuge, pers. comm. 1998).

0 0

Post a comment