Invasion, control, and distribution of medusahead [Taeniatherum caput-medusae (L.) Nevski] in California grasslands

Using Medusahead (Mh) [Taeniatherum caput-medusae (L.) Nevski] as an important example, we conducted studies to test (1) the effect of small-scale (10-1 m2) environmental change (accumulation or loss of thatch) on potential for Mh invasion and control, (2) the efficacy of Mh control through field-scale (103-10 4 m2) changes in grazing management, and (3) the use of remote sensing technologies to determine the intensity of Mh presence across California's Central Valley and surrounding foothills (1011-10 12 m2). These studies not only evaluated average effects environmental and management changes on Mh and the total area infested by Mh across the state, but also provided insight into the dynamics of individual small-scale "hotspots" of Mh seed productivity, spatial and temporal "escape" of Mh from grazing treatment within the field scale, and regional "epicenters" of Mh infestation.;In our first set of experiments, we tested the effects of thatch presence on Mh, non-Mh grass and forb seed productivity and mean Mh seed mass in two contrasting Mh-invaded sites (open and shaded) and two contrasting non-invaded sites (drained and riparian). Thatch presence greatly increased Mh seed productivity (from 2 to 30-fold increases) at most sites, but had relatively small effects on seed productivity of forbs and non-Mh grasses. Thatch addition in the non-invaded sites also facilitated development of "hotspots" of Mh seed productivity - individual plots with up to 10,000 Mh seeds m-2. In Mh-invaded sites, the effect of thatch removal on Mh seed productivity from depended on site identity. Thatch removal resulted in a 50% increase and decrease of Mh seed productivity at the shaded site and open site, respectively. This difference between sites may have been driven by differences in soil moisture availability during the late growing season, suggesting that thatch removal will is most conducive to Mh control at relatively dry sites and/or in years with relatively dry spring seasons.;In our second experiment, we applied four targeted forage utilization rates (50, 60, 70, and 80%) combined with two targeted grazing durations (7 and 14 days) and reseeding on 0.2-ha plots using 100-kg ewes. Compared to exclosures and continuously grazed control areas, precision grazing reduced Mh seed productivity by 75-80% as well as Mh and Mh thatch coverage by 72-88% in the year of treatment and/or during the spring and summer of the subsequent year. Across all precision-grazed plots, 11% of post-treatment forage canopy height observations remained above 9 cm -- a defoliation height that greatly suppressed Mh reproduction in a simultaneous clipping experiment. Within precision-grazed plots, differences in apparent forage utilization and grazing duration generated little or no variation in terms of measures of Mh infestation (such as Mh coverage and Mh and non-Mh seed productivity during year of treatment and the subsequent year). Likewise, reseeding had no impact on measures of Mh infestation.;Finally, we used Landsat ETM+ imagery to predict Mh absolute coverage in 1.3 Mha of rangeland across the northern California Central Valley. Quadratic discriminant functions of visible, near-infrared (NIR), and thermal reflectances in May and June were fit to land use class (such as rangeland, wheat/barley, vineyard) and rangeland Mh coverage classes (<5, 5-40, or >40% absolute Mh coverage) using over 2000 observations at 360-m2 resolution within the Dunnigan Hills region of the Central Valley. Another 264 observations were collected on rangelands outside of the Dunnigan Hills. Primarily due to expected patterns of red band reflectance in June, rangeland was easily separated from other land use classes when the relevant model was applied to independent observations. Of models fit to Mh coverage classes on rangeland in the Dunnigan Hills, only one based on green, red, and NIR reflectances in May and June provided unbiased predictions when extrapolated to other regions. High (>40%) Mh coverage could be assigned to little (1.7% or 0.6%) of the 1.3 Mha of rangeland area when the overall prediction error rate was held to 15% or 5% (respectively). However, a greater proportion of rangeland was infested with high Mh coverage in the northernmost Landsat scene, and 10-20 "epicenters" of Mh invasion on scales of 0.005-0.05 Mha were distributed along the length of the modeled Central Valley. These patterns suggest researchers should evaluate the potential for Mh to intensify in the southern portion of the valley; effective control of Mh may require local efforts to be placed in a regional context corresponding to the Mh invasion epicenter. (Abstract shortened by UMI.)...