Estimation of Quercus garryana dougl. ex. hook. (Fagaceae) size- class densities and their relationship to historical grazing pressure and several environmental variables of the Mad River Ranger District of Six Rivers National Forest, Humboldt County, California

Grazing allotment records for the Mad River Ranger District (MRRD), Six Rivers National Forest (SRNF), Humboldt County, CA were used to derive two grazing allotment subsets indicative of historical grazing pressure. The LOW grazed-class was represented by a group of allotments with 42.55 acres per mean annual head of cattle. The HI grazed-class group of allotments had 20.43 acres per mean annual head of cattle. Nine Quercus garryana (Oregon white oak) stands were randomly selected (without replacement) from each allotment subset with probabilities proportional to the size of the remaining oak stands within each group (PPS) using the list sequential method of Sunter (1977). A second stage of sampling was employed where three simple random samples (SRS) of 100 m2 plots were selected from each of the 18 first stage units (oak stands). Densities for three Q. garryana size-classes (seedling, sapling, and mature) were found significantly different between the two grazed-classes (P0.05). Greater seedling and mature tree densities were found for the HI grazed-class, while greater sapling densities were found for the LOW grazed-class. Sapling densities were roughly double mature tree densities regardless of grazed-class, providing evidence that regeneration is occurring for this species across graze pressure levels. Multiple regression models were fit for each of the three size-classes versus the independent variables: 1) available water-holding capacity of the soil; 2) surface horizon soil pH; 3) graze use (oak acres/mean annual head); 4) percent canopy cover; 5) vegetative understory type; and, 6) slope. The seedling density model was found significant (P=0.0069, R2=0.8432); sapling density and mature tree density models were not (P=0.0847, R2=0.683; P=0.2824, R2=0.534, respectively). The power of these models to detect significant differences ranged from excellent (seedling model = 0.9999, sapling model = 0.9633) to fair (mature tree model = 0.7457). Significant regressors of the seedling density model were: graze use (P=0.0048); understory (P=0.0173); available water-holding capacity (P=0.0189); and, canopy cover (P=0.0229). Local regression (loess) conditioning plots were examined to assess interactions between independent variables and to aid in interpretation of the impacts of significant regressors on seedling densities. Seedling density was reduced with increased graze use and AWC while increased canopy cover varied positively with seedling density. Increased seedling density was also found in plots dominated by annual grasses. Results provide evidence that in annual grass understories of the MRRD, moderate levels of livestock grazing reduce interspecific competition for plant resources between short-lived annual grasses and Q. garryana seedlings, thereby enhancing Q. garryana seedling recruitment and survival.