Point Sampling to Stratify Biomass Variability in Sagebrush Steppe Vegetation

Cover and yield are two of the most commonly monitored plant attributes in rangeland vegetation surveys. These variables are usually highly correlated and many previous authors have suggested point-intercept estimates of plant cover could be used as a surrogate for more expensive and destructive methods of estimating plant biomass. When measurement variables are highly correlated, double sampling can be used to prestratify variability in the measurement that is more difficult or costly to obtain, thus improving sampling efficiency. The objective of this study was to examine the cost effectiveness of using point-intercept data to prestratify variability in subsequent clipped-biomass sampling on a sagebrush–bunchgrass rangeland site in southern Idaho. Point-intercept and biomass data were obtained for shrub, grass, and forb vegetation in 90 1-m plots. These data were used to develop a synthetic population of 10 000 simulated plots for conducting sensitivity analysis on alternative doublesampling scenarios. Monte Carlo simulation techniques were used to determine the effect of sampling design on cost and variability of biomass estimates as a function of point-intercept sample size (i), number of point-intercept sample strata (s), and number of biomass samples per stratum (m). Minimization of variability in biomass estimates were always obtained from double-sampling scenarios in which a single median biomass estimate was obtained for a given stratum in the point-intercept data. Double-sampling strategies in which half of the point-intercept plots were also measured for biomass yielded a cost savings of 39% with a reduction in biomass-sample precision of 18% 6 4 SD. The relative loss of precision in biomass estimates (62% 6 12 SD) became equal to the relative cost savings of double sampling for scenarios in which the ratio of point-intercept/ biomass samples exceeded a value of five.