Laserfiche WebLink
as the mean (over 1000 iterations) of the ratio of the estimated half-width of the density <br />confidence interval, divided by the estimated density. Low values of this ratio indicate <br />more precision associated with the estimates. <br />Relative bias of the tree density estimates ranged from 40.9% for the estimates <br />with 1 sample point to 1.7% for the estimates with 20 sample points. Relative bias of the <br />shrub /sapling density estimates ranged from 3653.1% for the estimates with 1 sample <br />point to 3.3% for the estimates with 35 sample points. As the number of sample points <br />increased, the relative bias decreased for both tree and shrub /sapling density estimates <br />(Figures 2 and 3). This decrease was most pronounced for the first 5 points for the tree <br />and shrub /sapling density estimates. <br />Confidence interval coverage for tree density ranged from 49.4% for the estimates <br />with 2 sample points to 89.5% for the estimates with 26 sample points. Confidence <br />interval coverage for shrub /sapling density ranged from 37.7% for the estimates with 2 <br />sample points to 93.3% for the estimates with 27 sample points. As the number of <br />sample points increased, the confidence coverage increased for both tree and <br />shrub /sapling density estimates (Figure 4). This increase was most pronounced for the <br />first 9 points for both the tree and shrub /sapling estimates. <br />The half -width of the confidence interval on tree density, an estimate of relative <br />precision, ranged from 1.12 times the true value (112 %) for the estimates with 2 sample <br />points to 0.20 times the true value (20 %) for the estimates with 35 sample points. The <br />half -width of the confidence interval on shrub /sapling density ranged from 1.72 times the <br />true value (172 %) for the estimates with 2 sample points to 0.37 times the true value <br />(37 %) for the estimates with 35 sample points. As the number of sample points <br />increased, the precision increased (ratio decreased) for both tree and shrub /sapling <br />density estimates (Figure 5). This increase was most pronounced for the first 8 points for <br />the tree estimates and for the first 12 points for the shrub /sapling estimates. <br />Phase 1 Step -point Level of Effort <br />We evaluated the adequacy of the 17 survey locations to sample species richness <br />with the step -point data. We recorded the number of new species detected at each point <br />cumulatively through a random ordering of the points. A graph of species richness <br />obtained cumulatively though the 17 points (Figure 6) demonstrates a decrease in the <br />number of new species detected in each successive sampling point. When sampling is <br />adequate for species richness estimates, the species -point curve will level off so that <br />adding new points does not provide the dataset with any new species. With this data, the <br />last 3 step -point samplings did not add any new species. <br />We evaluated the bias, confidence interval coverage, and precision of relative <br />percent plant cover estimates under alternate sampling intensities. The relative percent <br />plant cover estimates were recalculated using a dataset where the step point data (species <br />and number of hits) had been randomly re- sampled with replacement from the original <br />dataset (17 points). The re- sampled datasets were created to simulate from 2 (1 point per <br />44 acres) to 35 (1 point per 2.5 acres) points with 50 or 100 step - points taken at each <br />point. The sampling procedure and calculation of the estimates was repeated 1000 times. <br />Relative bias, confidence interval coverage, and relative precision of relative percent <br />plant cover were calculated as above for each species. <br />