Laserfiche WebLink
when there were fewer transects, but seemed to have little effect when the number of <br />transects increased. <br />Confidence interval coverage of the tree density estimates ranged from 97.2% to <br />100% for Cottonwood Ranch and from 95.2% to 100% for Jeffrey Island. Confidence <br />interval coverage of the shrub /sapling density estimates ranged from 78.2% to 96.6% for <br />Cottonwood Ranch and from 87.7% to 99.8% for Jeffrey Island. Confidence interval <br />coverage increased as the number of sample transects increased and as the spacing of <br />points along the transect decreased (Figures 14 through 17). <br />The half -width of the confidence interval on tree density, an estimate of relative <br />precision, ranged from 0.51 times the true value (51 %) to 0.15 times the true value (15 %) <br />for Cottonwood Ranch and from 0.66 times the true value (66 %) to 0.19 times the true <br />value (19 %) for Jeffrey Island. The half -width of the confidence interval on <br />shrub /sapling density ranged from 1.08 times the true value (108 %) to 0.35 times the true <br />value (35 %) for Cottonwood Ranch and from 0.92 times the true value (92 %) to 0.26 <br />times the true value (26 %) for Jeffrey Island. Precision of the estimates increased (ratio <br />decreased) as the number of transects increased (figures 18 through 21). In general, the <br />spacing of the point- centered quarter plots had little effect on the ratio because the <br />estimate of variance is based on the number of transects. <br />Program -level Step -point Level of Effort <br />Bias and precision of the step -point estimates could not be calculated because the data <br />were compiled together in the field for each land cover transect and lacked a sample unit <br />for re- sampling within a transect. <br />Recommendations for Modifications to the Management -Level <br />Vegetation Monitoring Protocol <br />1. Design vegetation monitoring for each land and water management action <br />separately. <br />Since management actions will be implemented with the intention to clear a forest to an <br />open channel, clear a forest to a grassland, create an emergent wetland, or remove <br />seedlings from an island; different vegetation sampling methods should be designed to <br />assess the effectiveness of management. Ideally, the vegetation monitoring will be <br />designed with sampling both before and after the management action in the assessment <br />and associated control areas to provide information for making adaptive management <br />decisions related to the impact of land and water management actions on the vegetation <br />community. <br />2 Recommend reducing the amount of data collected with the point- centered- <br />quarter and step -point sampling. <br />The analyses presented in this report indicate a decrease in the density of sample points <br />for quantifying tree and shrub /sapling density and percent plant cover in the herbaceous <br />vegetation layer in forests would not increase the bias or decrease the precision to <br />unacceptable levels. <br />