10.07.2015 Views

Recovery Plan for the Northern Spotted Owl - DRAFT

Recovery Plan for the Northern Spotted Owl - DRAFT

Recovery Plan for the Northern Spotted Owl - DRAFT

SHOW MORE
SHOW LESS
  • No tags were found...

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Table A.3. Powera <strong>for</strong> various designs of <strong>the</strong> roadside surveys.Number ofStationsProportion of Stations Surveyed Every Fourth YearbSurveyedper Year 0.00 0.20 0.40 0.60 0.80 1.00200 0.32 0.33 0.34 0.35 0.37 0.38400 0.51 0.53 0.54 0.56 0.58 0.60600 0.66 0.68 0.69 0.71 0.73 0.75800 0.77 0.78 0.80 0.81 0.83 0.851,000 0.84 0.86 0.87 0.88 0.90 0.91probability of detecting <strong>the</strong> trend. In <strong>the</strong>se simulations, <strong>the</strong> average proportional cThe change in <strong>the</strong> population was -3.4 percent per year.9 The remaining stations were surveyed every year., .... ... .. .. ................ ............ ~~~~~~~~~~~~~~~~~~......... ...........256stations were required <strong>for</strong> 80 percent power when all stations were visited eachyear, but only 370 stations were required <strong>for</strong> <strong>the</strong> same power when all stationswere visited at 4-year intervals. The lattice design thus requires that moreroutes be selected, but permits a smaller number to be surveyed in each year(to achieve a given power) than a design in which each route is surveyed eachyear.The decision on whe<strong>the</strong>r to adopt a lattice design can be postponed until <strong>the</strong>second year of monitoring. At that time, a decision must be made to revisitevery route surveyed in <strong>the</strong> first year or to temporarily drop some routes andintroduce a corresponding set of new routes. If <strong>the</strong> new set were spatiallyinterpenetrating with those that were dropped, <strong>the</strong>n <strong>the</strong> logistics of <strong>the</strong> programwould not be compromised by this tactic and <strong>the</strong> geographic dispersionof <strong>the</strong> sample would remain essentially <strong>the</strong> same. (Such a spatially interpenetratingdesign on a 4-year cycle is being implemented by <strong>the</strong> U.S. EnvironmentalProtection Agency in its newly instituted Environmental Monitoring andAssessment Program, EMAP.)Sample size requirementsSample size includes <strong>the</strong> number of stations (or routes) monitored per year and<strong>the</strong> number of years during which monitoring continues. Many factors affectsample size requirements. Among <strong>the</strong> most important are <strong>the</strong> sampling designused to collect <strong>the</strong> data, <strong>the</strong> desired level of precision or power, and <strong>the</strong> degreeto which <strong>the</strong> true trend con<strong>for</strong>ms to <strong>the</strong> statistical (e.g., linear) model. Atpresent, many of <strong>the</strong> factors needed to estimate sample size requirements <strong>for</strong><strong>the</strong> roadside survey method are unknown. As a result, our analyses arepreliminary and undoubtedly will be altered as more in<strong>for</strong>mation is obtained.Despite <strong>the</strong>se problems, <strong>the</strong> analyses provide an indication of approximatesample size requirements which will be useful in deciding on <strong>the</strong> initial specifications<strong>for</strong> <strong>the</strong> monitoring program.Two analyses of sample size requirements are presented. The first is based ona series of computer simulations in which statistical methods were used tocreate data sets with known underlying trends. Estimates of <strong>the</strong>se trends <strong>the</strong>nwere calculated and compared to <strong>the</strong> known, true values. The process wasrepeated with different parameter sets to investigate sample size requirementsand identify factors that affect <strong>the</strong>m. This analysis permitted a detailed investigationof sample size requirements but required an assumption that <strong>the</strong>computer model was realistic. For <strong>the</strong> second analysis, we obtained data from

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!