• Precision, Repeatability, and Efficiency of Two Canopy-Cover Estimate Methods in Northern Great Plains Vegetation

      Symstad, Amy J.; Wienk, Cody L.; Thorstenson, Andy D. (Society for Range Management, 2008-07-01)
      Government agencies are subject to increasing public scrutiny of land management practices. Consequently, rigorous, yet efficient, monitoring protocols are needed to provide defensible quantitative data on the status and trends of rangeland vegetation. Rigor requires precise, repeatable measures, whereas efficiency requires the greatest possible information content for the amount of resources spent acquiring the information. We compared two methods—point frequency and visual estimate—of measuring canopy cover of individual plant species and groups of species (forbs vs. graminoids, native vs. nonnative) and plant species richness. These methods were compared in a variety of grassland vegetation types of the northern Great Plains for their precision, repeatability, and efficiency. Absolute precision of estimates was similar, but values generally differed between the two sampling methods. The point-frequency method yielded significantly higher values than the visual-estimate method for cover by individual species, graminoid cover, and total cover, and yielded significantly lower values for broadleaf (forb + shrub) cover and species richness. Differences in values derived by different sampling teams using the same method were similar between methods and within precision levels for many variables. Species richness and median species cover were the major exceptions; for these, the point-frequency method was far less repeatable. As performed in this study, the visual-estimate method required approximately twice the time as did the point-frequency method, but the former captured 55% more species. Overall, the visual-estimate method of measuring plant cover was more consistent among observers than anticipated, because of strong training, and captured considerably more species. However, its greater sampling time could reduce the number of samples and, therefore, reduce the statistical power of a sampling design if time is a limiting factor.