PS 74-132
A handy method for estimating the cover of forest herbs

Thursday, August 13, 2015
Exhibit Hall, Baltimore Convention Center
Christopher A. Walter, Biology, West Virginia University, Morgantown, WV
Mark B. Burnham, Biology, West Virginia University, Morgantown, WV
Frank S. Gilliam, Department of Biological Sciences, Marshall University, Huntington, WV
William T. Peterjohn, Biology, West Virginia University, Morgantown, WV

Herbaceous cover data are essential in addressing several ecological phenomena, including responses to experimental manipulations, successional change, and mapping the spread of invasive species. The more popular methods for measuring herbaceous cover use cover-abundance classes. Although these methods are quick, they rely on subjective classification which can lead to errors as large as 20%. Line-intercept sampling and image-analysis methods are more accurate, but only effective in lower-density vegetation. The most accurate method, point-intercept sampling, has the major disadvantage of being very time consuming, particularly in areas with complex canopy structure such as the forest herbaceous layer. A simple approach to estimate cover is presented, whereby an observer compares the area of their hand to the area of foliar surfaces. We tested the accuracy of this hand-area method for estimating cover and aboveground biomass in the forest herbaceous layer at four scales: (i) among individual plants of the same species (Rubus allegheniensis), (ii) among individual plants of differing species, (iii) across groups of individual species, and (iv) all plants within 1-m2 plots. To test the method at each scale, hand-measured leaf area was regressed against the same leaves measured using a leaf area meter, and against biomass.


The hand-area method (HA) was very accurate and consistent in predicting leaf area measured via meter (LA) at all scales (average r2 = 0.921). When LA was regressed against HA in R. allegheniensis, the r2 was 0.957. Among individual plants of differing species, r2 was 0.976 and 0.930 for groups of individual species. Within 1-m2 subplots, r2 of the LA vs. HA regression was 0.821. HA was also a good predictor of aboveground biomass for R. allegheniensis (r2 = 0.866), but not among the groups of individual species (r2 = 0.638) or within 1-m2 subplots (r2 = 0.193). The slope of the regression lines of LA vs. HA were usually approximately 38% lower than the actual area of the observer’s hand – indicating a consistent overestimation of actual leaf area. However, this bias can be subtracted to improve prediction accuracy while preserving precision. The hand-area method is a quick, accurate, and reliable way to estimate the cover of forest herbs. We believe the precision of the method is due to the broad-leafed nature of temperate forest herbs and suspect that other reference shapes could be used in areas that contain plants with less ovate leaves or particularly sinuous leaf margins.