Evaluating Multiple Rating Methods Utilized in Turfgrass Weed Science

Abstract

Turfgrass weed scientists commonly use visual ratings (VR) to assign a numerical value to a turfgrass or weed response. These ratings lack quantifiable numerical values and are considered subjective. Alternatives to VR, including line intersect analysis (LIA) and digital image analysis (DIA), have been used to varying extents in turfgrass research. Alternatives can be expensive, labor intensive, and can require extensive calibration and increased time for data acquisition. Minimal research has been conducted evaluating rating methods used in turfgrass weed science. Trials were conducted in 2007 and 2008 to evaluate ratings methods used to quantify large crabgrass populations as influenced by tall fescue mowing height (2.5, 5.1, 7.6, and 10.2 cm). Percent large crabgrass cover was assessed utilizing VR, LIA, and DIA to determine if differences existed among evaluation methods. Pairwise comparisons, Pearson's correlation, and linear regression were performed to compare evaluations. All rating methods were significantly correlated to one another. Differences of large crabgrass cover estimates existed between LIA and DIA data at all mowing heights and between VR and DIA data at the 7.6 and 10.2 cm mowing heights in 2007. Authors believe that shadows produced by the turf canopy at higher (≥ 7.6 cm) mowing heights increased DIA estimates of large crabgrass cover. At trial initiation in 2007, researchers did not capture calibration images because the methodology to eliminate a shadow influence using a standard digital image had not been published. Additional DIA calibration in 2008 corrected for canopy shadows, and no differences were observed in large crabgrass cover between all evaluation methods indicated by nonsignificance pairwise comparisons and estimated regression parameters. These data indicate VR are no different than LIA or DIA in estimating large crabgrass cover as affected by tall fescue mowing height.