LAUSR

Abstract Diet composition of free-roaming animals has been persistently difficult to quantify. Fecal DNA (fDNA) metabarcoding is an emerging novel technology to reconstruct botanical composition of diets. Validation studies comparing fDNA metabarcodion to fed diets are lacking. A blind 6-week feeding study using 5 heifers (416–527 kg) fed different diets weekly with fecal samples collected from individuals 7 days after the introduction of each new diet was conducted. Diets were: (1) C3 grass hay (creeping meadow foxtail (Alopecurus arundinaceus)), (2) C4 grass hay (foxtail millet (Setaria italica), (3) C3 grass hay + C4 grass hay + alfalfa hay (Medicago sativa) (equal 1/3 proportions dry matter, 17.0%, 30.5%, and 52.5% by crude protein contribution, respectively), (4) C3 grass hay + C4 grass hay + alfalfa hay (equal proportions) + minor component of Wyoming big sagebrush (Artemisia tridentata ssp. wyomingensis) leaves (fed at 25 grams per heifer per day equal to 0.29% by dry matter contribution; crude protein content unknown), (5) alfalfa hay + minor components of crested wheatgrass (Agropyron cristatum), western wheatgrass (Pascopyrum smithii), and blue grama (Bouteloua gracilis) that included steams, leaves, and seed heads (varying proportions by dry matter ranging from 0.02 to 0.15% but 0.05%, 0.04%, and 0.01% by crude protein contribution), and (6) alfalfa hay. Single species diets were fed ad libitum. Diets with multiple forage species were fed at metabolic weight-based amounts with minor diet components introduced 2-3 days after major diet components. At the species level, fed diets and fDNA metabarcoding diets were dissimilar due to misidentification of major C3 and C4 diet components. For C3, A. arundinaceus was under the same Operational Taxonomic Unit (OTU) as Phleum pratense (not present in the C3 sourced hay), so blind identification was initially wrong. For C4, S. italica was identified as Setaria palmifolia, which does not grow in the region. Microscopic evaluation of seeds for both misidentifications verified the correct species identification and the fDNA error. Minor dietary components were detected in only 1 of 5 heifers for all species. Increasing the utility of fDNA metabarcoding for accurately determining the diet composition of free-roaming animals will necessitate: 1) improving the reference library of voucher species specimens, 2) understanding by technical and scientific staff that multiple species may be attributed to the same OTU, and 3) greater verification of fDNA metabarcoding results using field-derived plant composition data, especially in rangeland settings where plant diversity can be intrinsically high.