Validation of DNA metabarcoding of fecal samples using cattle fed known rations
Introduction
Quantifying diet composition of free-roaming livestock and wildlife can inform decision-making by producers to achieve desired objectives through predicting outcomes of grazing strategies and identifying key plant species to guide management (Holechek et al., 1982). Such information is also critical because livestock and wildlife consume different quantities of nutrients temporally as forage quality and animal nutritional needs asynchronously fluctuate (Scasta et al., 2016). While quantitative dietary information is known to be useful, diet composition of free-roaming animals in extensive and spatially heterogeneous environments has been persistently difficult to quantify (Garnick et al., 2018). This challenge persists because (1) animals can be problematic to locate spatially, (2) plant material differentially breaks down in ruminant and non-ruminant animals, and (3) identification of plant fragments is difficult, complex, and time consuming (Holechek et al., 1982).
Diet composition quantification has become increasingly sophisticated (Garnick et al., 2018). Early techniques used direct observation of animal bite counts (Sanders et al., 1980) with subsequent development of more invasive methods including surgical stomach analysis, and rumen and esophageal fistulation (Holechek et al., 1982). More recently, non-invasive post-digestive techniques such as fecal microhistology and near-infrared reflectance spectroscopy (fNIRS) technology has been used (Lyons and Stuth, 1992; Decruyenaere et al., 2015; Kartzinel et al., 2015; Ottavian et al., 2015; Núñez-Sánchez et al., 2016). Despite technological advances, individual techniques continue to have challenges and constraints (Garnick et al., 2018). For the bite count method, plant identification skills are required, animals must be spatially located, and only a single animal is observed at a time (Sanders et al., 1980). For surgical techniques, plant material may be masticated beyond recognition, sample size is small, methods are time intensive, and problems can arise from surgical procedures (Holechek et al., 1982). Constraints of fecal microhistology are that plant material in the feces may not be proportional to that consumed by the animal, trained observers are required, slow turn around time for sample analyses is slow, and bias and errors can be high (Holechek et al., 1982; Garnick et al., 2018). Accurately quantifying diet botanical composition at the species-level has been problematic for fNIRS (Garnick et al., 2018). In addition, many fNIRS studies only quantified a single plant species in the diet (Walker et al., 1998, 2002; Valiente et al., 2004) or very simple diets with only 3 or so plant species (Landau et al., 2004), although more recent attempts have developed fNIRS calibrations for complex diets for up to 25 species in the diet (Glasser et al., 2008).
Technological advances now permit reconstruction of diet botanical composition using fecal samples from free-roaming animals through the use of DNA (fDNA) metabarcoding using a single-locus identification system with implications for dietary protein composition (Valentini et al., 2009; Craine et al., 2015; Kartzinel et al., 2015). Advantages of this technique are similar to fNIRS with non-invasivity and large sample sizes, but determination of diet botanical composition at the species level is quicker given the automation of reading numerous plant DNA sequences at a single time compared to the need for individual fNIRS calibrations per plant species (Pompanon et al., 2012). fDNA metabarcoding has focused on the trnL intron located in the plant chloroplast using c-h primers and estimating species diet composition through associations to their plant species-specific dietary protein content (Craine et al., 2015; Kartzinel et al., 2015).
Recent fDNA applications have (1) enumerated dietary niche partitioning among African herbivores (Kartzinel et al., 2015), (2) identified seasonal and regional diet fluctuations for bison (Bison bison; Craine et al., 2015), and (3) quantified geographical patterns of cattle (Bos taurus) diets in North America (Craine et al., 2016). Validation of the fDNA metabarcoding technique is lacking through comparison of known fed diets to laboratory results. Previous work has generated notable surprises in diet botanical composition of free-roaming herbivores (Craine et al., 2015), including previously unknown and non-graminoid major diet components in cattle diets at a continental scale (Craine et al., 2016), and greater composition of non-graminoid plants in bison diets (Craine et al., 2015). In addition, identification of diet components with resolution only to the family or genus level in some cases has been identified as a limitation, even if additional primers were used (Garnick et al., 2018). Thus, many questions exist regarding more widespread use of fDNA in scientific and managerial applications of free-ranging animals. We therefore conducted a blind cattle feeding trial using known fed rations (Pompanon et al., 2012) with the objective to validate the use of fDNA metabarcoding. We asked: Does fDNA metabarcoding accurately identify known major and minor diet components?
Section snippets
Animal management
Five non-gestating and non-lactating 2-year old Angus-cross heifers (weights ranging from 416 to 527 kg) were used in a 6-week feeding study. Animal care and use was approved under University of Wyoming IACUC protocol # 20170208DS00258-01. Heifers were penned individually in adjoining 160 m2 (4.45 m × 36 m) well-drained pens that included shade, outdoor access, and automatic water availability.
Diet construction and feeding
A unique diet was fed each week over the 6-week study period and every heifer received every diet
Can fecal DNA metabarcoding accurately identify known major diet components?
Fed diets and laboratory non-discretionary diet results differed due to misidentification of major C3 and C4 fed diet components with fDNA techniques. For the homogeneous C3 diet comprised of A. arundinaceus, the main species identified using fDNA techniques was timothy (Phleum pratense (L.)). This incorrect attribution was due to one OTU co-representing P. pratense and A. arundinaceus, along with several other Alopecurus species, at the 97% base pair matching level. Incorrect blind
Discussion
Our findings indicate several important limitations in the use of current fDNA methodology to quantify dietary composition of livestock. Specifically, fDNA results obtained and used without local knowledge of plant species in the sampled area or known diet (the case in our experiment) could lead to incorrect species level dietary compositions. It is possible to reach erroneous conclusions if fDNA is applied carte blanche in a manner similar to how more user-ready analyses of carbon, nitrogen,
Conclusions
fDNA metabarcoding is a novel technology with much potential for reconstruction of botanical composition of diets for free roaming animals, but caution must be emphasized when using this methodology. Our validation study suggests that the results can be incorrect and that three steps can help ensure more correct interpretation of the results. First, improving the content of the reference library to which OTUs are compared can enhance accuracy. Second, understanding that a single OTU can be
Animal care and welfare
Animal care and use was approved under University of Wyoming Institutional Animal Care and Use Committee (IACUC; Protocol # 20170208DS00258-01).
Funding
Support was provided by Western Sustainable Agriculture Research and Education (SARE) Graduate Student Project grant (GW17-059) and the University of Wyoming – Agricultural Experiment Station (UW-AES) through a USDA National Institute of Food and Agriculture McIntire Stennis Project—“Animal-Plant Interaction Ecology Grant on Wyoming Rangelands” (2015-2020, Project# WYO-559-15).
Declaration of Competing Interest
After project completion, T.J. began working for Jonah Ventures, the laboratory that conducted the fDNA metabarcoding analyses. The authors declare no other conflicts of interest.
Acknowledgements
Support was provided by Western Sustainable Agriculture Research and Education (SARE) Graduate Student Project grant (GW17-059) and the University of Wyoming – Agricultural Experiment Station (UW-AES) through a USDA National Institute of Food and Agriculture McIntire Stennis Project—“Animal-Plant Interaction Ecology Grant on Wyoming Rangelands” (2015-2020, Project# WYO-559-15).
References (23)
- et al.
Prediction error and repeatability of near infrared reflectance spectroscopy applied to faeces samples in order to predict voluntary intake and digestibility of forages by ruminants
Anim. Feed Sci. Tech.
(2015) - et al.
Assessment of animal-based methods used for estimating and monitoring rangeland herbivore diet composition
Rangeland Ecol. Manag.
(2018) - et al.
Evaluation of botanical and chemical composition of sheep diet by using faecal near infrared spectroscopy
Anim. Feed Sci. Tech.
(2016) - et al.
Application of near infrared reflectance spectroscopy (NIRS) on faecal samples from lactating dairy cows to assess two levels of concentrate supplementation during summer grazing in alpine pastures
Anim. Feed Sci. Tech.
(2015) - et al.
Meta-analysis of diet composition and potential conflict of wild horses with livestock and wild ungulates on western rangelands of North America
Rangeland Ecol. Manag.
(2016) - et al.
DNA barcoding for ecologists
Trends Ecol. Evol. (Amst.)
(2009) - et al.
Climatic warming and the future of bison as grazers
Sci. Rep.
(2015) - et al.
Continental-scale patterns reveal potential for warming-induced shifts in cattle diet
PLoS One
(2016) - et al.
A fecal near-infrared reflectance spectroscopy-aided methodology to determine goat dietary composition in a Mediterranean shrubland
J. Anim. Sci.
(2008) - et al.
Botanical composition determination of range herbivore diets: a review
J. Range Manage.
(1982)