Feedlot Hair Sheep Models Predict Fecal Nitrogen Better Than Urinary Loss

A multi-study analysis has produced equations for estimating nitrogen losses from feedlot hair sheep, with the clearest performance for fecal nitrogen, moderate accuracy for total nitrogen and substantially weaker prediction of urinary nitrogen. The development database combined 380 individual observations from 18 studies of confined hair sheep weighing 10.3 to 48.8 kilograms. It included 223 intact males, 77 castrated males and 80 females.

The most represented genotypes were Santa Ines, Santa Ines–Dorper crosses, Brazilian Somali, Dorper and Morada Nova. The resulting models were tested against an external database of 162 observations from seven additional studies, revealing that nitrogen intake was the most useful overall predictor but that no equation performed equally well for every route of excretion. The study addressed a practical gap in animal-nutrition research.

Equations available for sheep had mainly been developed in temperate production systems, where breeds, feeds and nutritional conditions differ from those encountered in tropical feedlots. Reliable estimates are needed because nitrogen excretion represents both a loss of feed protein and a potential environmental burden. In the assembled development database, mean dry matter intake was 947.95 grams per day and mean nitrogen intake was 21.28 grams per day.

Fecal nitrogen averaged 6.26 grams per day, urinary nitrogen 4.92 grams per day and total nitrogen 11.17 grams per day. Mean body weight was 26.90 kilograms, and the diets contained an average of 13.4% crude protein. Researchers tested whether relationships between intake and excretion differed among intact males, castrated males and females, or among the five represented genotypes.

The coefficients did not differ significantly by sex class or genotype, allowing general equations to be developed rather than separate models for each group. The statistical analysis used meta-analytical techniques to account for random effects among the 18 source experiments, including random intercepts and slopes. Candidate predictors included body weight, metabolic body weight, dry matter intake, total digestible nutrient intake and nitrogen intake.

Relationships were examined for both linear and nonlinear forms, and the final equations were evaluated with the Model Evaluation System using the independent database. Fecal nitrogen was strongly associated with all three major intake variables. Pearson correlation coefficients were 0.83 for dry matter intake, 0.78 for nitrogen intake and 0.76 for total digestible nutrient intake, with all three relationships significant at P < 0.001.

The analysis generated linear equations based on dry matter intake, nitrogen intake and total digestible nutrient intake. Fecal nitrogen, dry matter intake and nitrogen intake were expressed in grams per day. In external testing, the dry matter model produced an R² of 0.65, a concordance correlation coefficient of 0.75 and a root mean squared error of prediction of 1.76 grams per day.

The equation driven by nitrogen intake performed best among the fecal models, achieving an R² of 0.68, a concordance correlation coefficient of 0.82 and a root mean squared error of prediction of 1.63 grams per day. The total-digestible-nutrient model was less successful, with an R² of 0.55, concordance of 0.63 and an error of 1.94 grams per day. Nitrogen-intake and total-digestible-nutrient predictions showed significant departures from the ideal intercept of zero and slope of one, indicating residual mean or proportional bias.

The investigators nevertheless concluded that dry matter intake should be recommended for predicting fecal nitrogen because it is readily available in feeding systems and does not require laboratory measurement of feed nitrogen. Urinary nitrogen was more difficult to predict. Its strongest observed relationship was with nitrogen intake, with a correlation of 0.76, while dry matter intake was also positively associated with urinary output at 0.72.

The researchers fitted allometric and exponential equations to nitrogen intake. In the independent validation, both urinary models had an R² of 0.48. The allometric equation had a concordance correlation coefficient of 0.36 and a root mean squared error of prediction of 4.44 grams per day, whereas the exponential version achieved concordance of 0.49 and an error of 3.30 grams per day.

The validation statistics explain why urinary nitrogen received a lower performance rating. The exponential urinary model did not differ significantly from an intercept of zero or a slope of one, suggesting better overall agreement than the allometric version, but its concordance coefficient remained below 0.5. The allometric model had a significant mean bias associated with an intercept different from zero.

The study’s authors cautioned that variation among experiments and differences in urine-collection methods may have contributed to this uncertainty. Some source studies used continuous collections over four to seven consecutive days, while others relied on spot sampling with creatinine-based estimates of urinary volume. Continuous collection integrates changes over a full day, whereas spot samples can be influenced by post-meal nitrogen metabolism and daily patterns of renal clearance.

Total nitrogen, calculated as the sum of fecal and urinary excretion, was modeled as a function of nitrogen intake. Both the allometric and exponential total-nitrogen models achieved an R² of 0.72 in external testing. The allometric model had a concordance correlation coefficient of 0.69 and a root mean squared error of prediction of 4.40 grams per day.

The exponential model performed better, with concordance of 0.73 and an error of 3.63 grams per day, although both models showed significant mean and proportional bias. Nitrogen intake was the strongest common predictor for total output, with a correlation of 0.86; dry matter intake had a correlation of 0.85. Across the database, fecal and urinary losses represented 29.41% and 23.12% of nitrogen intake, respectively, for a combined measured excretion of 52.53% of intake.

The mean urinary-to-fecal nitrogen ratio was 0.79, lower than values reported in several earlier sheep studies. The authors said this could reflect the wider range of genotypes and diets represented in the database, or it could arise from methodological challenges in measuring excretion that increased variability. The results therefore provide estimation tools rather than universal guarantees.

Their scope is confined to the feedlot hair sheep represented in the source studies, and the equations should not automatically be treated as validated for other species, grazing animals, climates or production systems. Additional dietary and physiological variables may be needed to improve precision, particularly for urinary nitrogen. The findings give researchers and feedlot managers a clearer picture of where prediction models can be most useful.

A linear equation based on nitrogen intake performed best for fecal nitrogen in the external evaluation, while the dry matter equation offered a practical alternative when feed nitrogen analysis is unavailable. Nitrogen intake was also the predictor used for the urinary and total-nitrogen models, with the exponential versions outperforming the allometric alternatives in the reported validation. However, the urinary models remained limited by lower concordance and larger prediction errors than the fecal and total-nitrogen equations.

Subject of Research: Prediction of fecal, urinary and total nitrogen excretion in feedlot hair sheep using intake variables

Article Title: Prediction models of nitrogen excretion in feedlot hair sheep

Article References: Clarindo, A. J. C., Brito Neto, A. S., Viana Neto, A. M., Araujo, V. F. C., Silva, L. W. L., Herbster, C. J. L., Alves, F. G. S., Rego, A. C., Moura, A. A. A. N., Ferreira, M. A., Moreno, G. M. B., Oliveira, R. L., Bezerra, L. R., Chay-Canul, A. J., Ferreira, E. M., Santos, S. A., & Pereira, E. S. (2026). Prediction models of nitrogen excretion in feedlot hair sheep. Frontiers in Animal Science, 7, Article 1970735. https://doi.org/10.3389/fanim.2026.1970735

Image Credits: AI Generated

DOI: 10.3389/fanim.2026.1970735

Keywords: hair sheep, feedlot nutrition, nitrogen excretion, fecal nitrogen, urinary nitrogen, total nitrogen, prediction equations, meta-analysis, animal nutrition, tropical livestock, Prediction, models