Using blood urea nitrogen to predict nitrogen excretion and efficiencyof nitrogen utilization in cattle, sheep, goats, horses, pigs, and rats1

R. A. Kohn2, M. M. Dinneen, and E. Russek-Cohen

Department of Animal and Avian Sciences, University of Maryland, College Park 20742

ABSTRACT: The objectives of this study were toevaluate the potential for using blood urea N concentra-tion to predict urinary N excretion rate, and to developa mathematical model to estimate important variablesof N utilization for several different species of farmanimals and for rats. Treatment means (n = 251) from41 research publications were used to develop mathe-matical relationships. There was a strong linear rela-tionship between blood urea N concentration (mg/100mL) and rate of N excretion (g�d−1�kg BW−1) for all ani-mal species investigated. The N clearance rate of thekidney (L of blood cleared of urea�d−1�kg BW−1) wasgreater for pigs and rats than for herbivores (cattle,sheep, goats, horses). A model was developed to esti-mate parameters of N utilization. Driving variables forthe model included blood urea N concentration (mg/100mL), BW (kg), milk production rate (kg/d), and ADG (kg/d), and response variables included urinary N excretion

Key Words: Blood Urea Nitrogen, Nitrogen Excretion, Nitrogen Utilization Efficiency

2005 American Society of Animal Science. All rights reserved. J. Anim. Sci. 2005. 83:879–889

Introduction

Nitrogen losses from agriculture to air and water areperceived to be a major global environmental problem(NRC, 2003). Livestock farming is a significant sourceof reactive N in the environment. Of all the NH3 andN2O released into the environment because of humanactivity, approximately 70 and 30%, respectively, areestimated to arise from livestock farming (van Aar-denne et al., 2001). The purpose of livestock farming isto convert the carbohydrates and proteins in animalfeed to food sources for humans; however, only 5 to 30%of animal feed N usually meets this goal. The rest isexcreted by animals and can escape into the envi-ronment.

1A contribution from the Maryland Agric. Exp. Stn.2Correspondence: Animal Sciences Building, #142 (phone: 301-

405-4583; fax: 301-314-9059; e-mail: rkohn@umd.edu).Received August 4, 2004.Accepted January 4, 2005.

879

rate (g/d), fecal N excretion rate (g/d), rate of N intake(g/d), and N utilization efficiency (N in milk and gainper unit of N intake). Prediction errors varied widelydepending on the variable and species of animal, withmost of the variation attributed to study differences.Blood urea N concentration (mg/100 mL) can be usedto predict relative differences in urinary N excretionrate (g/d) for animals of a similar type and stage ofproduction within a study, but is less reliable acrossanimal types or studies. Blood urea N concentration(mg/100 mL) can be further integrated with estimatesof N digestibility (g/g) and N retention (g/d) to predictfecal N (g/d), N intake (g/d), and N utilization efficiency(grams of N in milk and meat per gram of N intake).Target values of blood urea N concentration (mg/100mL) can be backcalculated from required dietary N (g/d) and expected protein digestibility. Blood urea N canbe used in various animal species to quantify N utiliza-tion and excretion rates.

The ability to estimate N excretion rate of animalsmight be used to minimize pollution as well as decreasethe use of excess feed protein (NRC, 2003). For example,accurate estimates of manure N production rate mayimprove models to predict farm-level losses of N, andimprove planning for manure handling. In addition,plasma or blood urea N (BUN) concentration may beuseful as an indicator of protein status within a groupof animals, and could help to fine-tune diets or identifyproblems with a feeding program. Milk urea N concen-tration is used to predict N excretion in dairy cows(Jonker et al., 1998; Kohn et al., 2002). Blood andplasma urea N concentration are proportional to milkurea N in dairy cows (Baker et al., 1995) and thereforemay be useful as a predictor in much the same way.Differences in predictor values for various species mayprovide insight into differences in N utilization inthese species.

Our first objective was to evaluate the potential forpredicting urine N excretion rate (g/d) from BUN con-centration (mg/100 mL) and BW (kg) for several domes-tic species. The second objective was to develop and

Kohn et al.880

Table 1. Dataset from the literature for development of the models

Reference Means Animals Observations Designa

CattleArchibeque (2001)b 4 8 32 RLSArchibeque (2002)c 4 8 32 RLSBailey (1989) 15 45 540 RMDinn et al. (1998) 3 18 54 RLSEstermann et al. (2002)b 6 32 32 RBFisher et al. (2000) 3 18 54 RLSHaig et al. (2002)c 3 18 18 CRHoffman (2001)b 4 40 40 RBHuntington (1984)b 2 4 8 RMKaiser et al. (1983)b 9 18 54 RLSKnaus et al. (1998) 4 4 16 LSKoenig et al. (1982) 4 8 32 RLSKoster et al. (1997)b 5 5 25 LSLykos et al. (1997)b 3 6 18 RLSTheurer et al. (2002)c 2 7 14 RLS

SheepAhmad et al. (1995) 4 24 24 RBAlawa (1991)b 22 8 88 RMBohnert et al. (1999)c 4 8 32 RLSCole (1999) 10 10 50 RLSFerrell et al. (1999)b 5 6 30 LSForster et al. (1991) 5 30 30 RBHettiarachchi et al. (1999) 6 16 33 RLSSmith and Boling (1984) 4 8 32 RLSSutoh et al. (1993)b 2 4 4 CRWoodward and Reed (1997) 7 7 49 LS

GoatsLindberg (1989)b 17 6 51 RMWoodward and Reed (1997) 7 7 49 LSVerstegen et al. (1991)b 8 32 32 CR

HorsesMartin et al. (1996) 3 3 9 LS

PigsVan Barneveld et al. (1995)c 4 8 8 IBDunn and Speer (1991) 11 11 61 RLSMosenthin et al. (1992b) 2 5 10 RLSMosenthin, et al. (1992a) 2 4 8 RLSZerves and Zijlstra (2002b)c 12 36 36 RBZerves and Zijlstra (2002a)c 6 42 42 CR

RatsBetran et al. (1992)c 5 40 40 CREsteve et al. (1992)b 8 56 56 CRHarmuth-Hoene andSchwerdtfeger (1979)b 6 72 72 CRRubio et al. (1998)c 5 20 20 RBYounes et al. (1991) 1 10 10 CRYounes et al. (1996) 6 60 60 CR

aCR = completely randomized; RB = randomized block; IB = incomplete block; LS = Latin square; RLS =repeated Latin square; and RM = repeated measures.

bUsed only for N digestibility and metabolic fecal N predictions.cUsed only for N clearance rate predictions.

evaluate a model to predict fecal N excretion rate (g/d), rate of N intake (g/d), and N utilization efficiency(grams of N in milk and gain per gram of N intake).

Materials and Methods

Published research data (Table 1) on several animalspecies were examined for N measurements, specifi-cally rate of N intake (g/d), urinary N excretion rate (g/d), fecal N excretion rate (g/d), BUN concentration (mg/

100 mL), and BW (kg). The species included beef anddairy cattle, goats, sheep, horses, pigs, and rats. Studiesthat included total urinary N excretion rate, BUN con-centration, and BW or rate of N intake, fecal N excretionrate, and BW were included. Plasma urea N concentra-tion was assumed to be equivalent to BUN concentra-tion because urea diffuses freely into and out of bloodcells. In some cases, fecal N excretion rate (g/d) wascalculated from the reported apparent digestibility ofprotein (g/g) and intake N (g/d). The means (n = 251)

Predicting nitrogen excretion from blood urea 881

Table 2. Observational data for studies used to develop the models

Range

Itema No.b Mean SD Low High

CattleBW, kg 71 380 185 100 685UN, g/d 38 90 67 21 264NI, g/d 62 203 188 49 729FN, g/d 62 70 6 9 269BUN, mg/100 mL 38 12 6 4 25

SheepBW, kg 69 40 14 21 68UN, g/d 40 7 3 1 16NI, g/d 65 12 6 3 37FN, g/d 65 3 2 0.3 10BUN, mg/100 mL 40 17 9 2 34

GoatsBW, kg 32 14 8 5 23UN, g/d 7 5 0.6 4 6NI, g/d 32 8 5 0.8 17FN, g/d 32 3 3 0.04 7BUN, mg/100 mL 7 31 5 25 38

HorsesBW, kg 3 400 0 400 400UN, g/d 3 49 29 18 74NI, g/d 3 74 40 28 98FN, g/d 3 27 1 27 29BUN, mg/100 mL 3 14 7 7 21

PigsBW, kg 37 70 55 24 161UN, g/d 37 15 7 6 36NI, g/d 15 31 10 12 50FN, g/d 15 3 0.6 2 4BUN, mg/100 mL 37 10 5 2 22

RatsBW, kg 31 0.15 0.08 0.06 0.4UN, g/d 17 0.10 0.08 0.01 0.2NI, g/d 21 0.37 0.19 0.10 0.8FN, g/d 21 0.06 0.04 0.02 0.2BUN, mg/100 mL 17 11 8 2 31

aUN = urinary N excretion; NI = N intake; FN = fecal N excretion; and BUN = blood urea N.bNumber of means.

from 41 research publications were analyzed. Table 2provides a summary of the data. Statistics were per-formed using JMP v. 5 (SAS Inst., Inc., Cary, NC),using the approach recommended by St-Pierre (2001)as described below.

Renal clearance rate is the rate at which blood iscleared completely of some substance by the kidney.The clearance rate of waste N (volume of blood clearedper unit time) was determined according to Swensonand Reece (1993) as:

CR = UN/BUN

where CR represents clearance rate (liters of bloodcleared completely of urea per day), UN represents uri-nary N excretion rate (g/d), and BUN represents BUNconcentration (g/L). Clearance rate was expressed as afraction of BW because the animals used in the datasetvaried in size. Kauffman and St-Pierre (2001) found

that adjusting the coefficient to predict UN from milkurea N concentration with BW eliminated breed effectsbetween Holsteins and Jerseys. We also considered ex-pressing urinary N excretion rate as a fraction of meta-bolic BW (BW0.75), but fits were not improved, so resultsare not shown. The following statistical model was usedwithin species:

UN/BW = mean + CR × BUN+ study (random) + error

where terms were as defined previously. Clearancerates were determined for all species as the regressioncoefficients for BUN. The intercept did not differ fromzero for any species (BUN concentration = 0 impliesurinary N excretion rate = 0) except cattle. Based onthe physiology, we did not expect a significant intercept.Thus, the intercept was forced through zero for eachspecies.

Kohn et al.882

Table 3. Model equations for prediction of nitrogen excretion

Prediction Equationa

Urinary N, g/d CR × BUN × BWN intake, g/d (UN + milk N + growth N + M × BW0.75)/TDFecal N, g/d NI − UN − milk N − growth NN utilization efficiency, % (milk N + growth N ) × 100/NI

aCR = N clearance rate, L of blood cleared of BUN�d−1�kg BW−1; BUN = blood urea N, g/L; M = metabolicfecal N, g�d−1�kg BW−0.75; TD = true protein digestibility, kg/kg DMI; NI = N intake, g/d; and UN = urinaryN excretion, g/d.

Fecal N was predicted using the Lucas test to calcu-late true digestibility (grams of N absorbed per gramof N intake) and metabolic N (g/d) from apparent digest-ibility with changing N intake (Van Soest, 1994). Appar-ent digestibility of N was defined by the equation:

AD = (NI − FN)/NI

where AD is apparent digestibility (g of N disappear-ance/g of intake N), NI is N intake (g/d), and FN is fecalN (g/d). True digestibility accounted for metabolic Nas follows:

TD = AD + M/NI

where TD is true digestibility (g of N digested/g of intakeN), and M is metabolic fecal N (g/d). This equation wasrearranged to yield:

AD × NI = TD × NI − M

where terms were as defined previously. To standardizethe equation for animals of vastly different sizes, bothsides were divided by metabolic BW in kilograms(BW0.75). The NRC (1996) calculates maintenance N,which includes endogenous N as a function of BW0.75

for beef cattle. In the present study, the parametersTD and M/BW0.75 were determined as the slope andintercept from regression analysis for each species us-ing the statistical model:

AD × NI/BW0.75 = mean + TD (NI/BW0.75)+ M/BW0.75 + study (random) + error

where terms were as defined for previous equations.The N retained per unit of ADG (g/kg) was also esti-

mated by regression of grams of N retained againstADG in kilograms, with the intercept forced through 0and study included as a random effect. This approachminimized the effect of animals growing at low ratesof gain, where the N retained divided by ADG ishighly unreliable.

There were limited data (109 means) that includedall variables of interest for the same treatment means.When different model parameters are derived from dif-ferent datasets and used together in a single predictionmodel, it is important that the relationships among

variables are similar in both datasets. To evaluatewhether data available for urea clearance rate esti-mates were equivalent to data available for digestibilityestimates, additional statistical analyses were per-formed using only studies including all variablesneeded for both estimates, and these estimates werecompared with those obtained by including all data (251means) to make either estimate. Only results from thelarger dataset are shown in tables, but other resultsare described when different.

Based on these estimates for urinary N excretion rate(g/d), true N digestibility (g/g), and N retention per unitof ADG (g/kg), the N intake (g/d), fecal N (g/d), and Nutilization efficiency (g of N in milk and meat/g of Nintake) were predicted as described in Table 3. Theseestimates were derived from the definitions of apparentand true digestibility and N utilization efficiency. TheN utilization efficiency also was determined directly asthe sum of N retained (g/d) in milk and gain dividedby N intake (g/d).

The accuracy of the model was evaluated by calculat-ing the residuals (predicted − observed) for each obser-vation (Bibby and Toutenburg, 1977). The root meansquare prediction error (RMSPE) is composed of meanbias and residual error and is defined as:

RMSPE =√(Σ[predicted − observed]2/number of samples)

For this study,

RMSPE = √[(mean bias)2 + (residual error)2]

The mean of residuals was taken as the mean bias.The residual error, after accounting for mean bias, wascomputed by multiplying the standard deviation of re-siduals by √[(n − 1)/n] to correct for the loss of 1 dfdue to fitting the mean when determining the standarddeviation. This correction is required for small samplesizes. The fraction of variance attributed to mean biaswas calculated as the mean bias squared divided by themean square prediction error, and the fraction attrib-uted to remaining dispersion was residual error2/meansquare prediction error. Mean square prediction errorattributed to the study effect was estimated as the par-tial sum of squares for the study divided by the sum of

Predicting nitrogen excretion from blood urea 883

all partial sums of squares and the error sum of squares.Most studies are designed to minimize variation withinstudy by using similar animals or the same animals,but this practice increases the possibility that the sam-ple does not represent a broad population, and there-fore, samples for different studies are likely to varymore from each other than from within studies.

Results and Discussion

Nitrogen Clearance Rate

As expected, BUN concentration was linearly relatedto urinary N excretion rate for all species (Figure 1).The intercept of regression was not different from zerofor any species except cattle (P < 0.05). The estimatedN clearance rates (L of blood cleared of N�d−1�kg BW−

1) for each species are shown in Table 4. Similar results(not shown) were obtained by directly calculating UN/(BW × BUN), but using the regression approach putsgreater weight on higher values. This weighting is de-sirable because the values in which the BUN is closeto 0 are less repeatable and skew the clearance rateestimates. For all species except rats, similar resultswere obtained when only using the limited dataset thatincluded studies that reported all variables of interest,including those for both digestibility and clearancerates. Clearance rate averaged 4.8 L�d−1�kg BW−1 whenusing the complete dataset with all available data onrats (Table 4). However, clearance rate was 12.6 L�d−

1�kg BW−1 for the limited dataset that included onlyseven means from the two studies with complete data(including digestibility data) for rats. The seven meansyielding steep slopes can be identified in Figure 1 for therat data. Greater than 80% of the variance for randomeffects was attributed to studies for cattle, pigs, sheep,and rats (only one study was used for each of horsesand goats), as shown in Table 4. This substantial studyeffect reflects differences in age and breed of animals,treatments, and analysis procedures. The results showthe consistent relationship between BUN concentrationand urinary N excretion rate, and the general coeffi-cients that could be used to estimate urinary N excre-tion rate from blood measurements. Although these re-lationships held across studies and species, the relativedifference within a study can be more consistently esti-mated than the differences across studies.

Estimates of urinary N (UN) could be calculated as:

UN = CR × BUN × BW

where CR represents the calculated clearance rates.For a 500-kg cow, with a BUN of 0.20 g/L, and clearancerate of 1.3 (Table 3), UN would be 130 g/d (calculatedas 1.3 × 0.2 × 500), but the confidence interval on theestimate would range from 89 to 171 g/d (calculated as:[1.3 ± 1.96 × SE] × 0.2 × 500) for an unknown study.However, the random effect of study contributes to mostof this uncertainty. For comparisons within a study,

the error would be substantially decreased. For exam-ple, for cattle, the SE within a study would be 0.05(calculated as √[SE2 × 0.08]), and the range with 95%

confidence would be from 123 to 137 g/d.To test differences in N clearance rates among spe-

cies, ANOVA was performed on the clearance rates de-termined as regression coefficients for each study whenregressing UN/BW on BUN. Data were transformed bytaking the natural logarithm because variances werenot equal across all species. The Levene test (JMP, v. 5)showed that unequal variances before transformationwere corrected by transformation. Clearance rates dif-fered among species according to the ANOVA conductedon the regression coefficients for each study (P < 0.001).Student’s t-tests revealed that the clearance rate forrats was higher than for all other species except pigs(P < 0.05), the clearance rates for pigs was higher (P <0.05) than for sheep and goats, and others did not differ(P > 0.10). Figure 1 shows that clearance rates for ratswere similar to those of pigs for two of three studiesusing rats. Dairy and beef cattle data were combinedbecause the estimates were not different among typesof cattle (regression slopes ± 2 × SE overlapped).

The lower clearance rates for herbivores as comparedwith the omnivorous species may have resulted fromdivergent needs for each species due to the diets theyhave adapted to consume. Herbivores benefit from re-cycling of N to the gut from the blood, and thereby cansurvive on low amounts of low-quality protein. Nitrogenrecycled to the rumen can be used to support fermenta-tion and synthesis of microbial protein, whereas N recy-cled to the hindgut supports fermentation processesthat provide energy, but do not supply protein to theanimal. Keeping urea in the blood would be advanta-geous for herbivores, making greater levels of N recycl-ing to the gut possible. The omnivores, on the otherhand, have a greater need to excrete N, which maybecome toxic to them. For all species, the blood ureaconcentrations were in the same range regardless ofthe quantity of protein consumed (Table 2). The poorerfit of data to the regression lines within rat studies mayhave resulted from the use of completely randomizeddesigns for rat studies rather than designs with treat-ments nested within animal as was the case for theother species.

Digestibility

True digestibility and metabolic fecal N are shownfor each species in Table 5 and the relevant regressionequations are shown in Figure 2. For all species exceptrats, similar results also were obtained when only usingstudies that reported variables for both digestibilityand clearance rates. When using all the data availablefor N digestibility in rats, the true digestibility was only79% (Table 5); however, true digestibility was 99% forthe seven means with complete data for rats. Thesedifferences for the limited vs. complete datasets for ratsmay be attributed to the small sample size in both cases.

Kohn et al.884

Figure 1. Urinary N (g�d−1�kg BW−1) vs. blood urea nitrogen (mg/100 mL) for each species. Data used for modeldevelopment are shown with different symbols and lines representing different studies.

With both datasets, metabolic N was not significantlydifferent from zero for rats. The lower magnitude ofmetabolic fecal N for pigs and rats may result from thelower amount of microbial residues in the feces and lesscomplex system of protein and fiber digestion. High true

digestibility of N and the large variance in digestibilityreflect the diverse types of protein sources used in thesestudies, as well as the diverse types of animals. Use ofmature forages for several studies with cattle resultedin decreased protein digestibility. The digestibility in

Predicting nitrogen excretion from blood urea 885

Table 4. Nitrogen clearance rate (L of blood cleared�d−1�kg BW−1) determined by re-gressiona

Random effectNitrogen clearance rate variance, %

Species No.b Estimate SE P <c Study Residual

Cattle 38 1.3 0.12 0.001 92 8Sheep 40 1.2 0.09 0.001 89 11Goats 7 0.7 0.08 0.001 — —Horses 3 0.9 0.06 0.005 — —Pigs 37 3.1 0.43 0.001 83 17Rats 17 4.8 1.40 0.004 74 26

aModel by species: UN/BW = mean + CR × BUN + study(random) + error, where clearance rate estimatewas calculated as the coefficient for BUN, intercept = 0; UN/BW = urinary N, g�d−1�kg BW−1; and BUN =blood urea N, g/L.

bNumber of study means.cClearance rate estimate (slope) ≠ 0.

these studies seems similar to that assumed by dietformulation recommendations (NRC, 1996, 2001).

Nitrogen Retention

For species in which data were available, the N reten-tion per kilogram BW was determined as shown in Ta-ble 6. These values were slightly higher than reportedusing changes in carcass N. In those studies, retainedN ranged from 14 to 16 g/kg for swine (Mahan andShields, 1998) and 20 to 22 g/kg for beef cattle (Ferrelland Jenkins, 1998). Retained N in balance studies, likethe ones used in the present analysis, is determined bydifference between N intake and output, so that losses,including volatilization of fecal or urinary N duringcollection, are considered part of the retention. Hence,these procedures may overestimate N retention com-pared with results from slaughter studies.

Nitrogen Utilization Predictions

The purpose of deriving these parameters was to de-termine the feasibility of using BUN and production

Table 5. True crude protein digestibility (%) and metabolic fecal N (g/kg BW0.75) calculatedfor different species by regressiona

Random effectTrue N digestibility Metabolic N variance, %

Species No.b % SE P <c g SE P <d Study Residual

Cattle 62 76 3.3 0.001 0.04 0.013 0.001 93 7Sheep 65 92 1.8 0.001 0.02 0.005 0.001 97 3Goats 32 96 2.5 0.001 0.03 0.017 0.08 98 2Horses 3 98 2.3 0.02 0.03 0.002 0.05 — —Pigs 15 96 2.0 0.001 0.005 0.002 0.07 61 39Rats 21 79 3.8 0.001 −0.009 0.007 NS 55 45

aModel by species: AD × NI/BW0.75 = mean + TD (NI/BW0.75) + M/BW0.75 + study(random) + error, whereAD = apparent N digestibility, g/100 g; NI = N intake, g/d; TD = true N digestibility, g/100 g; M = metabolicfecal N, g/kg BW0.75; and BW0.75 = metabolic BW, kg.

bNumber of study means.cTrue digestibility estimate (slope) ≠ 0.dMetabolic N estimate (intercept) ≠ 0.

data to estimate values of interest in modeling herd orflock N utilization, and whole-farm N losses. The meanexcretion and efficiency predictions for each species, theRMSPE, and the components of variation in the errorare shown in Table 7. In this case, the coefficients re-ported in the previous tables were used to make esti-mates for all cases where data were available to evalu-ate the estimates. The RMSPE varied from 7 to 112%of the mean estimate for urinary N excretion rate, butmost of this variation was attributed to study differ-ences. The substantial mean bias for cattle resultedfrom forcing the intercept of the line through zero forthe clearance rate estimate. There is no physiologicalreason to explain a significant intercept, and it was notsignificant for data from other species. Prediction errorsfor rate of N intake and fecal N excretion rate alsowere largely attributed to differences among studies.Substantial variation was attributed to mean bias incases where the residual analysis could only use a smallsubset of the data that were used in model development.In essence, this effect was also a bias imposed by studydifferences. Because most of the error in predictions

Kohn et al.886

Figure 2. Apparent N absorbed (g�d−1�kg BW−0.75) vs. N intake (g�d−1�kg BW−0.75) for each species. The slope representstrue digestibility (%), and the y-intercept is the negative value of metabolic fecal N (g/kg BW0.75). Data used for modeldevelopment are shown with different symbols and lines representing different studies.

was attributed to study differences, the proposed modelis likely to be a better tool to detect relative differenceswithin studies with similar animals and diets, than todetermine absolute estimates of N utilization. Much of

the variation in predictions was due to imposed treat-ments that varied greatly in N digestibility to a degreethat may exceed what would be observed on commer-cial farms.

Predicting nitrogen excretion from blood urea 887

Table 6. Nitrogen retained in grams per kilogram of average daily live weight gaindetermined by regression for different speciesa

Retained N per ADG Random effect variance, %

Species No.b g/kg SE P <c Study Residual

Cattle 25 26 5.3 0.001 93 7Sheep 23 29 7.8 0.001 81 19Pigs 37 17 2.1 0.001 90 10Rats 16 25 11.5 0.05 91 9

aModel by species: N balance, g/d = mean + NR × ADG + study(random) + error, where intercept = 0, andNR = nitrogen retained, g/kg of gain.

bNumber of study means.cN retained estimate (slope) ≠ 0.

This study demonstrates the feasibility of using BUNconcentration to predict N excretion rate for various do-mestic animals. With adequate estimates of N digestibil-ity, one can also determine rate of N intake, fecal Nexcretion rate, and N utilization efficiency (grams of N inmilk and gain per gram of N intake). A similar approachusing milk urea N concentration was used to character-ize herd N utilization and excretion on 450 commercialdairy farms (Jonker et al., 2002a). By estimating the

Table 7. Partitioning of root mean square prediction error (RMSPE) for each animal species

RMSPEa Variance, %Mean

Predictionb No.c prediction Total CV Mean bias Study bias Residual error

CattleUN, g/d 38 64 42 66 38 57 5NI, g/d 29 183 58 31 14 79 7FN, g/d 29 75 22 29 55 17 28NUE, % 28 23 2 9 86 4 10

GoatsUN, g/d 7 5.0 0.4 7 0 — 100NI, g/d 7 9.1 2.0 22 70 — 30FN, g/d 7 3.4 0.7 21 70 — 30NUE, % 7 15.6 8.6 55 57 — 43

SheepUN, g/d 40 6.3 2.1 33 2 86 12NI, g/d 36 10.0 4.0 41 25 65 10FN, g/d 36 3.3 1.7 51 5 88 7NUE, % 18 16 14 86 98 2 0

HorsesUN, g/d 3 50 6.5 13 3 — 97NI, g/d 3 76 22 29 1 — 99FN, g/d 3 26 2.1 8 54 — 46

PigsUN, g/d 37 17 8.9 52 6 90 4NI, g/d 15 40 16 39 32 62 5FN, g/d 15 4 1.8 42 81 7 12NUE, % 15 35 13 38 97 0 3

RatsUN, g/d 17 0.09 0.10 112 1 69 30NI, g/d 7 0.2 0.2 79 93 0 7FN, g/d 7 0.03 0.03 104 58 16 26NUE, % 7 60 10 17 87 2 11

aRoot mean square prediction error = √(Σ[observed − predicted]2/n); expressed as a total or as a percentageof the mean prediction (CV).

bUN = urinary N excretion; NI = N intake; FN = fecal N excretion; and NUE = N utilization efficiencyas defined by equations in Table 3.

cNumber of study means.

urinary N excretion rate that is expected when feedingrecommended quantities of protein, one can back-calcu-late target BUN concentrations for given productionrates. This approach has been taken for milk urea Nconcentration (Jonker et al., 1999), and farms that missthis target can be identified for assistance. An extensionprogram using milk urea N concentration in this waywas able to document changes in management resultingin decreased feed costs (Jonker et al., 2002b).

Kohn et al.888

Implications

Concentration of blood urea nitrogen was highly cor-related with urinary nitrogen excretion rate. Bloodsamples may enable prediction of nitrogen excretionrate, and subsequently fecal nitrogen excretion rate,rate of nitrogen intake, and nitrogen utilization effi-ciency (nitrogen in milk and gain per nitrogen intake).Target blood urea nitrogen concentration can be calcu-lated for different species for different production rates.Herbivorous species (cattle, sheep, goats, and horses)had lower nitrogen clearance rates than pigs and rats.Herbivores have adapted to conserving urea nitrogenfor recycling to the gut, which may explain the lowernitrogen clearance rates.

Literature Cited

Ahmad, M. R., V. G. Allen, J. P. Fontenot, and G. W. Hawkins. 1995.Effect of sulfur fertilization on chemical composition, ensilingcharacteristics, and utilization by lambs of sorghum silage. J.Anim. Sci. 73:1803–1810.

Alawa, J. P. 1991. Nitrogen excretion by sheep fed low-protein rough-age diets. Small Rumin. Res. 6:279–284.

Archibeque, S. L., J. C. Burns, and G. B. Huntington. 2001. Ureaflux in beef steers: Effects of forage species and nitrogen fertiliza-tion. J. Anim. Sci. 79:1937–1943.

Archibeque, S. L., J. C. Burns, and G. B. Huntington. 2002. Nitrogenmetabolism of beef steers fed endophyte-free tall fescue hay:Effects of ruminally protected methionine supplementation. J.Anim. Sci. 80:1344–1351.

Bailey, C. B. 1989. Rate and efficiency of gain, body composition,nitrogen, metabolism, and blood composition of growing Holsteinsteers given diets of roughage or concentrate. J. Anim. Sci.69:707–725.

Baker, L. D., J. D. Ferguson, and W. Chalupa. 1995. Responses inurea and true protein of milk to different protein feeding schemesfor dairy cows. J. Dairy Sci. 78:2424–2434.

Betran, M. A., E. Estornell, T. Barber, and J. Cabo. 1992. Nitrogenmetabolism in obesity induced by monosodium-L-glutamate inrats. Int. J. Obes. 16:555–564.

Bibby, J., and H. Toutenburg. 1977. Pages 16–19 in Prediction andImproved Estimation in Linear Models. John Wiley & Sons,London, U.K.

Bohnert, D. W., B. T. Larson, S. J. Lewis, C. J. Richards, K. C.Swanson, D. L. Harmon, and G. E. Mitchell, Jr. 1999. Net nutri-ent flux in visceral tissues of lambs fed diets differing in supple-mental nitrogen source. J. Anim. Sci. 77:2545–2553.

Cole, N. A. 1999. Nitrogen retention by lambs fed oscillating dietaryprotein. J. Anim. Sci. 77:215–222.

Dinn, N. E., J. A. Shelford, and L. J. Fisher. 1998. Use of the CornellNet Carbohydrate and Protein System and rumen-protected ly-sine and methionine to reduce nitrogen excretion from lactatingdairy cows. J. Dairy Sci. 81:229–237.

Dunn, J. M., and V. C. Speer. 1991. Nitrogen requirement of pregnantgilts. J. Anim. Sci. 69:2020–2025.

Estermann, B. L., F. Sutter, P. O. Schlegel, D. Erdin, H. R. Wettstein,and M. Kreuzer. 2002. Effect of calf age and dam breed on intake,energy expenditure, and excretion of nitrogen, phosphorus, andmethane of beef cows with calves. J. Anim. Sci. 80:1124–1134.

Esteve, M., I. Rafecas, X. Remesar, and M. Alemany. 1992. Nitrogenbalance of lean and obese Zucker rats subjected to a cafeteriadiet. Int. J. Obes. 16:237–244.

Ferrell, C. L., and T. G. Jenkins. 1998. Body composition and energyutilization by steers of diverse genotypes fed a high-concentratediet during the finishing period: I. Angus, Belgian Blue, Here-ford, and Piedmontese Sires. J. Anim. Sci. 76:637–646.

Ferrell, C. L., K. K. Kreikemeier, and H. C. Freetly. 1999. The effectof supplemental energy, nitrogen, and protein on feed intake,digestibility, and nitrogen flux across the gut and liver in sheepfed low-quality forage. J. Anim. Sci. 77:3353–3364.

Fisher, L. J., N. E. Dinn, J. A. Shelford, and J. W. Paul. 2000. Theeffect of concentration and form of dietary nitrogen on the utiliza-tion and excretion of nitrogen by lactating dairy cows. Can. J.Anim. Sci. 80:207–209.

Forster, L. A., J. P. Fontenot, H. D. Perry, J. G. Foster, and V. G. Allen.1991. Apparent digestibility and nutrient balance in lambs feddifferent levels of flatpea hay. J. Anim. Sci. 69:1719–1725.

Haig, P. A., T. Mutsvangwa, R. Spratt, and B. W. McBride. 2002.Effects of dietary protein solubility on nitrogen losses from lac-tating dairy cows and comparison with predictions from theCornell Net Carbohydrate and Protein System. J. Dairy Sci.85:1208–1217.

Harmuth-Hoene, A. E., and E. Schwerdtfeger. 1979. Effect of indigest-ible polysaccharides on protein digestibility and nitrogen reten-tion in growing rats. Nutr. Metab. 23:399–407.

Hettiarachchi, M., R. M. Dixon, and J. V. Nolan. 1999. Effect of intra-ruminal urea infusions and changing digestible organic matterintake on nitrogen kinetics in sheep fed rice straw. J. Agric. Sci.(Camb.) 133:109–121.

Hoffman, P. C., N. M. Esser, L. M. Bauman, S. L. Denzine, M. Eng-strom, and H. Chester-Jones. 2001. Effect of dietary protein ongrowth and nitrogen balance of Holstein heifers. J. Dairy Sci.84:843–847.

Huntington, G. B. 1984. Net absorption of glucose and nitrogenouscompounds by lactating Holstein cows. J. Dairy Sci. 67:1919–1927.

Jonker, J. S., R. A. Kohn, and R. A. Erdman. 1998. Using milk ureanitrogen to predict nitrogen excretion and utilization efficiencyin lactating dairy cows. J. Dairy Sci. 81:2681–2692.

Jonker, J. S., R. A. Kohn, and R. A. Erdman. 1999. Milk urea nitrogentarget concentrations for lactating dairy cows fed according toNational Research Council recommendations. J. Dairy Sci.82:1261–1273.

Jonker, J. S., R. A. Kohn, and J. High. 2002a. Dairy herd managementpractices that impact nitrogen utilization efficiency. J. Dairy Sci.85:1218–1226.

Jonker, J. S., R. A. Kohn, and J. High. 2002b. Use of milk ureanitrogen to improve diary cows diets. J. Dairy Sci. 85:939–946.

Kaiser, A. G., D. F. Osbourn, and P. England. 1983. Intake, digestionand nitrogen retention by calves given ryegrass silages: Influ-ence of formaldehyde treatment and supplementation withmaize starch or maize starch and urea. J. Agric. Sci. (Camb.)100:63–74.

Kauffman, A. J., and N. R. St-Pierre. 2001. The relationship of milkurea nitrogen to urine nitrogen excretion in Holstein and Jerseycows. J. Dairy Sci. 84:2284–2294.

Knaus, W. F., D. H. Beerman, T. F. Robinson, D. G. Fox, and K. D.Finnerty. 1998. Effects of dietary mixture of meat and bone meal,feather meal, blood meal, and fishmeal on nitrogen utilization infinishing Holstein steers. J. Anim. Sci. 76:1481–1487.

Koenig, J. M., N. W. Bradley, and J. A. Boling. 1982. Liver arginaseactivity and urea metabolism in beef heifers fed urea diets andabomasally infused with arginine and(or) ammonium acetate.J. Anim. Sci. 54:426–433.

Kohn, R. A., K. F. Kalscheur, and E. Russek-Cohen. 2002. Evaluationof models to estimate urinary nitrogen and expected milk ureanitrogen. J. Dairy Sci. 85:227–233.

Koster, H. H., R. C. Cochran, E. C. Titgemeyer, E. S. Vanzant, T. G.Nagaraja, K. K. Kreikemeier, and G. St. Jean. 1997. Effect ofincreasing proportion of supplemental nitrogen from urea onintake and utilization of low-quality, tallgrass-prairie forage bybeef steers. J. Anim. Sci. 75:1393–1399.

Lindberg, J. E. 1989. Nitrogen metabolism and urinary excretion ofpurines in goat kids. Brit. J. Nutr. 61:309–321.

Lykos, T., G. A. Varga, and D. Casper. 1997. Varying degradationrates of total nonstructural carbohydrates: Effects on ruminalfermentation, blood metabolites and milk production and compo-

Predicting nitrogen excretion from blood urea 889

sition in high-producing Holstein cows. J. Dairy Sci. 80:3341–3355.

Mahan, D. C., and R. G. Shields, Jr. 1998. Macro- and micro-mineralcomposition of pigs from birth to 145 kilograms of body weight.J. Anim. Sci. 76:506–512.

Martin, R. G., N. P. McMeniman, B. W. Norton, and K. F. Dowsett.1996. Utilization of endogenous and dietary urea in the largeintestine of the mature horse. Br. J. Nutr. 76:373–386.

Mosenthin, R., W. C. Sauer, and C. F. M. de Lange. 1992a. Tracerstudies of urea kinetics in growing pigs. I. The effect of intrave-nous infusion of urea on urea recycling and the site of ureasecretion into gastrointestinal tract. J. Anim. Sci. 70:3458–3466.

Mosenthin, R., W. C. Sauer, H. Henkel, F. Ahrens, and C. F. M. deLange. 1992b. Tracer studies of urea kinetics in growing pigs.II. The effect of starch infusion at the distal ileum on urearecycling and bacterial nitrogen excretion. J. Anim. Sci.70:3467–3472.

NRC. 1996. Nutrient Requirements of Beef Cattle. 7th ed. Natl. Acad.Press, Washington, DC.

NRC. 2001. Nutrient Requirements of Dairy Cattle. 7th ed. Natl.Acad. Press, Washington, DC.

NRC. 2003. Air Emissions from Animal Feeding Operations: CurrentKnowledge, Future Needs. Natl. Acad. Press, Washington, DC.

Rubio, L. A., G. Grant, P. Dewey, D. Brown, M. Annand, S. Bardocz,and A. Pustai. 1998. Nutritional utilization by rats of chickpea(Cicer arietinum) meal and its isolated globulin proteins is poorerthan that of defatted soybean or lactalbumin. J. Nutr.128:1042–1047.

Smith, S. I., and J. A. Boling. 1984. Lipid coating as a mode ofprotecting free methionine from ruminal degradation. J. Anim.Sci. 58:187–193.

St-Pierre, N. R. 2001. Integrating quantitative findings from multiplestudies using mixed model methodology. J. Dairy Sci. 84:741–755.

Swenson, M. J. and W. O. Reece. 1993. Water balance and excretion.Pages 573–604 in Duke’s Physiology of Domestic Animals. 11thed. Cornell Univ. Press, Ithaca, NY.

Sutoh, M., Y. Obara, and T. Yoneyama. 1993. The effects of feedingregimen and dietary sucrose supplementation on natural abun-

dance of 15N in some components of ruminal fluid and plasmaof sheep. J. Anim. Sci. 1993. 71:226–231.

Theurer, C. B., G. B. Huntington, J. T. Huber, R. S. Swingle, and J.A. Moore. 2002. Net absorption and utilization of nitrogenouscompounds across ruminal, intestinal, and hepatic tissues ofgrowing beef steers fed dry-rolled or steam-flaked sorghumgrain. J. Anim. Sci. 80:525–532.

van Aardenne, J. A., F. J. Dentener, C. G. M. Klijn Goldewijk, J.Lelieveld, and J. G. J. Olivier. 2001. A 1°–1° resolution datasetof historical anthropogenic trace gas emissions for the period1890–1990. Global Biogeochem. Cycles 15:909–920.

Van Barneveld, R. J., E. S. Batterham, and D. C. Skingle. 1995. Theeffect of heat on amino acids for growing pigs. 4. Nitrogen balanceand urine, serum, and plasma composition of growing pigs fedon raw or heat-treated field peas (Pisum sativum). Br. J. Nutr.73:259–273.

Van Soest, P. J. 1994. Nutritional Ecology of the Ruminant. 2nd ed.Cornell Univ. Press, Ithaca, NY.

Verstegen, M. W. A., D. Zwart, W. van der Hel, B. O. Brouwer, andT. Wensing. 1991. Effect of Tryppanosoma vivax infection onenergy and nitrogen metabolism of west African dwarf goats. J.Anim. Sci. 69:1667–1677.

Woodward, A., and J. D. Reed. 1997. Nitrogen metabolism of sheepand goats consuming Acacia brevispica and Sesbania sesban. J.Anim. Sci. 75:1130–1139.

Younes, H., C. Demigne, S. R. Behr, K. A. Garleb, and C. Remesy.1996. A blend of dietary fibers increases urea disposal in thelarge intestine and lowers urinary nitrogen excretion in rats feda low protein diet. J. Nutr. Biochem. 7:474–480.

Younes, H., K. A. Garleb, S. R. Behr, C. Demigne, and C. Remesy.1998. Dietary fiber stimulates the extra-renal route of nitrogenexcretion in partially nephrectomized rats. J. Nutr. Biochem.9:613–620.

Zerves, S., and R. T. Zijlstra. 2002a. Effects of dietary protein andfermentable fiber on nitrogen excretion patterns and plasmaurea in grower pigs. J. Anim. Sci. 80:3247–3256.

Zerves, S., and R. T. Zijlstra. 2002b. Effects of dietary protein andoathull fiber on nitrogen excretion patterns and postprandialplasma urea profiles in grower pigs. J. Anim. Sci. 80:3238–3246.