Evaluation of the feed intake models in the Nordic feed evaluation system NorFor

Krizsan, S.J.; Höjer, Annika; Huuskonen, Arto; Hetta, Mårten; Huhtanen, Pekka

doi:10.1080/09064702.2014.929168

Cited by 2 publications

(2 citation statements)

References 24 publications

(60 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Accurate predictions of voluntary feed intake are important in the formulation of optimal diets to support the productivity of cows (Allen, 1996). However, evaluations (Krizsan et al, 2014;Jensen et al, 2015a) of the NRC (2001), Zom et al (2012), Gruber et al (2004), TDMI (Huhtanen et al, 2011), and NorFor models showed that these models all had a fairly high prediction error (SD) of 1.5 to 3 kg of DMI/d, and a systematic overprediction at high DMI and underprediction at low DMI. These discrepancies could be due to variable substitution rates between forages and concentrates depending on the energy balance and intake (Faverdin et al, 1991(Faverdin et al, , 2011.…”

Section: Introductionmentioning

confidence: 99%

Description and evaluation of a net energy intake model as a function of dietary chewing index

Jensen

Markussen

Nielsen

et al. 2016

Journal of Dairy Science

View full text Add to dashboard Cite

Previously, a linear relationship has been found between net energy intake (NEI) and dietary chewing index (CI) of the diet for different types of cattle. Therefore, we propose to generalize and calibrate this relationship into a new model for direct prediction of NEI by dairy cows from CI values (CI; min/MJ of NE). Furthermore, we studied the forage-to-concentrate substitution rate in this new NEI model. To calibrate the model on a diverse set of situations, we built a database of mean intake from 14 production experiments with a total of 986 primi- and multiparous lactating dairy cows of different breeds fed 136 different diets ad libitum. The NEI were estimated by the Nordic feed evaluation system. The CI value of diets was estimated from the intake of concentrate, intake of forage neutral detergent fiber (NDF), particle length of forage, indigestible NDF/NDF, body weight, NDF/body weight, and the content of NE in DM. We show that the slope values in this regression are proportional to the squared intercepts, giving the nonlinear equation NEI=NEI-k×NEI×CI, where the parameter k represents the decline in NEI with the increasing CI of the diet and a was estimated to have a value of 2, implying a constant maximum daily chewing time. The intercept NEI in the regression of NEI on CI may be interpreted as metabolic net energy intake capacity of the cows fed without physical constraints on intake. Based on experimental data, the maximum chewing time was estimated as 1/(4 × k). The NEI values were parameterized as a linear function of metabolic body size, energy-corrected milk yield (kg/d), days in milk, and days in milk squared. Prediction accuracy was evaluated by mean square prediction error (MSPE) and its decomposition into central tendency, regression, and disturbance, across and within experiments on independent data from 19 experiments including 812 primi- and multiparous lactating dairy cows of different breeds fed 80 different diets ad libitum. The NEI model predicted NEI with an MSPE of 8% of observed, and across the 19 experiments the error central tendency, error regression, and error disturbance were 4.2, 40.6, and 84.9% of MSPE, respectively. The described intake model implies a variable forage-to-concentrate substitution rate as a nonlinear function of NEI, CI of forage, and supplementation of concentrate.

show abstract

Section: Introductionmentioning

confidence: 99%

Description and evaluation of a net energy intake model as a function of dietary chewing index

Jensen

Markussen

Nielsen

et al. 2016

Journal of Dairy Science

View full text Add to dashboard Cite

show abstract

“…The fit of the models were evaluated from the root-mean-square error (RMSE), from Akaike's information criterion (AIC), and from the significance of regression parameters. The smaller the RMSE and AIC are, the stronger the relationship of the residuals to the variables examined is (Krizsan et al, 2014).…”

Section: Mef=1mentioning

confidence: 99%

Ruminal pH predictions for beef cattle: Comparative evaluation of current models

Sarhan

Beauchemin

2015

Journal of Animal Science

View full text Add to dashboard Cite

This study evaluated 8 empirical models for their ability to accurately predict mean ruminal pH in beef cattle fed a wide range of diets. Models tested that use physically effective fiber (peNDF) as a dependent variable were Pitt et al. (1996, PIT), Mertens (1997, MER), Fox et al. (2004, FOX), Zebeli et al. (2006, ZB6), and Zebeli et al. (2008, ZB8), and those that use rumen VFA were Tamminga and Van Vuuren (1988, TAM), Lescoat and Sauvant (1995, LES), and Allen (1997, ALL). A data set of 65 published papers (231 treatment means) for beef cattle was assembled that included information on animal characteristics, diet composition, and ruminal fermentation and mean pH. Model evaluations were based on mean square prediction error (MSPE), concordance correlation coefficient (CCC), and regression analysis. The prediction potential of the models varied with low root MSPE (RMSPE) values of 4.94% and 5.37% for PIT and FOX, RMSPE values of 9.66% and 12.55% for ZB6 and MER, and intermediate RMSPE values of 5.66% to 6.26% for the other models. For PIT and FOX, with the lowest RMSPE, approximately 96% of MSPE was due to random error, whereas for ZB6 and MER, with the highest RMSPE, 15.85% and 23.42% of MSPE, respectively, was due to linear bias, and 37.19% and 60.12% of the error, respectively, was due to deviation of the regression slope from unity. The CCC was greatest for PIT (0.67) and FOX (0.62), followed by 0.60 for LES and TAM, 0.52 for ZB8, 0.39 for MER, 0.34 for ALL, and 0.22 for ZB6. Residuals plotted against model-predicted values showed linear bias (P < 0.001) for all models except PIT (P = 0.976) and FOX (P = 0.054) and mean bias (P < 0.001) except for FOX (P = 0.293), LES (P = 0.215), and TAM (P = 0.119). The study showed that the empirical models PIT and FOX, based on peNDF, and LES and TAM, based on VFA, are preferred over the others for prediction of mean ruminal pH in beef cattle fed a wide range of diets. Several animal (BW and intake), diet (forage and OM contents), and ruminal (ammonia and acetate concentrations) factors were (P < 0.001) related to the residuals for each model. We conclude that the accuracy of prediction of mean ruminal pH was relatively low for all extant models. Consideration of factors in addition to peNDF and total VFA, as well as the use of data from studies with continuous measurement of ruminal pH over 24 h or more, would be useful in the development of improved models for predicting ruminal pH in beef cattle.

show abstract

Evaluation of the feed intake models in the Nordic feed evaluation system NorFor

Cited by 2 publications

References 24 publications

Description and evaluation of a net energy intake model as a function of dietary chewing index

Description and evaluation of a net energy intake model as a function of dietary chewing index

Ruminal pH predictions for beef cattle: Comparative evaluation of current models

Contact Info

Product

Resources

About