Advanced search
1 file | 1.47 MB Add to list

Nutrient digestibility and predicting the energy content of pig feeds

(2022)
Author
Promoter
(UGent) and Johan De Boever
Organization
Abstract
With 11 million slaughters per year, pigs are the most important production animal in Belgium. Feed is the biggest cost factor in pig farming, especially the energy component. The aim of feed formulation and production is to meet the nutrient requirements of pigs under the constraint of the available feed ingredients (Chapter 1). Correct energy and nutrient evaluation is critical, especially given the increasing amounts of fiber-rich co-products from human food production that are being included in pig feed. Another difficulty is that feed digestibility and utilization depend on many factors related to either the animal or to its growing conditions (Chapter 1). The net energy (NE) reflects best the true available energy for pigs, but NE determination requires time-consuming and expensive in vivo digestibility experiments. Potential methods to estimate NE in feeds include the use of tabular values of ingredients, empirical models based on chemical parameters and/or the in vitro digestibility of organic matter and near infrared spectroscopy (NIRS). The following research objectives (Chapter 2) were defined: • Study the interactions between fat and fiber level on nutrient digestibility and energy utilization of pig feeds (Chapter 4) • Evaluate the usefulness and accuracy of NIRS calibrations based on spectra from pig feed and feces to predict their chemical composition as well as nutrient digestibility and NE (Chapter 5) • Compare a) the use of feed tables, b) empirical models based on chemical and in vitro analyses and c) NIRS calibrations based on feed and feces spectra for their accuracy in predicting feed quality and more particularly NE, their applicability in practice and their limits (Chapter 6) Data related to these objectives were gathered via three in vivo digestibility trials. In Chapter 3, the protocol and details of the various methods used are described and justified. Main conclusions are: the acid insoluble ash (AIA), that is naturally present in pig feeds, is reliable as marker to determine nutrient digestibility via spot sampling of feces; AIA and TiO2 are equally reliable as external markers; the NE of a feed increases with 0.0021 MJ per kg higher bodyweight; once-daily sampling and pooling feces from two or three animals per pen is sufficient to determine in vivo digestibility. The interactions between fiber-rich co-products and fat were investigated, as well as their possible impact on nutrient digestibility and NE (Chapter 4). For this experiment, a low fiber (LF) and a high fiber (HF) diet were formulated. From these two basal diets, 8 additional diets were derived by adding 20 or 40 g/kg of either pig fat or soy oil. Compared to LF diets, HF diets had a lower (P<0.05) standardized digestibility for all nutrients except for NDF. However, higher fiber levels in the diet did not affect the digestibility of the added fat. The standardized digestibility of fat in the HF diets with added fat was higher (P<0.05) than that of the non-fat supplemented basal diet, whereas no significant (P>0.05) differences in fat digestibility among the LF diets were observed. Increasing fat level appeared to negatively affect the fat-free organic matter digestibility, in particular NDF. On the other hand, the negative effect of fat addition on digestibility and NE was relatively small in comparison to the extra energy added by the fat. Therefore, no special consideration should be given when adding extra fat to a fiber-rich diet to meet pigs’ energy requirements. For prediction of the chemical composition of pig feed and feces, as well as nutrient digestibility and NE, the usefulness and accuracy of NIRS calibrations based on combined spectra were investigated (Chapter 5). Near infrared spectroscopy appeared to accurately predict (residual prediction deviation, RPD>3; R²>0.8) most organic components and moisture content of pig feed and freeze-dried feces. The NE was better estimated using feed spectra than feces spectra, with a standard error of cross validation (SECV) of 0.33 and 0.46 MJ/kg, respectively. Combining feed and feces spectra resulted in an overall better estimation of the digestibility and NE, especially with merging (SECV=0.26 MJ/kg) and subtracting (SECV=0.27 MJ/kg). Finally, this study highlighted the importance of the cross-validation method when the dataset contains several feces spectra (alone or combined) from the same feed. Indeed, cross-validation may lead to over-optimistic results when the group to be validated contains spectra that are not independent. The use of tabular values, empirical models and NIRS to estimate nutrient composition and NE were compared (Chapter 6). This evaluation was based on 62 compound feeds, and 28 among them had known ingredient composition. In addition to the accuracy of each approach, their applicability in practice and practical limitations were compared. Feed tables predicted most chemical components accurately (R²>0.90), with exception of crude ash, sugar, acid detergent lignin and moisture, the latter being strongly underestimated (-16 g/kg). The standard error of estimate (SEE) for NE with tabular values amounted to 0.29 MJ/kg. Using empirical models based solely on chemical parameters resulted in a SEE of 0.21 MJ/kg. Incorporating in vitro digestibility resulted in a decrease of SEE to 0.18 MJ/kg. The SEE of a NIRS calibration based on feed spectra alone resulted in a SEE of 0.33 MJ/kg, whereas the combination of feed and feces spectra decreased this error to 0.26 MJ/kg. The use of feed tables allows a cheap and fast estimation of the chemical composition and NE, but it is only suitable for feeds of known and common ingredient composition and is not so accurate. The use of empirical models allows to accurately estimate the NE of feeds even with unknown composition, but relies on quite expensive and time-consuming analyses. Finally, NIRS calibration based on feed and feces spectra allows for a fast estimation of NE, but is less precise than empirical models and requires a specific sample preparation. Finally, the general discussion, conclusions and recommendations are presented (Chapter 7), where the results are discussed in depth and put into perspective. Important attention points are the limitation of NIRS accuracy for evaluating compound feeds due to the complexity and heterogeneity of the matrix; the error (0.033 MJ/kg) in the determination of the in vivo NE reference value due to sampling, animal variation and laboratory analysis, which represents the maximum achievable accuracy for the predicting models; and the practical limitation and applicability of each method. Future research could consider a combination of estimation methods and it would be interesting to integrate the energy and protein (amino acid) balance.

Downloads

  • (...).pdf
    • full text (Published version)
    • |
    • UGent only (changes to open access on 2027-03-08)
    • |
    • PDF
    • |
    • 1.47 MB

Citation

Please use this url to cite or link to this publication:

MLA
Paternostre, Louis. Nutrient Digestibility and Predicting the Energy Content of Pig Feeds. Ghent University. Faculty of Veterinary Medicine, 2022.
APA
Paternostre, L. (2022). Nutrient digestibility and predicting the energy content of pig feeds. Ghent University. Faculty of Veterinary Medicine, Merelbeke, Belgium.
Chicago author-date
Paternostre, Louis. 2022. “Nutrient Digestibility and Predicting the Energy Content of Pig Feeds.” Merelbeke, Belgium: Ghent University. Faculty of Veterinary Medicine.
Chicago author-date (all authors)
Paternostre, Louis. 2022. “Nutrient Digestibility and Predicting the Energy Content of Pig Feeds.” Merelbeke, Belgium: Ghent University. Faculty of Veterinary Medicine.
Vancouver
1.
Paternostre L. Nutrient digestibility and predicting the energy content of pig feeds. [Merelbeke, Belgium]: Ghent University. Faculty of Veterinary Medicine; 2022.
IEEE
[1]
L. Paternostre, “Nutrient digestibility and predicting the energy content of pig feeds,” Ghent University. Faculty of Veterinary Medicine, Merelbeke, Belgium, 2022.
@phdthesis{8744988,
  abstract     = {{With 11 million slaughters per year, pigs are the most important production animal in Belgium. Feed is
the biggest cost factor in pig farming, especially the energy component. The aim of feed formulation
and production is to meet the nutrient requirements of pigs under the constraint of the available feed
ingredients (Chapter 1). Correct energy and nutrient evaluation is critical, especially given the increasing
amounts of fiber-rich co-products from human food production that are being included in pig feed.
Another difficulty is that feed digestibility and utilization depend on many factors related to either the
animal or to its growing conditions (Chapter 1). The net energy (NE) reflects best the true available
energy for pigs, but NE determination requires time-consuming and expensive in vivo digestibility
experiments. Potential methods to estimate NE in feeds include the use of tabular values of ingredients,
empirical models based on chemical parameters and/or the in vitro digestibility of organic matter and
near infrared spectroscopy (NIRS).
The following research objectives (Chapter 2) were defined:
• Study the interactions between fat and fiber level on nutrient digestibility and energy utilization
of pig feeds (Chapter 4)
• Evaluate the usefulness and accuracy of NIRS calibrations based on spectra from pig feed and
feces to predict their chemical composition as well as nutrient digestibility and NE (Chapter 5)
• Compare a) the use of feed tables, b) empirical models based on chemical and in vitro analyses
and c) NIRS calibrations based on feed and feces spectra for their accuracy in predicting feed
quality and more particularly NE, their applicability in practice and their limits (Chapter 6)
Data related to these objectives were gathered via three in vivo digestibility trials. In Chapter 3, the
protocol and details of the various methods used are described and justified. Main conclusions are: the
acid insoluble ash (AIA), that is naturally present in pig feeds, is reliable as marker to determine nutrient
digestibility via spot sampling of feces; AIA and TiO2 are equally reliable as external markers; the NE
of a feed increases with 0.0021 MJ per kg higher bodyweight; once-daily sampling and pooling feces
from two or three animals per pen is sufficient to determine in vivo digestibility.
The interactions between fiber-rich co-products and fat were investigated, as well as their possible
impact on nutrient digestibility and NE (Chapter 4). For this experiment, a low fiber (LF) and a high
fiber (HF) diet were formulated. From these two basal diets, 8 additional diets were derived by adding
20 or 40 g/kg of either pig fat or soy oil. Compared to LF diets, HF diets had a lower (P<0.05)
standardized digestibility for all nutrients except for NDF. However, higher fiber levels in the diet did
not affect the digestibility of the added fat. The standardized digestibility of fat in the HF diets with
added fat was higher (P<0.05) than that of the non-fat supplemented basal diet, whereas no significant
(P>0.05) differences in fat digestibility among the LF diets were observed. Increasing fat level appeared
to negatively affect the fat-free organic matter digestibility, in particular NDF. On the other hand, the
negative effect of fat addition on digestibility and NE was relatively small in comparison to the extra
energy added by the fat. Therefore, no special consideration should be given when adding extra fat to a
fiber-rich diet to meet pigs’ energy requirements.
For prediction of the chemical composition of pig feed and feces, as well as nutrient digestibility and
NE, the usefulness and accuracy of NIRS calibrations based on combined spectra were investigated
(Chapter 5). Near infrared spectroscopy appeared to accurately predict (residual prediction deviation,
RPD>3; R²>0.8) most organic components and moisture content of pig feed and freeze-dried feces. The
NE was better estimated using feed spectra than feces spectra, with a standard error of cross validation
(SECV) of 0.33 and 0.46 MJ/kg, respectively. Combining feed and feces spectra resulted in an overall
better estimation of the digestibility and NE, especially with merging (SECV=0.26 MJ/kg) and
subtracting (SECV=0.27 MJ/kg). Finally, this study highlighted the importance of the cross-validation
method when the dataset contains several feces spectra (alone or combined) from the same feed. Indeed,
cross-validation may lead to over-optimistic results when the group to be validated contains spectra that
are not independent.
The use of tabular values, empirical models and NIRS to estimate nutrient composition and NE were
compared (Chapter 6). This evaluation was based on 62 compound feeds, and 28 among them had known
ingredient composition. In addition to the accuracy of each approach, their applicability in practice and
practical limitations were compared. Feed tables predicted most chemical components accurately
(R²>0.90), with exception of crude ash, sugar, acid detergent lignin and moisture, the latter being
strongly underestimated (-16 g/kg). The standard error of estimate (SEE) for NE with tabular values
amounted to 0.29 MJ/kg. Using empirical models based solely on chemical parameters resulted in a SEE
of 0.21 MJ/kg. Incorporating in vitro digestibility resulted in a decrease of SEE to 0.18 MJ/kg. The SEE
of a NIRS calibration based on feed spectra alone resulted in a SEE of 0.33 MJ/kg, whereas the
combination of feed and feces spectra decreased this error to 0.26 MJ/kg. The use of feed tables allows
a cheap and fast estimation of the chemical composition and NE, but it is only suitable for feeds of
known and common ingredient composition and is not so accurate. The use of empirical models allows
to accurately estimate the NE of feeds even with unknown composition, but relies on quite expensive
and time-consuming analyses. Finally, NIRS calibration based on feed and feces spectra allows for a
fast estimation of NE, but is less precise than empirical models and requires a specific sample
preparation.
Finally, the general discussion, conclusions and recommendations are presented (Chapter 7), where the
results are discussed in depth and put into perspective. Important attention points are the limitation of
NIRS accuracy for evaluating compound feeds due to the complexity and heterogeneity of the matrix;
the error (0.033 MJ/kg) in the determination of the in vivo NE reference value due to sampling, animal
variation and laboratory analysis, which represents the maximum achievable accuracy for the predicting
models; and the practical limitation and applicability of each method. Future research could consider a
combination of estimation methods and it would be interesting to integrate the energy and protein (amino
acid) balance.}},
  author       = {{Paternostre, Louis}},
  language     = {{eng}},
  pages        = {{X, 152}},
  publisher    = {{Ghent University. Faculty of Veterinary Medicine}},
  school       = {{Ghent University}},
  title        = {{Nutrient digestibility and predicting the energy content of pig feeds}},
  year         = {{2022}},
}