Regression analysis to predict the impact of dietary neutral detergent fiber on carcass yield in swine

Abstract

Research has shown that carcass yield in swine is reduced when ingredients with high neutral detergent fiber (NDF) content. Carcass yield reduction from feeding high-fiber ingredients results from an increase in the weight of intestinal contents. NDF has been shown to result in the digestive contents to swell in the large intestine by absorbing water thus increasing the fecal volume in the large intestine. Considering the financial implications of changing carcass yield, the objective of this project was to develop a regression equation to estimate carcass yield from dietary NDF and strategies where high-NDF ingredients are taken out of the diet in the last dietary phases before slaughter (withdrawal period; WP). Data from 8 experiments (43 observations) originated from 6 journal articles and 1 technical memo were used to develop the regression equation. The WP of high NDF ingredients was either none or ranged from 5 to 63 d in the experiments. Treatment diets of each trial were reformulated to obtain dietary nutrient content using the NRC ingredient library (NRC, Nutrient requirements of swine. 11th ed, 2012). Composition of experimental diets was used to calculate dietary net energy, crude protein, crude fiber, NDF, and acid detergent fiber in the last two dietary phases. These dietary compositions along with the number of days of WP were used to develop regression equations. The model was determined using a step-wise selection procedure starting with guided forward selection through individual predictor variables, with a statistical significance at P < 0.05 used to determine inclusion of terms in the final model. The regression analysis showed that WP, NDF level in the dietary phase prior to the final phase (NDF1), NDF level in the last dietary phase before marketing (NDF2), and the interaction between NDF2 and WP were the most important variables in the dataset to predict carcass yield. The resulting regression equation was as follows: carcass yield, % = 0.03492 ± 0.02633 × WP (d) – 0.05092 ± 0.02862 × NDF1 (%) – 0.06897 ± 0.02931 × NDF2 (%) – 0.00289 ± 0.00216 × (NDF2 [%] × WP [d]) + 76.0769 ± 1.33730. In conclusion, high levels of NDF up to slaughter had a negative impact on carcass yield. Increasing the length of the WP improved carcass yield; however, the effect of WP was dependent on the level of NDF2. The equation herein provides a tool to estimate of the impact of dietary NDF on carcass yield.

INTRODUCTION

Multiple studies have investigated the impact of high-fiber ingredients on swine growth and carcass characteristics. Reports have indicated that up to 30% distillers dried grains with solubles (DDGS) can be fed without compromising growth performance (De Decker et al., 2005; Jacela et al., 2009; Stein and Shurson, 2009). However, research has also shown that carcass yield is reduced when DDGS or other high neutral detergent fiber (NDF) ingredients are fed to slaughter (Linneen et al., 2008). This effect on carcass yield occurs because high NDF diets increase the weight of the gastrointestinal tract mass and the volume and weight of intestinal contents at harvest (Turlington, 1984; Anugwa et al., 1989; Agyekum et al., 2012; Asmus et al., 2014). One successful strategy to ameliorate the negative effects on carcass yield is removing high-NDF ingredients from the diet before harvest. Research has reported that pigs transitioned from a high-NDF diet to a corn–soybean meal diet before harvest had similar carcass yield compared with pigs fed a corn–soybean meal diet throughout the entire finishing phase (Asmus et al., 2014; Graham et al., 2014; Coble et al., 2018). However, the optimal duration of the fiber reduction period is variable and primarily depends on the NDF content of the diets. Because of the financial implications of improving carcass yield, the objective of this project was to develop a prediction equation to accurately estimate the change in carcass yield from dietary NDF and NDF withdrawal strategies.

MATERIALS AND METHODS

Meta-Analysis

A literature review was conducted to compile studies that examined the effects of high insoluble fiber ingredients and withdrawal strategies on carcass yield. The literature search was conducted via the Kansas State University Libraries utilizing the CABI search engine. The following keywords were applied in the search: ([“neutral detergent fiber” OR “NDF” OR “fiber” OR “distillers” OR “DDGS”] AND [“growing-finishing” OR “grower-finisher” OR “finishing” OR “finisher”] AND [“pig” OR “swine”] AND [“yield” OR “dressing”]), and results were refined by: language (“English”), organism descriptors (“pigs”), and document type (“journal article” and “conference paper”). The search retrieved 174 publications from 1992 to 2018.

Selection Criteria for Inclusion and Exclusion

In order to be included in the final database, experiments had to meet the following criteria: 1) experimental treatments included a removal of high-NDF ingredients treatment and a corn–soybean meal diet as a control treatment; 2) provided information including duration of the feeding period, initial body weight, final body weight, average daily gain, average daily feed intake, gain:feed ratio, duration of withdrawal period (WP), and carcass yield; 3) provided percentage of dietary ingredients and NDF levels throughout the experiment; and 4) pigs used in experiments had ad libitum access to feed and water. Additional information regarding carcass characteristics or processing methods were not taken into consideration for the selection of studies due to the scarcity of information available in published literature. Considering these criteria, the final database resulted in 7 papers and 8 different studies with a total of 43 treatment observations (Table 1).

Table 1.

Summary of papers used in the regression analysis to predict carcass yield in finishing pigs

First author, year	Source1	NDF12, %	NDF23, %	WP4, d	Initial body weight, kg	Final body weight, kg	Carcass yield, %
Asmus, 2014	J	8.79–20.18	8.82–20.21	0–47	41.0	120.6–122.8	71.6–73.2
Coble, 2018 (experiment 1)	J	8.79–20.18	8.82–20.20	0–20	38.4	124.6–126.0	71.2–72.7
Coble, 2018 (experiment 2)	J	8.76–20.17	8.79–20.29	0–24	44.5	128.3–132.5	74.3–75.4
Gaines, 2007	J	8.72–15.25	8.75–15.28	0–42	66.1	126.9–128.5	75.9–77.1
Graham, 2014	J	8.79–20.18	8.83–20.20	0–24	55.8	122.9–126.8	72.8–74.2
Jacela, 2009	M	8.53–15.00	8.43–14.90	0–41	39.0	118.6–121.5	75.1–75.9
Nemechek, 2015	J	8.79–20.18	8.82–20.20	0–17	49.6	127.5–129.0	74.7–75.1
Xu, 2010	J	8.76–15.26	8.82–15.31	0–63	30.0	121.0–125.0	75.8 - 77.0

¹Source type: J = journal, M = technical memo.

²Range of neutral detergent fiber (NDF) concentration in dietary phase before the final phase.

³Range of NDF concentration in final dietary phase before marketing.

⁴Range of withdrawal period.

Diet Composition Calculations

Treatment diets of each trial were reformulated using a spreadsheet-based software program (Kansas State University Diet Formulation Program V.8.1) to obtain dietary nutrient content based on values obtained from NRC ingredient library (NRC, 2012). Composition of experimental diets was used to calculate dietary net energy (kcal/kg), crude protein (%), crude fiber (%), NDF (%), and acid detergent fiber (%) concentrations (as-fed basis) in the last two dietary phases and were recorded in the template for each dietary treatment. In addition, NDF WP in days as well as the SE of reported carcass yield were collected for each treatment in all experiments.

Statistical Analysis

The GLIMMIX procedure of SAS (SAS Institute, Inc., Cary, NC) was used to develop regression equations to predict carcass yield for finishing pigs. The method of restricted maximum likelihood was used in the model selection to evaluate significance of fixed effect terms. The model was determined using a step-wise selection procedure starting with guided forward selection through individual predictor variables, with a statistical significance at P < 0.05 used to determine inclusion of terms in the model. Throughout the selection process, studentized residuals plots were observed to determine if quadratic or interaction terms needed to be tested in the model. Residual plots were also used to investigate outliers. For development of the statistical model, study was included as a random effect according to procedures suggested by St-Pierre (2001). In addition, observations were weighted across studies according to the within study pooled SE for carcass yield. To determine the weighting, the SE of each mean was inverted and squared, and subsequently divided by the original SE to express the results on the same scale as the original data. Finally, the WEIGHT statement in SAS (SAS Institute, Inc.) provided a weight for each of these transformed values. Thus, observations with a smaller SE were weighted heavier, thus, having greater influence in the results than observations with larger SE.

RESULTS AND DISCUSSION

Prediction Equation for Carcass Yield Add the Equation

The regression analysis revealed that WP, NDF level in the dietary phase before the final phase (NDF1), NDF level in the WP before marketing (NDF2), and the interaction between NDF2 and WP were significant variables in the dataset to explain changes in carcass yield. The resulting regression equation was as follows: carcass yield, %= 0.03492 ± 0.02633 × WP (d) – 0.05092 ± 0.02862 × NDF1 (%) – 0.06897 ± 0.02931 × NDF2 (%) – 0.00289 ± 0.00216 × (NDF2 [%] × WP [d]) + 76.0769 ± 1.33730 (Table 2). To the best of our knowledge, this is the first published meta-regression that offers a model to quantify the impacts of dietary fiber and withdrawal strategy on carcass yield.

Table 2.

Regression equation to predict carcass yield from dietary neutral detergent fiber (NDF) and withdrawal strategies

Carcass yield, % = 0.03492 ± 0.02633 × WP (d) – 0.05092 ± 0.02862 × NDF1 (%) – 0.06897 ± 0.02931 × NDF2 (%) – 0.00289 ± 0.00216 × (NDF2 [%] × WP [d]) + 76.0769 ± 1.33730

Data from eight trials were used as a database for the statistical analysis to develop the model. NDF1 = NDF concentration in dietary phase before final dietary phase; NDF2 = NDF concentration in final dietary phase before marketing; WP = withdrawal period.

As expected, high NDF had a negative impact on carcass yield. According to Turlington (1984), the reduction in carcass yield from feeding high-fiber ingredients results from an increase in the weight of intestinal contents in the colon and cecum. The increase in gut fill is a result of the type of fiber in the ingredient. NDF has been shown to result in the digestive contents to swell in the large intestine by absorbing water thus increasing the fecal volume in the large intestine (Coble et al., 2018). Literatures generally agree that withdrawing high-fiber ingredients before marketing improve carcass yield (Asmus et al., 2014; Graham et al., 2014; Coble et al., 2018). Our model suggests that increasing the length of the WP improved carcass yield; however, the effect of WP was dependent on the level of NDF2, as indicated by the interaction term. The lower the NDF content is fed during the withdraw period, the greater withdrawal effect can be expected.

Application of Prediction Equations

An example using this equation is presented in Fig. 1. In the simulation, pigs were fed with moderate and high NDF1 diets (16% and 21% NDF; equivalent to 35% and 50% DDGS, respectively), and then transitioned to diets with either 9% or 13% NDF during the last dietary phase (NDF2) fed anywhere from 5 to 40 d before marketing. Predicted carcass yield when pigs are fed a corn–soybean meal (9% NDF) diet during both dietary phases was 75.0%. There is an estimated yield decrease of 0.84% and 1.44% when no withdrawal was implemented and 16% and 21% NDF diets were fed throughout the last two dietary phases, respectively.

Figure 1.

Predicted carcass yield of pigs fed varying neutral detergent fiber (NDF) levels (9%, 16%, or 21%) in the last dietary phase before marketing (NDF2) and for pigs transitioned from a 21% or 16% NDF diet (NDF1) to a 9% or 13% NDF diet (NDF2).

Partial carcass yield recovery is apparent when pigs are fed a 16% or 21% NDF diet and transitioned to a 9% NDF diet, depending on the length of the WP. However, the model predicted that yield is not continually improved when the diet in the last phase contains 13% NDF. In this situation, the entire benefit is found in the first 5 d of feeding the 13% NDF diet with no further improvement thereafter. The minimal WP where pigs were switched from a high-NDF diet to a different diet in the experiments used to develop the equation was 5 d. Consequently, the equation should not be used to predict withdrawal time less than 5 d.

In summary, fiber withdrawal strategies appear to recover carcass yield with the magnitude depending on the NDF level of the last two dietary phases as well as the fiber withdrawal length. This study offers a predication model to estimate the carcass yield of pigs fed various dietary NDF contents with given fiber WP.

Conflict of interest statement. None declared.