METABOLISM AND NUTRITION Using a computer-controlled simulated digestion system to predict the energetic value of corn for ducks1 F. Zhao,2 L. Zhang, B. M. Mi, H. F. Zhang, S. S. Hou, and Z. Y. Zhang The State Key Laboratory of Animal Nutrition, Institute of Animal Sciences, Chinese Academy of Agricultural Sciences, Beijing 100193 ABSTRACT Two experiments were conducted to develop a computer-controlled digestion system to simulate the digestion process of duck for predicting the concentration of ME and the metabolizability of gross energy (GE) in corn. In a calibration experiment, 30 corn-based calibration samples with a previously published ME concentration in 2008 were used to develop the prediction models for in vivo energetic values. The linear relationships were established between in vivo ME concentration and in vitro digestible energy (IVDE) concentration, and between in vivo metabolizability of GE (ME/GE) and in vitro digestibility of GE (IVDE/ GE), respectively. In a validation experiment, 6 sources of corn with previously published ME concentration in 2008 randomly selected from the primary corn-growing regions of China were used to validate the prediction models established in the calibration experiment. The results showed that in calibration samples, the IVDE concentration was positively correlated with the AME (r = 0.9419), AMEn (r = 0.9480), TME (r = 0.9403),
and TMEn concentration (r = 0.9473). Similarly, the IVDE/GE was positively correlated with the AME/GE (r = 0.95987), AMEn/GE (r = 0.9641), TME/GE (r = 0.9588), and TMEn/GE (r = 0.9637). The coefficient of determination greater than 0.88 and 0.91, and residual SD less than 45 kcal/kg of DM and 1.01% were observed in the prediction models for ME concentrations and ME/GE, respectively. Twenty-nine out of 30 calibration samples showed differences less than 100 kcal/ kg of DM and 2.4% between determined and predicted values for 4 ME (AME, AMEn, TME, and TMEn) and for 4 ME/GE (AME/GE, AMEn/GE, TME/GE, and TMEn/GE), respectively. Using prediction models developed from 30 calibration samples, 6 validation samples further showed differences less than 100 kcal/ kg of DM and 2% between determined and predicted values for ME and ME/GE, respectively. Therefore, the computer-controlled simulated digestion system can be used to predict the ME and ME/GE of corn for ducks with acceptable accuracy.
Key words: corn, duck, in vitro digestible energy, metabolizable energy, simulated digestion system 2014 Poultry Science 93:1410–1420 http://dx.doi.org/10.3382/ps.2013-03532
INTRODUCTION Corn is the major source of energy for more than 3 billion ducks raised annually in China. Previous studies have shown a high variable AMEn content in corn for poultry (ranges from 2,926 to 4,093 kcal/kg of DM, Leeson et al., 1993; Lessire et al., 2003). Therefore, accurate determination of the energetic value of corn is critical to confirm the ME content of a corn based diet for ducks. ©2014 Poultry Science Association Inc. Received July 31, 2013. Accepted January 21, 2014. 1 This project was financially supported by the Basic Scientific Research Program (2010jc-3-3), State Innovation Method Project (2009IM033100), Scientific Research New Star Plan of Beijing (2011098), and the Special Fund for the Innovation Team of the Chinese Academy of Agricultural Sciences (Beijing, P. R. China). 2 Corresponding author:
[email protected]
In general, the in vivo energy balance technique has been accepted as the most accurate method to determine the ME concentration or metabolizability of gross energy (GE) of feed ingredient (Hill and Anderson, 1958; Sibbald, 1976; Farrell, 1978; Bourdillon et al., 1990). However, it is time-consuming, costly, labor-intensive, and requires lots of birds, which makes the efficiency for ME determination unsatisfactory in practice (Yegani et al., 2013). To rapidly predict the ME concentration or metabolizability of GE of feed ingredient, nutritionists have developed several in vitro digestion procedures for roosters (Sakamoto et al., 1980; Clunies et al., 1984; Valdes and Leeson, 1992; Losada et al., 2009, 2010; Yegani et al., 2013). However, results show that the accuracy of prediction of ME is not always satisfactory. Two factors could contribute to the lack of correlation between in vivo ME and in vitro digestible energy (IVDE) or between in vivo metabolizability of GE (ME/GE) and
1410
PREDICTION OF ENERGETIC VALUE
in vitro digestibility of GE (IVDE/GE). One is that the simulated small intestinal fluid of poultry is made of porcine small intestinal fluid (Sakamoto et al., 1980; Clunies et al., 1984; Clunies and Leeson, 1984) or porcine pancreatin (Valdes and Leeson, 1992), and it is unclear whether its digestion ability is equal to that of in vivo small intestinal fluid of poultry. Another is that manual conduction of in vitro digestion procedures such as pH regulation, digestive enzyme injection, and separation of digested and undigested substance (Sakamoto et al., 1980; Clunies et al., 1984; Clunies and Leeson, 1984; Valdes and Leeson, 1992) could produce imprecise results. Considering these disadvantages, we propose a new in vitro digestion procedure for simulating the digestion of poultry. The novel features for this procedure are the preparation of simulated small intestinal fluid with amylase, trypsin, and chymotrypsin reagents to make equal activities of these digestive enzymes to that of in vivo small intestinal fluid, and an automatic equipment for simulating the process of gizzard-intestinal digestion. Using the chicken as a test animal, a good repeatability and additivity of IVDE, and acceptable accuracy of predicted ME of feed were reported by Mi (2012) and Ren (2012), which indicates this new in vitro digestion procedure is a promising method to predict the energetic value of feed for poultry. However, few studies have been conducted on in vitro digestion procedure for ducks. The objective of present study was to validate whether this new in vitro digestion procedure can be used to predict the ME concentration and metabolizability of GE of corn for ducks.
MATERIALS AND METHODS Experimental Design, Corn Samples, and In Vivo ME Assay Calibration Experiment. The objective of this experiment was to develop linear prediction models between ME and IVDE, and between ME/GE and IVDE/ GE using 30 corn calibration samples. Each sample was made by combining different percentages of corn, corn gluten meal, corn hulls, corn germ, and corn starch. The specifications of these 30 corn calibration samples have been described and reported by Zhao et al. (2008). The details for the procedure of the precision-fed model for ME determination, sample collection, chemical analysis, and calculation of 4 ME (AME, AMEn, TME, and TMEn) have been described by Zhao et al. (2008). Validation Experiment. The objective of this experiment was to test whether the accuracy of these linear models developed in the calibration experiment were acceptable for predicting the ME concentration or ME/GE of validation corn samples. Six sources of corn were randomly selected from the primary corn-growing regions of China and used to validate the prediction models established in the calibration experiment. The details for chemical composition and the procedure of
1411
precision-fed model for ME determination were also described and reported by Zhao et al. (2008).
IVDE Determination Simulated Digestive Fluid and Buffer Solution. The simulated gastric fluid and simulated small intestinal fluid were used for the in vitro digestion of gizzard and small intestine, respectively. The simulated gastric fluid was made of 1,550 U of pepsin/mL (Sigma 10070, Sigma-Aldrich Co., St. Louis, MO), which was in accordance with the activity of pepsin in in vivo gastric fluid of poultry reported by Sturkie (1976). To avoid the errors from different pepsin assay procedures, the method for determining the activity of pepsin (Sigma 10070) was the same as that described by Sturkie (1976). The concentrated simulated small intestinal fluid (CSSIF) was composed of 4,417 U of amylase/mL (Sigma A3306), 1,196 U of trypsin/mL (Amersco 0785, Amersco Inc., Solon, OH), and 429 U of chymotrypsin/mL (Amersco 0164). The activities of enzymes in the CSSIF were equal to 11 times of these in the small intestinal fluid of white Pekin duck (Zhao et al., 2007). During the in vitro small intestinal digestion, in each digestion chamber, 2 mL of CSSIF was diluted by 20 mL of neutral residual simulated gastric fluid. The activities of amylase, trypsin, and chymotrypsin in solution for in vitro small intestinal digestion were equal to these in in vivo small intestinal fluid. To avoid errors from different enzyme assay procedures, the methods for determination of the activity of amylase (Sigma A3306), trypsin (Amresco 0785), and chymotrypsin (Amresco 0164) were all in accordance with these described by Zhao et al. (2007). The gastric buffer solution was composed of 16.9 mmol/L NaCl, 9.6 mmol/L KCl, and 10 mmol/L HCl to match the ions concentration in gastric fluid of poultry (Sturkie, 1976) and adjusted to pH 2.0 at 42°C by the addition of 200 mmol/L HCl. The upper small intestinal buffer solution was prepared with 85.1 mmol/L NaCl, 14.8 mmol/L KCl, 170 mmol/L NaH2PO4, and 30 mmol/L Na2HPO4 to match the ion concentration in upper small intestinal fluid of duck (Zhao et al., 2012), and adjusted to pH 6.52 at 42°C by the addition of 200 mmol/L NaOH. The lower small intestinal buffer solution was composed of 85.1 mmol/L NaCl, 14.8 mmol/L KCl, 30 mmol/L NaH2PO4, and 170 mmol/L Na2HPO4 to match the ions concentration in lower small intestinal fluid of duck (Zhao et al., 2012), and adjusted to pH 7.91 at 42°C by the addition of 200 mmol/L NaOH. Preparation of Digestion Chamber. The digestion chamber (Figure 1) consisted of a transparent glass barrel, a dialysis tubing, a solution entry, a solution exit, a one-way valve, 2 silicone plugs, and 2 ground glass ports. Before preparation of the digestion chamber, the dialysis tubing (Membra-cel MD 44–14, Viskase Companies Inc., Darien, IL) was cut into pieces of 25 cm. Glycerin and sulfur compounds were removed by treating the tubing with a solution of 0.2% (wt/ vol) NaHCO3 and 1 mmol/L EDTA at 100°C for 10
1412
Zhao et al.
Figure 1. Schematic diagram of digestion chamber. 1 = solution entry. 2 = silicone plug. 3 = one-way value. 4 = entry of concentrated simulated small intestinal fluid. 5 = ground glass port. 6 = dialysis tubing. 7 = barrel. 8 = solution exit. 9 = feed sample. 10 = digestive fluid.
min. Subsequently, the tubing was washed with deionized water and treated with a solution of 1 mmol/L EDTA at 100°C for 10 min. Finally, the dialysis tubing was soaked in distilled water at 4°C before usage. The treated dialysis tubing was inserted into the glass barrel and tied on the outside of ground ports with rubber bands. After one ground port was sealed with a silicone plug, 2-g samples and 20 mL of simulated gastric fluid were added from the other ground port into the dialysis tubing. Then, this port was sealed with a silicone plug, which was inserted into a small silicone tube and connected to a one-way valve. Computer-Controlled Simulated Digestion System. The computer-controlled simulated digestion system (CCSDS) was designed to automatically process the in vitro gastrointestinal digestion procedure for duck. In each assay, there were 5 replicated determinations for one sample. This system was composed of 5 digestion chambers, 2 single-channel peristaltic pumps, 10 electronic values, a multiple-channel peristaltic pump, a warmed-air shaking incubator, a cooled-air incubator, a water bath incubator, a decompression tube, 4 reagent bottles, 3 buckets, a single-chip microcomputer, a computer, and a control software (Figure 2). Five replicated digestion chambers with solution entry downward were connected in a sequence that the exit of the one chamber connected to the entry of the closest next chamber. They were fixed on a platform and placed in a warmed-air shaking incubator. The solution entry of the first digestion chamber was connected to a single-channel peristaltic pump (11, Figure 2) for pumping buffer solution. The solution exit of the fifth digestion chamber was connected to a decompression tube for releasing the pressure of the digestion chambers. All digestion chambers were connected to a multiple-channel peristaltic pump (13, Figure 2) for the injection of CSSIF. One liter of gastric buffer, 1 L of upper small intestinal buffer, and 1 L of lower small intestinal buffer solutions were all placed in a water bath incubator. The wash bottle connected to a 25-L bucket by a single-channel peristaltic pump (12, Figure 2). After all tubes were connected as shown in Figure
Figure 2. Schematic diagram of computer-controlled simulated digestion system for ducks. 1–10: electronic value; 11–12: single-channel peristaltic pump; 13: multiple-channel peristaltic pump. CAI = cooled-air incubator; CSSIF = concentrated simulated small intestinal fluid; DC = digestion chamber; DT = decompression tube; GB = gastric buffer; LSIB = lower small intestinal buffer; OWV = one-way valve; SCM = single-chip microcomputer; ST = silicon tube; USIB = upper small intestinal buffer; WASI = warmed-air shaking incubator; WB = wash bottle; WBI = water bath incubator.
2, 2 mL of CSSIF were added into each small bottle stored at 4 to 8°C in a cooled-air incubator. During in vitro digestion, the digestion chambers and buffer solutions were maintained a 42°C stable temperature by the warmed-air incubator and the water bath incubator, respectively. A shaking speed of 180 rpm was used to mix the sample and digestive fluid. This system simulate the digestion and absorption of gizzard and intestine by change of the solution outside the dialysis tubing and the digestive fluid inside the dialysis tubing of digestion chamber. The parameters for in vitro digestion procedures, including the time for gastric digestion and intestinal digestion, the volume of CSSIF, change of solution outside the dialysis tubing, flow rate for solutions, temperature, shaking speed, and volume of water and frequency for clearance of hydrolyzed products, were automatically processed by the control software. In Vitro Gastric Digestion. For the gastric digestion, electronic valves 1 and 2 were opened. Next, the gastric buffer solution was pumped (120 mL/min, 11, Figure 2) into digestion chambers and circulated for 4 h. When the digestion process was finished, the electronic valves 1 and 2 were closed. Then, an emptying procedure was conducted. This procedure consisted of opening the electronic valves 9 and 10 and reverse-pumping (120 mL/min, 11, Figure 2) solution into the waste bottle. Subsequently, 3 replicated wash procedures were conducted. Each wash procedure consisted of pumping (420 mL/min, 12, Figure 2) 1,500 mL of deionized water into the wash bottle, opening the electronic valves 7 and 8, circularly pumping (120 mL/min, 11, Figure 2) water from wash bottle into the digestion chamber for 40 min, and an emptying procedure. In Vitro Intestinal Digestion. After the gastric digestion, a small intestinal digestion, including upper and lower small intestinal digestion, was processed. In the upper small intestinal digestion, the electronic valves
PREDICTION OF ENERGETIC VALUE
3 and 4 were opened and the upper small intestinal buffer solution was circularly pumped (120 mL/min, 11, Figure 2) into digestion chambers. After the buffer solution was circulated for 0.5 h, 2 mL of CSSIF was injected into each digestion chamber by the multiplechannel peristaltic pump (1.5 mL/min, 13, Figure 2), and the buffer solution continued to circulate for 7.5 h. When the upper small intestinal digestion was finished, the electronic valves 3 and 4 were closed and an emptying procedure was processed. Subsequently, the lower small intestinal digestion was conducted. The electronic valves 5 and 6 were opened and the lower small intestinal buffer solution had been circularly pumped (120 mL/min, 11, Figure 2) into digestion chambers for 7.5 h. When the lower small intestinal digestion was finished, the electronic valves 5 and 6 were closed. Lastly, an emptying procedure and 6 replicated wash procedures were processed. Undigested Residues. After the simulated digestion finished, the undigested residues from each digestion chamber were transferred to a preweighed vessel and dried overnight at 65°C, then, dried at 105°C for 5 h to constant weight. Next, the dry residues were transferred to a preweighted sintered glass crucible (G4) to extract fat by 45 mL of ethanol for 4 times. Lastly, the defatted residues along with the crucible were dried at 105°C for 5 h to constant weight.
Chemical Analysis The calibration and validation samples were ground finely in a laboratory mill fitted with a 0.3-mm mesh screen before analysis. The DM (method 934.01; AOAC, 1990) concentration of each sample was determined by oven drying at 105°C for 5 h. The GE concentrations of the calibration and validation samples and residues were analyzed by bomb calorimeter using a Parr 1281 automatic adiabatic calorimeter (Parr Instrument Co., Moline, IL); benzoic acid was used as the standard. All calibration and validation samples were analyzed in duplicate. Residues from each replicate were analyzed only once.
Calculation and Statistical Analysis The IVDE was calculated using the following formula: IVDE = [(sample DM weight × sample GE) − (defatted residue DM weight × defatted residue GE)]/ sample DM weight. Each sample was determined for IVDE in 5 replicates. The in vitro digestibility of GE and metabolizability of GE, expressed as the proportion IVDE/GE (%) and 4 ME/GE (AME/GE, AMEn/ GE, TME/GE, and TMEn/GE; %), respectively, were also calculated for each sample. Pearson correlation coefficients between ME and IVDE, and between ME/GE and IVDE/GE of calibration samples were calculated using CORR procedure of SAS 9.0 software (SAS Inst. Inc., Cary, NC). The REG procedure was used to develop linear regression models
1413
to predict the AME, AMEn, TME, and TMEn from IVDE, and predict the AME/GE, AMEn/GE, TME/ GE, and TMEn/GE from IVDE/GE, which was in accordance with previous regression analysis between in vitro and in vivo values for DE in swine (Noblet and Jaguelin-Peyraud, 2007; Regmi et al., 2008) and ME in poultry (Clunies et al., 1984; Valdes and Leeson, 1992; Boisen and Fernández, 1997; Yegani et al., 2013). The UNIVARIATE procedure was used to examine for the presence of outliers and for homogeneity of the residuals from prediction models. The influence of each observation on the prediction models was examined by using DFFITS. Observations with |DFFITS| > 2 were considered to be significant outliers. However, all data had a normal distribution, constant variance, and no outliers. The coefficient of determination (R2) and residual SD (RSD) were used as the indicators of quality of the prediction models. This was done by the procedure described by Kaps and Lamberson (2004). For testing the accuracy of the prediction models, the differences less than 100 kcal/kg of DM between determined and predicted AME, AMEn, TME, and TMEn (Valdes and Leeson, 1992), or the differences less than 2.62% between determined and predicted AME/GE, AMEn/GE, TME/GE, and TMEn/GE (Bourdillon et al., 1990) were considered to be acceptable.
RESULTS AND DISCUSSION Developing ME and Metabolizability of GE Prediction Models for Corn Good correlation between in vivo and in vitro determinations is very important to develop a new in vitro digestion procedure for predicting the concentration of ME or metabolizability of GE in poultry feed (Boisen and Eggum, 1991). In 30 calibration corn samples (Table 1), the mean IVDE concentration was 3,485 kcal/kg of DM with a range from 3,203 to 3,771 kcal/ kg of DM. Similarly, the mean AME and AMEn concentration were 3,480 and 3,415 kcal/kg of DM with ranges from 3,214 to 3,851, and 3,162 to 3,779 kcal/kg of DM, respectively. Simple correlation analysis showed that the IVDE concentration was highly positively correlated with the AME and AMEn concentration (r = 0.9419 and 0.9480, respectively; P < 0.01). The mean TME and TMEn concentration were 3,742 and 3,590 kcal/kg of DM with ranges from 3,477 to 4,111, and 3,338 to 3,952 kcal/kg of DM, respectively, which were greater than the mean IVDE. However, the IVDE concentration was also highly positively correlated with the TME and TMEn concentration (r = 0.9403 and 0.9473, respectively; P < 0.01). Similar correlation coefficients between IVDE and 4 ME concentrations (AME, AMEn, TME, and TMEn) were observed because of highly correlation between any 2 of these 4 ME (r = 0.99, Zhao et al., 2008). The strong correlation between IVDE and 4 ME concentrations is due to several factors. First, corn is primarily composed of starch
1414
Zhao et al.
(67.4%), CP (8.0%), and fat (4.4%; Cowieson, 2005). The ME concentration of corn is highly dependent on the concentrations of these digestible components (Carré, 2004; Cowieson, 2005). Second, in in vivo digestion, main enzymes activities including α-amlyase, proteases (pepsin, trypsin, chymotrypsin), and lipase are involved in the digestion of starch, protein, and fat (Whittow, 2000). In in vitro digestion of present study, α-amlyase and proteases (pepsin, trypsin, and chymotrypsin) are responsible for the digestion of starch and protein, respectively. The fat digestion was estimated by extraction of fat from undigested residues with ethanol. Previous results showed that the in vitro value determined by CCSDS was comparable with the in vivo value for the digestibility of fat in corn-soybean meal diet for ducks (86.4 and 91.9% for in vitro and in vivo digestibility of fat, respectively; Yin, 2011). In consideration of the low fat content and the small difference between in vivo and in vitro digestibility of fat in corn (Yin, 2011), the digestible fat content is comparable between the values determined by CCSDS and in vivo bioassay. This implies that the contributions of in vitro digestible fat to IVDE and in vivo digestible fat to ME are also comparable. So, the disappearance of organic matter in in vivo digestion may be similar to that in the in vitro procedure. Last, moderate variation in independent and dependent variables is helpful to produce a greater correlation coefficient (Carré, 1990; Regmi et al., 2008), because the correlation is difficult to perform if the assay precision for independent or dependent variables is not good to differentiate calibration samples. In the present study, the wide ranges of ME and IVDE concentration (637 kcal/kg of AME and 568 kcal/kg of IVDE) in calibration samples are satisfactory from a statistical standpoint.
The IVDE/GE varied from 69.37 to 84.59% with a mean of 76.24%. Correspondingly, the AME/GE and AMEn/GE varied from 69.24 to 86.38 and 68.12 to 85.77% with means of 76.13 and 74.71%, respectively. The TME/GE and TMEn/GE varied from 74.90 to 92.22 and 71.91 to 88.65% with means of 81.86 and 78.54%, respectively. Pearson correlation coefficients showed that the IVDE/GE was highly positively correlated with the AME/GE, AMEn/GE, TME/GE, and TMEn/GE (r = 0.9598, 0.9641, 0.9588, and 0.9637, respectively; P < 0.01), indicating that in vitro digestibility of GE is a good predictor of the in vivo metabolizability of energy. Similar result was also reported in swine by Regmi et al. (2008) who observed the in vitro digestibility of GE was highly correlated with apparent total tract digestibility of energy in 21 barley, although the manual in vitro digestion procedure (Huang et al., 2003) was used in their study. Current results showed that the mean values were comparable between IVDE and AME, and between IVDE/GE and AME/GE. In addition, high correlation coefficients were observed between IVDE and 4 ME determinations, and between IVDE/GE and ME/GE. Therefore, the CCSDS simulates in vivo digestion quite well and can be useful to predict the ME or metabolizability of GE of corn for ducks. Regression analysis showed the significant linear effects of IVDE concentration on ME concentration and IVDE/GE on ME/GE, respectively (Table 2). The coefficient of determination (R2) greater than 0.88 and greater than 0.91 were observed in the regression models for predicting 4 ME concentrations and for predicting 4 ME/GE, respectively, which infers that only less than 12 and 9% of the observed variation in the 4 ME concentrations and in the 4 ME/GE of calibration sam-
Table 1. Pearson correlation coefficients (r) between ME1 and in vitro digestible energy (IVDE),2 and between metabolizability of gross energy (GE)3 and in vitro digestibility of GE in calibration samples r Item IVDE, kcal/kg AME, kcal/kg AMEn, kcal/kg TME, kcal/kg TMEn, kcal/kg IVDE/GE, % AME/GE,5 % AMEn/GE,6 % TME/GE,7 % TMEn/GE,8 % 1ME
Range in determined values
IVDE
IVDE/GE4
Least
Mean
Greatest
0.9419** 0.9480** 0.9403** 0.9473**
0.9598** 0.9641** 0.9588** 0.9637**
3,203 3,214 3,162 3,477 3,338 69.37 69.24 68.12 74.90 71.91
3,485 3,480 3,415 3,742 3,590 76.24 76.13 74.71 81.86 78.54
3,771 3,851 3,779 4,111 3,952 84.59 86.38 85.77 92.22 88.65
(DM basis) was previously determined by Zhao et al. (2008). (DM basis). 3GE (DM basis). 4IVDE/GE = in vitro digestibility of GE. 5AME/GE = ratio of AME to GE. 6AME /GE = ratio of AME to GE. n n 7TME/GE = ratio of TME to GE. 8TME /GE = ratio of TME to GE. n n **Indicates r is significantly different from zero at P < 0.01. 2IVDE
1415
PREDICTION OF ENERGETIC VALUE
Table 2. Equations for predicting ME1 from in vitro digestible energy (IVDE)2 and predicting metabolizability of gross energy (GE)3 from in vitro digestibility of GE in calibration samples Item
Equation
AME, kcal/kg AMEn, kcal/kg TME, kcal/kg TMEn, kcal/kg AME/GE,6 % AMEn/GE,8 % TME/GE,9 % TMEn/GE,10 %
275 + 0.9196 × IVDE 241 + 0.9180 × IVDE 546 + 0.9170 × IVDE 423 + 0.9089 × IVDE 3.91 + 0.9473 × IVDE/GE7 3.13 + 0.9389 × IVDE/GE 8.68 + 0.9599 × IVDE/GE 6.33 + 0.9472 × IVDE/GE
SE of the intercept
SE of the slope
R2
RSD4
Adjusted RSD5
P-value
216 201 219 203 3.99 3.73 4.10 3.79
0.0620 0.0578 0.0627 0.0581 0.0523 0.0489 0.0537 0.0496
0.8871 0.8986 0.8841 0.8974 0.9213 0.9294 0.9193 0.9287
45 42 45 42 0.99 0.92 1.01 0.94
0.071 0.068 0.071 0.068 0.058 0.052 0.058 0.056
