Understanding the variation in the energy value of South African maize to improve diet formulation Peter Plumstead, Ph.D., PAS peter@chemunique.co.za Your Animals, Our Science
How do we grow this genetics to its optimum potential?
Understanding Variation in Commercial Operations = Value Potential Meat (kg/house) Sub-optimal production Sub-optimal diet digestibility Feed Ingredient variation Environment High microbial loads Sub-clinical disease Clinical disease Genetic Potential Commercial Performance Reduced performance + mortality O Focus on Maize / Corn Value Potential Feed, time, other inputs
Assessment of Variation of Corn in SA SAGL routinely analyse corn samples collected from Silo s Data from 27 regions over 4 harvest years from 2012-2017 2188 maize samples analysed (1162 White; 1026 Yellow)
Why care about the energy value of maize? Maize supplies approximately ~62-70% of the dietary AMEn and ~20% of dietary protein in SA broiler formulations To optimize feed cost and cost/kg gain, it is important to accurately predict digestibility of energy and amino acids from maize used in feed formulation AMEn contribution to feed Protein contribution to Feed MAIZE YELLOW MAIZE YELLOW SOY O/C 47% SOY O/C 47% 23,87% FULLFAT SOYA SUN O/C 38% VEG OIL OTHER 65,75% FULLFAT SOYA SUN O/C 38% VEG OIL OTHER *Broiler diet formulated to 1,05%dLys and 12,2 MJ ME broiler
Where does corn energy (AMEn) come from? Corn Morphology AMEn from corn Pericarp Fiber ~66% from starch ~8% from oil ~7% from protein ~19% from fibre & free sugars Endosperm Starch & Low Quality Protein Embryo/Germ Oil & High Quality Protein Reference: Noblet (INRA) 6
Maize Grain Composition and DE trends (1999-2010) >10,000 corn samples analyzed / year, >150,000 samples in dataset 5,0 %Oil 11,0 %Crude Protein 4,5 10,0 % 4,0 % 9,0 3,5 3,0 1999 y = -0,039x + 4,0082 R² = 0,4664 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 8,0 7,0 1999 y = -0,1083x + 9,7037 R² = 0,8035 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 75 74 % Starch 4050 DE, kcal/kg 73 4025 72 71 70 69 y = 0,1897x + 71,093 R² = 0,7402 4000 3975 y = -2,0924x + 4017,5 R² = 0,6572 68 3950 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2003 2004 2005 2006 2007 2008 2009 2010 All data on 100% DM basis Slide From Dan Jones et al., 2015 Pioneer Seeds
Selection for higher starch and lower protein? No, but there are strong genetic correlations between grain traits and yield Table 1. Genetic correlations between yield and quality traits Trait Name Corn Soy Oil 0.16 ± 0.037 0.13 ± 0.027 Protein 0.41 ± 0.032 0.24 ± 0.026 Starch 0.43 ± 0.037 DE 0.10 ± 0.040
Assessment of Variation of Corn in SA R 2 = 0,736 Crude Protein (% DM) Starch (% DM) N = 2188 Maize samples from 2012-2017 Harvest
Assessment of Variation of Corn in SA Crude Protein (% DM) Crude Fat (% DM) Harvest year Harvest year No seasonal downward trend in Crude Protein or Crude Fat Variation within Harvest Year was much larger than between years N = 2188 Maize samples from 2012-2017 Harvest
We can adjust AMEn for variation in proximate composition WPSA (1986) AMEn (kj/kg) = 15.51 x CP + 34.31 x C.Fat + 16.69 x starch + 13.01 x sugars CVB 2007: AME n (kj/kg) = 15.56xdig.CP + 38.83xdig.C.Fat + 17.32xdig.NfE Rostock: AME n (kj/kg) = 18.8xdig.CP +39.8xdig.C.Fat +17.3xdig.Starch +16.0xdig.Sugars +17.2xdig.NFR Brazilian tables (2011) ME = 18,04xdigCP + 38,88xdig.C.Fat + 17,32 x dignfed 3,5 Crude Fat (%) 3,0 2,5 2,0
We can adjust AMEn based on variation in proximate composition Rostock equation: AME n (kj/kg) = 18.8xdig.CP +39.8xdig.C.Fat +17.3xdig.Starch +16.0xdig.Sugar+17.2xdig.NFR Nutrient Corn A Corn B ME value Digestibility (%) ME contribution (KCAL/KG) Corn A Corn B DM 88 88 CP 7,5 8,5 4,50 81,6 275 312 Fat 3,5 4 9,51 86,4 288 329 Starch 68 67 4,13 92,17 2 592 2 553 Sugars 2 2 3,82 35 27 27 NFR 4,5 4 4,11 35 65 58 Ash 2,5 2,5 3 246 3 278 Difference 32 kcal/kg Energy contribution Amount of substrate Increment of ileal digestibility of substrate = x x (g substrate / (kcal/kg feed) kg feed) (g substrate / kg feed) Energy contribution of substrate (kcal / g substrate)
83.0 Prediction Equations to estimate AMEn AME n (kj/kg) = 18.8xdig.CP +39.8xdig.C.Fat +17.3xdig.Starch +16.0xdig.Sugar+17.2xdig.NFR Same chemical values. = same AMEn or NE content? 93.0 92.0 91.0 90.0 89.0 88.0 87.0 86.0 85.0 What about variation in digestibility of Protein, Fat, or starch? 84.0
Not all Maize Protein and Starch is Made The Same Starch granule Prolamin Zein Protein matrix
Energy Digestibility is Highly Variable for Hybrids 93.0 Pioneer Seeds used In-Vivo studies on the more than 300 corn samples fed to broilers to develop direct prediction of GE and GE digestibility (AMEn) using NIT (Near Infra Red Transmittance) 92.0 91.0 90.0 89.0 88.0 87.0 86.0 85.0 84.0 83.0 Energy digestibility of >300 characterized hybrids Slide from D.Jones,2015
How do we manage variation? Measured in the past Protein Starch Oil Crude Fibre Ash DE or NE calculated by Prediction equations* Measured now Direct measurement DM Protein Fat AMEn and DE via NIT Near Infra-Red Transmittance Prediction * Fixed digestibility of Protein, Fat, Starch and NFR
Validation of NIT Predicted vs. Actual AMEn Hruby, 2015
NIT models to predict nutritional traits Poultry AMEn vs Swine DE
NIT Calibrations allow us to directly predict the AMEn value of South African Maize 100 kcal 15 kcal N = 2188 Maize samples from 2012-2017 Harvest
absolute differences in the average AMEn between regions were small vs. variation within regions Geography matters, but Heat map of White Maize AMEn by region (Average of 2017 Harvest year)
Do White and Yellow Maize have the same nutritional value? White Maize Yellow Maize Prob > F N analyzed 1026 1162 Moisture 10,98 +/- 0,88 10,92 +/- 0,83 NS Protein (% DM) 9,07 +/- 0,73 9,28 +/- 0,79 <0,001 Starch (% DM) 69,57 +/- 0,87 69,07 +/- 1,04 <0,001 Fat (% DM) 4,28 +/- 0,26 4,04 +/- 0,33 <0,001 AMEn (kcal/kg DM)* 3941 +/- 20 3931 +/- 20 <0,001 DE (kcal/kg DM)* 4040 +/- 18,5 4031 +/- 18,4 <0,001 * AMEn and DE were determined using DuPont NIT analysis of whole maize kernels
Potential Variation of AMEn value of maize in SA within white or yellow hybrids is far greater than differences in the average values White Maize Yellow Maize
Gross Energy and AMEn of 60 SA Corn Samples 4200 Gross Energy =3991 kcal/kg AMEn =3508 kcal/kg Bomb Calorimetry NIT Gross Energy or AMEn (kcal/kg) 4000 3800 3600 3400 3200 483 kcal/kg undigested 3000 1 3 5 7 9 11131517192123252729313335373941434547495153555759
Energy Digestibility of 60 SA corn samples Total Tract Gross Energy Digestibility (%) 92 91 90 89 88 87 86 85 84 83 82 81 1% difference in GE digestibility = 39,9 kcal/kg AMEn 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 Determined by measuring Gross energy of maize (bomb calorimetry) & AMEn (NIT) 60
Corn variability >>10 day old broilers vs. Roosters AME kcal/kg 4300 4100 3900 3700 3500 3300 3100 2900 2700 2500 chick rooster 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Sample Collins and Moran, 1998
How accurate are prediction equations? Determined AMEn (kcal/kg) 3600 3500 3400 3300 3200 3100 3000 2900 Determined vs. Predicted AMEn of Corn Data from Leeson et al., 1993 vs. CVB 2007 Prediction equations y = 1,9869x - 3273,8 R² = 0,2017 2800 3190 3200 3210 3220 3230 3240 3250 3260 3270 3280 3290 3300 3310 3320 3330 3340 Predicted AMEn (kcal/kg, CVB 2007)
How accurate are prediction equations? Determined NIT AMEn vs. Predicted AMEn of Corn Data from NIT vs. CVB 2007 Prediction equations R 2 = 0,298 RMSE 19,99 kcal/kg
CVB AMEn prediction resulted in low GE digestibility Gross Energy Digestibility (%)* * Calculated as Predicted AMEn from NIT or CVB / Analyzed Gross Energy of Corn
Multiple regression model to predict AMEn Stepwise regression of analysed parameters of 2200 white and yellow maize samples Term Estimate Std Error Prob > F Intercept 7418,67 310,1 <0,001 Moisture -35,76 0,32 <0,001 Protein (% DM) 11,80 0,60 <0,001 Starch (% DM) 1,16 0,50 <0,021 Fat 34,97 0,80 <0,001 Colour 0,23 0,25 0,3495 Region -0,18 0,03 <0,001 Harvest Year -1,91 0,15 <0,001 Model R 2 = 0.92 with Root Mean Square Error of 9.2 kcal/kg. Colour not significant! Not including region and year reduced R 2 to 0.72
Comparison of NIT AMEn vs. Predicted AMEn* Measured NIT AMEn (kcal/kg) R 2 = 0,917 RMSE =9,29 Y = a + b*moisture+c*starch+d*fat+e*region+f*year Predicted AMEn (kcal/kg DM) using Model N = 2188 Maize samples from 2012-2017 Harvest
What is the opportunity? Gross energy digestibility of SA maize is not as constant as previously thought Differences in hybrids, as well as seasonal and regional growing conditions significantly affect AMEn value of corn. This is not currently factored into existing prediction equations of AMEn. CVB Predictions don t seem to account for all of the potential variation in AMEn NIT callibrations that directly determine GE and AMEn allow better estimates of changes in digestibility of incoming corn vs. formulated values Understanding variation in nutrient digestibility / AMEn of maize will also allow better predictions of enzyme response Improved performance and carcass yield by estimating the digestible energy more accurately Near Infra-Red Transmittance Prediction
The Gap between Genetic potential and actual In Conclusion performance is variable, and under-estimated. Sub-optimal production Sub-optimal diet digestibility Feed Ingredient variation Metabolic stress (ascites) High microbial loads More accurate determination of Maize AMEn and NE using NIT Production frontier Meat (kg/house) Sub-clinical disease Clinical disease Commercial Performance Value Potential Reduced performance + mortality O Feed, time, other inputs
Thank you / Merci beaucoup Research $ & Facilities Special Thanks Wiana Louw - SAGL Corinda Erasmus (Stats) QUESTIONS?