Building new feed unit systems for ruminants based on

absorbed nutrients and animal responses in France.

D.Sauvant1, P.Noziere2, 1AgroParistech-INRA UMR MoSAR,

2INRA URH-Theix

Modern story of ruminant Feeding Systems in France:

1978: « Alimentation des Ruminants » (R.Jarrige & 25 co-W)

1. A new Energy System for milk and meat (with NL and Switz.) 2. Launching 1st functional Protein Unit System (PDI) (adopted It, Sp, Pol, Switz…; adapted NL 94…) 3. Launching 1st Fill Unit System 4. All the requirements and feed values INRATION computer program

Further updatings 1. 1988: 1st updating + English version 2. 2002/2004 INRA-AFZ multispecies tables for concentrates & by-products (french, english, spanish, chineese) 3. 2007 last updating: requirements, responses,dynamic events, feed values, new computerized version …

REMAINING ISSUES…CHALLENGES… Feed unit systems (energy, protein…) are becoming obsolete despite marginal updatings Mechanistic models cannot (yet ?) replace the practical interest of the feed unit systems Knowledge is continuously accumulating and new challenges are emerging: impact on environment, quality of products, wellbeing…

“Systali” working group in France to update the french Feed unit systems (1st step: 2010-2013)

Head of the project: P.Noziere Assistants: J.L. Peyraud and D.Sauvant and 28 other research workers from 4 INRA lab (Theix, Rennes, Paris, Guadeloupe)

Targets of systali ? - More precise prediction of NE, PDI, AADI… - Prediction of the flows of nutrients: VFA + Gaz + Glucose + Fatty acids + Ess.Am.Ac… - Prediction of the animal responses to these flows - Enlarge the fields of applications (warm countries…) and the feed data bases 1st necessity to improve the accuracy of the current basic unit systems

SYSTALI PROJECT: PLAN 1. Organization of the project 2. Methods of approach 3. Digestive Part 4. Animal Part

FEED CHARACTERISTICS ANIMAL CHARACTERISTICS

FEED VALUE

DIET VALUE

REQUIREMENTS RESPONSES

INTEGRATED MODEL OF GUT

FEEDING PRACTICES

1. Global systali project

FEED CHARACTERISTICS ANIMAL CHARACTERISTICS

FEED VALUE

DIET VALUE

REQUIREMENTS RESPONSES

INTEGRATED MODEL OF GUT

FEEDING PRACTICES

1. Global systali project

NUTRIENT FLOWS

2. methods of approach: 2.1. Building large data bases from experimental results from INRA, or from the literature. 2.2. Study of the meta-designs (representativity, orthogonality…) and experiment encoding 2.3. Meta-analyses of data empirical models (St Pierre, 2001, Sauvant et al., 2008) Integration of the empirical models ?

543210

100

90

80

70

60

50

40

30

DMI%LW

OM

Dig

estib

ility

%Cloud of 2650 data of measurements of OMD% and DMI%LW in ruminants

EXAMPLE OF META-ANALYSIS: Influence of DMI%LW on OMD%, phase 1

1 point = 1 treatment = 1 group of animals

43210

90

80

70

60

50

40

DMI%LW

OM D

iges

tibilit

y %

400 data from experiments focused on influence of level of DMI%LW

EXAMPLE OF META-ANALYSIS: phase 2

4,03,53,02,52,01,51,00,5

90

80

70

60

50 MSI%PV

DMI%LW

Intra-experiment influence of DMI%LW on diet OM digestibility

OMD%

152 experiments focused on influence of level of DMI Databases « Bovidig » & « Rumener », D.Sauvant & al., 2011

EXAMPLE OF META-ANALYSIS: phase 3

43210

77,5

75,0

72,5

70,0

67,5

65,0

DMI%LW

OM

dig

estib

ility

%

Database RumenerDatabase Bovidig

Y=76.0 - 2.74 Xn=400, nexp=152, RSD=1.6

Experiments focused on DMI influence

Intra-experiment influence of DMI%LW on diet OM digestibility

Standard level =2 For tables

Table>observed

Table<observed

3.Digestive part 3.1. Digestive interactions 3.2. Substrate degradation - N & starch - Fermentable OM 3.3. Microbial protein production 3.4. Integration of the various equations

3.1.Digestive interactions (I)

VALUEdiet = Σi pi TABLEVALUEfeedi ± I

Modelling I ? -Impacted item = OMD% -Causes Predictors ?

Major causes of digestive interactions altering OM Digestibility ?

OMD%

Feeding Level

(DMI%LW)

Proportion of Concentrate

(PCO)

Rumen Protein Balance (RPB)

From OMD to ME ME = GE * Edig - ECH4/GE - EU/GE

DMI%LW, PCO, RPB, (CP)

OMD

Tables

3. Substrate degradation 3.1. N and starch

EDN & EDst = a (100/100+kl) + b (kd/kd+kp) a, b, kd in situ data kl, kpf, kpc = f(DMI%LW, PCO) by metaA Evaluation in situ vs in vivo ?

120100806040200

140

120

100

80

60

40

20

0

CP*(1-0.01*EDN%) %DM

in v

ivo

UCP

(g/k

gDM

)

Y=14.2 + 0.96 Xn=311, nexp=121, R2spec=0.72, RSD=8.8

Intra-experiment prediction of UCP from in sacco data

Bovidig database (Sauvant, 2012)

14.2

. Experiments focused on Protein influence

. Slope = 1 Digestive partition of protein

Endogenous N

g/kgDM

250200150100500

300

250

200

150

100

50

0

Starch * (1-0.01*EDSt%)

Star

ch fl

ow a

t duo

denu

m (

g/kg

DM

)

Y=-3.95 + 1.08 Xn=283, nexp=116, R2spec=0.88, ETR=22.1

Intra experiment prediction of by-pass starch from in sacco data

Database Bovidig (Sauvant, 2012)

. Experiments focused on Starch influences

. Slope = 1 Digestive partition of starch

3.2.2. Fermented OM

- Former approach index: FOM = DOM – RUCP -RUSt – EE – Vol - New approach « true » FOM FOM = corDOM – Dig.Int.Fract of duod. flows of [CP + St + FA + NDF] – Vol

3.2. Substrate degradation

3.3. Major causes of variation of microbial protein production in the rumen?

Proportion of concentrate (or Chemostat

Effect)

Fermented organic matter

Rumen Protein Balance

20 30 40 50 60 70 80 90

5

10

15

20

(g/kgMS)N microbien

Y=7.24+0.1173 Xn=707, nexp=270, Rreg=0.77, ETR=1.2

Y=0.25X

Intra-experiment influence of TDOMru on microbial N

Microbial N (g/kgDM)

Truly digested OM in the rumen (%DM)

« Bovidig » database (D.Sauvant, 2012)

High PCO Low RPB

3.4. Integration of the various equations ?

A necessity to insure consistency across more than 100

empirical equations extracted from various contexts…

Equations used to predict UFL, PDI…

Other equations (chewing, fill, pH nutrients flows…)

Partition of equations issued from meta-analyses

0,80,70,60,50,40,30,20,10,0

8

6

4

2

0

-2 Proportion of concentrate in diet DM

Influence of the proportion of concentrate on digestive interaction

OMD%table - OMD%invivo

Y=6.5/(1+(0.35/X)**3)n=72, ETR=1.22

Bovidig Database: experiments focused on effect of concentrate (Sauvant, 2012)

250200150100500

300

250

200

150

100

50

0

Starch * (1-0.01*EDSt%)

Star

ch fl

ow a

t duo

denu

m (

g/kg

DM

)

Y=-3.95 + 1.08 Xn=283, nexp=116, R2spec=0.88, ETR=22.1

Intra experiment prediction of by-pass starch from in sacco data

Database Bovidig (Sauvant, 2012)

Equations used to predict UFL, PDI…

Other equations (chewing, fill, pH nutrients flows…)

Partition of equations issued from meta-analyses

0,80,70,60,50,40,30,20,10,0

8

6

4

2

0

-2 Proportion of concentrate in diet DM

Influence of the proportion of concentrate on digestive interaction

OMD%table - OMD%invivo

Y=6.5/(1+(0.35/X)**3)n=72, ETR=1.22

Bovidig Database: experiments focused on effect of concentrate (Sauvant, 2012)

250200150100500

300

250

200

150

100

50

0

Starch * (1-0.01*EDSt%)

Star

ch fl

ow a

t duo

denu

m (

g/kg

DM

)

Y=-3.95 + 1.08 Xn=283, nexp=116, R2spec=0.88, ETR=22.1

Intra experiment prediction of by-pass starch from in sacco data

Database Bovidig (Sauvant, 2012)

Integration into a new

mechanistic model

of the rumen and gut

NI

dMO

dE

EM

EN

F1

ECH4

EU

F2

F3

ED

F4

F5

F6EBF7

MM

NDF

EE

F8

F9

F10

MOD_

F11

F12

kl_km_kf_kmf_utilisation_métaboliqueF13

MOF

PDIA

AGDint

NDFDint

PF

F14

F15

F16

F17

F18

AmiDint F19

PIMENmic

F20

PDIMEF22F23

NDFduo

F21

kf_kc_kl_transit

DTNF26

DTN6

a_b_c_N

F27

PIA

F28

F29

drF30

MAT

F32

PIMN

PDIMN

F34

PDIE

F35

PDINF36

AG

F37

AGduo

F38F39

F40F41

F44

a_b_c_Amidon

AMuoF42

Amidon

DTAM

F43

F45

F46

F47

BalPROru

F50

PCO

F51

F52

EUsurEBF53 F54

gCH4surMOD

F55

F56

F57

F58F59

q F60

F61

F62

F63

F64

F65

F66

F69

F70F71

F31

DTAM6

NDFnd

F25

F48

F24

F33

UFL_UFVF67

Diagramme « systali »: equations PDI et UF

NI

dMO

dE

EM

EN

F1

ECH4

EU

F2

F3

ED

F4

F5

F6EBF7

MM

NDF

EE

F8

F9

F10

MOD_

F11

F12

kl_km_kf_kmf_utilisation_métaboliqueF13

MOF

PDIA

AGDint

NDFDint

PF

F14

F15

F16

F17

F18

AmiDint F19

PIMENmic

F20

PDIMEF22F23

NDFduo

F21

kf_kc_kl_transit

DTNF26

DTN6

a_b_c_N

F27

PIA

F28

F29

drF30

MAT

F32

PIMN

PDIMN

F34

PDIE

F35

PDINF36

AG

F37

AGduo

F38F39

F40F41

F44

a_b_c_Amidon

AMuoF42

Amidon

DTAM

F43

F45

F46

F47

BalPROru

F50

PCO

F51

F52

EUsurEBF53 F54

gCH4surMOD

F55

F56

F57

F58F59

q F60

F61

F62

F63

F64

F65

F66

F69

F70F71

F31

DTAM6

NDFnd

F25

F48

F24

F33

UFL_UFVF67

Diagramme « systali »: equations PDI et UF Integration in

« Systool » to calculate

the new values of the

experimental rations

(cf Poster)

NI

dMO

dE

EM

EN

F1

ECH4

EU

F2

F3

ED

F4

F5

F6EBF7

MM

NDF

EE

F8

F9

F10

MOD_

F11

F12

kl_km_kf_kmf_utilisation_métaboliqueF13

MOF

PDIA

AGDint

NDFDint

PF

F14

F15

F16

F17

F18

AmiDint F19

PIMENmic

F20

PDIMEF22F23

NDFduo

F21

kf_kc_kl_transit

DTNF26

DTN6

a_b_c_N

F27

PIA

F28

F29

drF30

MAT

F32

PIMN

PDIMN

F34

PDIE

F35

PDINF36

AG

F37

AGduo

F38F39

F40F41

F44

a_b_c_Amidon

AMuoF42

Amidon

DTAM

F43

F45

F46

F47

BalPROru

F50

PCO

F51

F52

EUsurEBF53 F54

gCH4surMOD

F55

F56

F57

F58F59

q F60

F61

F62

F63

F64

F65

F66

F69

F70F71

F31

DTAM6

NDFnd

F25

F48

F24

F33

UFL_UFVF67

Diagramme « systali »: equations PDI et UF

Intégration in the

simulator « Sirar »

to assess the global consistency across practical

contexts

EVALUATION ON SHEEP OF THE INRA-SYSTALI MODEL OF DIGESTIVE INTERACTIONS

P. Nozière1, P. Chapoutot2 and D. Sauvant2 3 INRA-VetAgroSup, UMR 1213 Herbivores, Theix, France

1 INRA-AgroParisTech, UMR 791 Modélisation Systémique Appliquée aux Ruminants, Paris, France email: pierre.noziere@clermont.inra.fr

4.Animal part 4.1. Reconsideration of the basic requirements for maintenance and production - According to the new nutritive values of feed & diets - Integration of new aspects & concepts - For ruminants on poor quality diets, - Consistancy with the INRA-CIRAD-AFZ-FAO « Feedipedia » 4.2. Responses to feed units ? - N.E.(UF) and PDI and EAA - VFA, glucose, fatty acids Responses = feed efficiency, performances, digestive risks, milk and carcasses composition, faecal and urinary N, P, CH4 excretions, behavior…

4540353025201510

900

800

700

600

500

400

300

PDI e

ntre

tien

g/j

CNCPS

Syst a

Syst b

INRA 2007

Syst c

NRC

NL NorFor

Maintenance protein requirements: comparison across proposals for cattle

Milk yield kg/d

Mai

nten

ance

g/d

50403020100-10-20

90

80

70

60

50

40

30

PDIE 100 /k MS

0

6461,

EffPDI=61.6*exp(-0.013*[PDI-100]/MS)

n=136, nexp=53, ETR=3.27

Influence of MP concentration on MP efficiency into milk protein

[PDI-100] g/kgDMI

Efficiency

Bovidig-PDI database, D.Sauvant, 2013 Similar trends with dairy goats Influence of energy balance

CONCLUSIONS 1. First step of systali is in 2013 2. A priority was accorded to model digestive processes: 3. Meta-analyses allowed to stress several new interesting relationships allowing to be more realistic and accurate 4. The integrative model allowed to check constancy across relationships issued from various contexts 5. The approach was completed/evaluated by 2 tools to calculate diets and to simulate practical situations

151050-5

10,0

7,5

5,0

2,5

0,0

-2,5

-5,0 Correction dMO par NI+PCO+BalPROru

Y=-0.23 + 0.67 Xn=848, ETR=1.0

Y=-0.36 +0.68Xn=848, nexp=162, ETR=0.9

Base "Rumener" (DS & SGR, 2011)

Necessity to integrate ECH4%GE and EU%GE to assess digestive interactions

Edig:Corrected by FL, PCO and RPB

ME/GE:Corrected by FL, PCO, RPB and ECH4%GE + EU%EGE

ECH4 and EU induce a compensation, curvilinear ?

RELATIONSHIP RESPONSES-REQUIREMENTS: Can we estimate the both from a same data base ?

SECRETION IN MILK

ABSORBED NUTRIENTS

REQUIREMENTS (HOMEORHESIS Inter-Exp, Pull) Abs = f(Secr)

RESPONSES (HOMEOSTASE Intra-Exp, Push) Secr = f(Abs)

The negative influence of digestive interactions

1.Feeding level (FL) or level of production: + 1 unit of DMI%LW - 2 to 3 points of OMD% 2.Concentrate supply: +10%CO - 0.3 to - 0.6 pts OMD% 3.Lack of nitrogen in the rumen: - 10g CP/DM - 0.2 to - 0.4 point of OMD% - 10g Rumen protein balance - 0.5 point of OMD%

Quantitative consequences ?

43210-1

5

4

3

2

1

0

-1

-2Correction systali (UFL/J)

Corr

ectio

n fin

land

aise

(eq

UFL/

J)

No significant differences between both proposals Calculated on the « Bovidig » database

Comparison of proposals from Finland and France to take into account digestive interactions caused by feeding level, concentrate supply and

dietary protein content, interpretation in milk feed unit/d (D.Sauvant, unpub)