Livestock Module (70_livestock)


The livestock module calculates how much and what kind of biomass products are required to satisfy the demand for livestock products (see $vm\_dem\_feed(i,kli,kbio)$). For this purpose, it provides for every time-step the feed baskets needed to produce one unit of livestock commodity. Feed demand is also needed within the pasture module (31_pasture) in order to derive required pasture areas. Interactions with residues (18_residues) and processing module (20_processing) are necessary to calculate the demand for substitutes of crop residues and conversion byproducts if generated byproducts are not sufficient to satisfy feed demand (see $vm\_dem\_res\_substitutes(i,kli,kcr)$ and $vm\_dem\_convby\_substitutes(i,kli,kcr)$). Demand for byproduct substitutes can increase the demand for primary products from cropland. Additionally, the module provides regional livestock production costs (see $vm\_cost\_prod(i,kli)$).



Name Description Unit A
$vm\_prod\_reg(i,kli)$ Regional livestock production mio. ton DM x
$fm\_prod\_cal(i,k)$ Calibration of production as defined in the FAO Food Balance Sheets (FBS) to production volume as derived from the FAO ProdSTAT database - x
$vm\_dem\_res\_substitutes(i,kli,kcr)$ Demand for substitutes of crop residues (if more crop residues are required than produced as by-product of primary products) mio. ton DM x
$vm\_dem\_convby\_substitutes(i,kli,kcr)$ Demand for substitutes of conversion byproducts (if more conversion byproducts are required than produced as by-product of primary products) mio. ton DM x
$vm\_cost\_prod(i,kli)$ Factor costs for livestock products mio. US$ x

The last columns of the table indicate the usage in the different realizations (numbered with capital letters)


Name Description Unit
$fm\_feeding\_convergence(t\_all)$ speed of the convergence process of regional livestock productions systems -
$vm\_dem\_feed(i,kli,kbio)$ Regional feed demand including byproducts mio. ton DM

Interface plot

Figure 0: Information exchange among modules.


(A) JAN15 (default)

This realization of the livestock module represents an update of the implementation as described in Weindl et al. 20101. An important part of the feed basket calculations is conducted outside of the core MAgPIE-GAMS code.

  • Feed basket calculations before entering the main MAgPIE-GAMS-code:

Before data on feed baskets enter the livestock module, they have to be calculated by a set of preprocessing routines. The supply of animal food commodities is divided into five livestock production activities (ruminant meat, pig meat, poultry meat, eggs and milk). The realization of each of the livestock production activities is based on the distinction of animal functions (reproducers, producers and replacing animals) and the specification of the energy content of each feed category in metabolizable energy (ME) for monogastrics, and in the case of ruminants in net energy (NE) for maintenance (NE.m), growth (NE.g) and lactation (NE.l). Specific feed energy requirements per commodity unit generated for each animal function and livestock production activity are taken from Wirsenius 20002, which include the minimum requirements for maintenance, growth, lactation, reproduction and other basic biological functions of the animals. In addition, they comprise a general allowance for basic activity and temperature effects and are complemented by extra energy expenditures for grazing. The specific feed energy requirements per unit output are consistent with available FAOSTAT data on animal productivity and reflect the conversion efficiency of feedstock to animal products. Together with information on regional livestock production (FAOSTAT, 2010), the specific feed energy requirements allow to calculate total regional feed energy requirements.

The next step in the calculation procedure consists in computing the corresponding total feed use in dry matter as well as the feed baskets for each animal function and livestock activity. Feed use data from the FAOSTAT Food Balance Sheets (FBS), which encompasses also conversion byproducts like meals, brans and molasses besides primary agricultural commodities like cereals, are used as an essential component to derive regional feed supply. Estimates of residues used as feed are based on harvest indices of food crops as well as recovery rates of different residues and assignment rates for feed use (based on Wirsenius, 2000; Krausmann, 2008; Smil) and are consistent with parameters and calculations of the Residues Module (18_residues).

The distribution of the described expanded data base on regional feed use of the entire livestock sector to single livestock activities and animal functions is obtained by an optimization model (Feed Distribution Model). The penalty function to be minimized includes two balancing feed categories for the feed energy balances (additional fodder crops and grazed biomass). Lower boundaries for the nutrient density of the resulting feed mixes from regional nutrient density recommendations are realized by introducing constraints. For developed countries, the nutrient density guidelines are based on NRC data (NRC 19963; NRC 19894), whereas for developing countries they are estimated by Wirsenius 20002. This additionally required fodder production is not added to the cropland based fodder production in MAgPIE (productin activity $foddr$), which is consistent with FAOSTAT, but declared as "scavenging" (similar terms are "roadside grazing" or "occasional feed") without cropland requirements within the model. Accordingly, the MAgPIE fodder production induced by this module realization meets the FAO fodder production. Moreover, the ruminant feed baskets in SAS are corrected based on values from Wirsenius 20002 to account for scavenging, i.e. a part of the feed demand for pasture biomass is declared as scavenging/roadside grazing, since a lot of authors report the use of occasional or "indefinite" feed resources for that region. Feed baskets which are the main result of the Feed Distribution Model are delivered as input for MAgPIE on DM basis [DM feed per DM generated product] without the detailed information on different types of energy requirements anymore. Figure 1 illustrates aggregated product-specific feed baskets, i.e. the feed requirements per product on dry matter basis and the share of different feedstock categories in the feed mix.

Figure 1: Regional feed requirements per product generated (DM /DM)and the share of different feedstock categories in the feed mix.

On the basis of these feed requirements, a regression analysis was conducted in order to reveal their relationship to the productivity of the respective livestock production sub-systems. Data on livestock productivity were taken from FAOSTAT containing a database on livestock production as well as numbers of animals in stock and producing animals. As measure for the productivity of the dairy-milk and the chicken-egg sub-system, the ratio of products generated per producing animals was calculated. For the remaining three sub-systems, productivity was calculated as ratio of products generated per animals in stock. Figure 2 displays the results of the regression analysis. By the means of the resulting equations, trend interpolations of historical productivity data or scenario assumptions on future developments of livestock productivity can be used to translate these projections into future time-dependent feed requirements. With a similar regression analysis, relations between productivity and the share of one selected group of feed categories could be observed, allowing also productivity and thus time-depending changes of feed mixes.

Figure 2: Feed conversion (defined as kg feed eaten per kg product generated on dry matter basis) for major animal food systems and MAgPIE world regions for 1995 and model estimation.

Figure 3: Comparison of data and model estimates with linear regression (solid line) and 1:1 line (dashed line).

Future scenarios of livestock productivity (according to the SSP story-lines) are derived based on informed guesses, taking into account past productivity improvements, GDP projections, cultural particularities, and the general scenario story-line. The scenarios were derived for each livestock product for the years 2030,2050,and 2100, and were interpolated in between. (The expert guesses were assembled in the file SVN\tools\livestock\Feed-Distribution-Model\output\feed baskets expert guesses.xlsx and processed with the R-script output_livestock_ssp.R in the same folder).

Figure 4: Scenarios (SSP 1,2,5 and 2,3,4) for the five magpie livestock commodities.

At the end of the preprocessing routines, the time-dependent parameter $i70\_feed\_bask(t,i,kli,kcr)$ is obtained in an aggregated form (no information on different types of energy requirements and animal functions) which is then used within the MAgPIE-GAMS-code.

  • Feed basket calculations within MAgPIE-GAMS-code (see equations.gms ):

Demand for different feed categories is essentially derived by multiplying the regional livestock production with the respective feed baskets. In the case of feed demand for primary crop products, the directly by the feed baskets induced crop demand is supplemented by an additional demand (see $vm\_dem\_res\_substitutes(i,kli,kcr)$ and $vm\_dem\_convby\_substitutes(i,kli,kcr)$ ) if the supply of residues and conversion byproducts is not sufficient:

Equation 1:

vm\_dem\_feed(i,kli,kcr) = vm\_prod\_reg(i,kli) \cdot ic70\_feed\_bask(i,kli,kcr) \cdot (1+f70\_feed\_losses\_share(kcr)) \\
+ vm\_dem\_res\_substitutes(i,kli,kcr)+ vm\_dem\_convby\_substitutes(i,kli,kcr)

Equation 2:

vm\_dem\_feed(i,kli,nonkcr) &= vm\_prod\_reg(i,kli) \cdot ic70\_feed\_bask(i,kli,nonkcr)

  • Livestock production costs calculations within MAgPIE-GAMS-code (see equations.gms ):

Factor requirement costs (labour, capital, ..., exclusive feed) of livestock production depend on the amount of production and the per-unit costs. The per-unit costs for non-ruminants are assumed to be fixed at 1500$ per ton (approximate global unweighted average over all production types) as there were no clear patterns observable. For ruminant products (milk and meet), we use a regression of per-unit factor costs from the GTAP database and the livestock productivity indicators. Here, factor costs clearly rise with intensification. Based on the regional production of livestock products, livestock production costs are calculated:

Equation 3:

vm\_cost\_prod(i,kli) \geq vm\_prod\_reg(i,kli)
\cdot (f70\_cost\_regr(kli,"cost_regr_a") + f70\_cost\_regr(kli,"cost_regr_b") \cdot ic70\_productivity(i,kli)

  • Speed of the convergence process of regional livestock productions systems (see input.gms ):

The interface $fm\_feeding\_convergence(t\_all)$ is read in within the livestock module and defines the speed of the convergence process of regional livestock productions systems. This transformation rate is relevant for several modules that are closely interconnected with livestock production, i.e. 53_methane and animal waste management systems 55_awms.

There are currently no known limitations of this realization.


Name Description Unit A
$i70\_feed\_bask(t,i,kli,kbio)$ Feed baskets per livestock product DM per DM x
$ic70\_feed\_bask(i,kli,kbio)$ Feed baskets per livestock product for the current time step DM per DM x
$i70\_feed\_bask\_sys(t,i,sys,kbio)$ Feed baskets per livestock product and animal function (with scavenging correction for ruminants in SAS) DM per DM x
$ic70\_productivity(i,kli)$ Livestock productivity in current time step (milk or meat yield per animal) DM per DM x
$f70\_feed\_losses\_share(kcr)$ Losses of different feed commodities as share of amount eaten (currently only fodder losses are taken into account) - x
$f70\_cost\_regr(kli,cost\_regr)$ Factor requirements for livestock production US\$04 per ton DM (A) and US\$(B) x
$sys$ Livestock production sub-systems (sys_beef,sys_dairy,sys_pig,sys_hen,sys_chicken) - x
$nonkcr$ Feed types excluding primary crop products - x
$cost\_regr$ Cost regression parameters (cost_regr_a,cost_regr_b) - x

The last columns of the table indicate the usage in the different realizations (numbered with capital letters)


Isabelle Weindl, Benjamin Bodirsky

See Also

core, 31_pasture, 18_residues, 20_processing, 16_demand, Overview


1 [Weindl et al. 2010] Weindl, I., Lotze-Campen, H., Popp, A., Bodirsky, B., and Rolinski, S.: Impact of livestock feeding technologies on global greenhouse
gas emissions. In: IATRC Public Trade Policy Research and Analysis Symposium, Climate Change in World Agriculture: Mitigation, Adaptation, Trade and Food Security, Stuttgart, Germany,

2 [Wirsenius 2000] Wirsenius, S: Human Use of Land and Organic Materials: Modelling the Turnover of Biomass in the Global Food System. Ph.D. Thesis, Chalmers University of Technology and Göteborg University, Göteborg, Sweden, 2000.

3 [NRC 1996] NRC, 1996. Nutrient requirements of beef cattle. National Academy Press, Washington.

4 [NRC 1989] NRC, 1989. Nutrient requirements of dairy cattle. National Academy Press, Washington.