1
Supplementary information
2
Text S1. Calculation of livestock numbers in Europe
3
Major-ruminant livestock numbers from statistics (e.g., FAOstat, 2013) are provided as head
4
of dairy cows, beef cattle, sheep and goats. The energy and feed requirement can be different
5
for animals of different categories or even the same species with different liveweight and milk
6
productivity. The Livestock Unit (LU) is used to compare or aggregate numbers of animals of
7
different
8
http://ec.europa.eu/agriculture/envir/report/en/lex_en/report_en.htm)
9
requirements of the different animals. In this study, livestock species are converted to LU
10
based on the calculation of metabolisable energy (ME) requirement of each type of animal.
11
ME (also called net energy) is the amount of energy an animal needs for maintenance and for
12
activities such as growth, lactation and pregnancy. All calculations hereafter are given for
13
each country per year. The total number of ruminant livestock (Ntotal; in LU) is calculated as:
14
𝑁𝑡𝑜𝑡𝑎𝑙 =
15
where MELU is the ME requirement by one LU defined in this study (an adult dairy cow
16
producing 3000 kg milk annually, with live body weight of 600 kg; Eurostat, 2013); MEtotal is
17
the annual total ME requirement by all major ruminants, and is given by:
18
𝑀𝐸𝑡𝑜𝑡𝑎𝑙 = ∑(𝑀𝐸𝑖,𝑗 × 𝑁𝑖,𝑗 )
19
where Ni,j is number (in head from FAOstat) of animals in category i, type j; and MEi,j is the
20
ME requirement of animals in category i, type j; here, the category indicates whether the
21
animals are cattle, sheep or goats, and the type indicates the animals are producing milk,
22
slaughtered for meat, or animals neither producing milk nor slaughtered for meat in that year
23
(called residual animals in this study). FAOstat (2013) provides annual country-averaged
species
or
categories
(source:
based
𝑀𝐸𝑡𝑜𝑡𝑎𝑙
𝑀𝐸𝐿𝑈
on
the
feed
(Eq. S1)
(Eq. S2)
24
statistical data on numbers of major ruminant live animals stocks, milking animals (producing
25
milk)/slaughtered animals (producing meat), and correspondent milk yield for milking
26
animals (i.e., dairy cows, and sheep and goats for milk) and meat yield (carcass weight) for
27
meat animals (i.e., beef cattle, and sheep and goats for meat). When the numbers of total live
28
animal stocks (cattle, sheep or goats) are larger than the sum of milk and slaughtered animal
29
numbers, the residual animal numbers are calculated as the difference between total live
30
animal stocks and the sum of milk and slaughtered animal numbers.
31
The calculation of ME requirement for animals of each category and type follows the IPCC
32
Tier 2 algorithms (IPCC, 2006 Vol 4, Chapter 10, Eqns 10.3 to 10.13). ME includes the net
33
energy for maintenance (NEm), activity (NEa), growth (NEg), lactation (NEl), draft power
34
(NEwork; cattle only), wool production (NEwool; sheep only), and pregnancy (NEp):
35
𝑀𝐸 = 𝑁𝐸𝑚 + 𝑁𝐸𝑎 + 𝑁𝐸𝑔 + 𝑁𝐸𝑙 + 𝑁𝐸𝑤𝑜𝑟𝑘 + 𝑁𝐸𝑤𝑜𝑜𝑙 + 𝑁𝐸𝑝
36
where NEl is only for milk animals. It is noteworthy that NEg and NEwool were not calculated
37
in this study due to the limitation on data availability (i.e., average daily weight gain and wool
38
productivity).
39
The most critical animal performance data required for each animal category to estimate ME
40
are liveweight (for the calculation of NEm, and further for NEa, NEwork and NEp) and average
41
daily milk production (for the calculation of NEl). The liveweight was calculated as:
42
𝐿𝑖𝑣𝑒𝑤𝑒𝑖𝑔ℎ𝑡 = 𝐷𝑟𝑒𝑠𝑠𝑖𝑛𝑔 𝑝𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒
43
where dressing percentage is the conversion factor between carcass weight and live weight,
44
and the values are derived from Opio et al. (2013; see their Appendix B, Tables B20 and
45
B21).
𝐶𝑎𝑟𝑐𝑎𝑠𝑠 𝑤𝑒𝑖𝑔ℎ𝑡
(Eq. S3)
(Eq. S4)
46
In addition, several assumptions were made for the ME calculation: 1) the milk animals and
47
the residual animals have the same liveweight as the meat animals; 2) the residual animals
48
have no ME requirement for lactation (NEl) or pregnancy (NEp); 3) each animal spends an
49
hour per day working to search for food (used for calculation of NEwork,).
50
The ME requirement by one LU (MELU) is calculated to be ca. 85 MJ day-1. Then the gross
51
energy (GE) requirement is derived based on the summed ME requirements and the energy
52
availability characteristics of the feed(s) given by Eqn 10.16 of IPCC (2006). In this study,
53
low quality forage (such as mature grass) with typical digestible energy expressed as a
54
percentage of gross energy (DE%) of 55% was assumed. Finally, to convert from GE in
55
energy units to dry matter intake (DMI), we divided GE by the energy density of the feed
56
(with a default value of 18.45 MJ kg-1 of dry matter; IPCC, 2006). As a result, the daily dry
57
matter intake (DMI) of one LU was calculated as ca. 18 kg.
58
Figure S2 shows the total number of ruminant livestock (Ntotal; in LU) in each of the major
59
agricultural regions of Europe (Olesen & Bindi, 2002) and their evolution during the period
60
1961-2010.
61
62
63
Text S2. Components of the GHG budget
64
In this study, grassland based farm systems are defined. All grass products (harvested biomass
65
or grazing pasture) are used to feed livestock (grass-fed livestock numbers). All harvested
66
grass is assumed to be consumed on-farm and all manure returned to the grassland. Therefore,
67
both the ecosystem and off-site (at farm scale) GHGs fluxes are considered. The IPCC
68
guidelines for national greenhouse gas inventories (IPCC, 2006) provide methods of
69
calculating GHGs emissions from managed soil, from livestock and from manure
70
management, which is the reference in the calculation described below except for those
71
simulated by ORCHIDEE-GM.
72
73
The GHG fluxes of grassland ecosystem
74
The net GHG exchange (NGE; Soussana et al., 2007) represents GHG balance of a grassland
75
ecosystem. It can be calculated by adding CH4 and N2O emissions to NEE using the global
76
warming potential of each of these gases at 100-year time horizon (IPCC, 2013):
77
𝑁𝐺𝐸 = −(𝑁𝐸𝐸 + 𝐹𝐶𝐻4 −𝑒𝑐𝑜 × 𝐺𝑊𝑃𝐶𝐻4 + 𝐹𝑁2 𝑂−𝑒𝑐𝑜 × 𝐺𝑊𝑃𝑁2 𝑂 )
78
where GWPCH4 = 12.36, as 1 kg C-CH4 = 12.36 kg C-CO2; GWPN2O = 127.71, as 1 kg N-N2O
79
= 127.71 kg C-CO2; FCH4-eco is CH4 emission by grazing animals; FN2O-eco is direct and
80
indirect N2O emission from managed soil (based on IPCC, 2006). A positive value indicates
81
the grassland is a net sink of GHG.
82
CO2 fluxes in a grassland ecosystem
83
In managed grassland, the NEE (net ecosystem exchange of CO2) is calculated as:
84
𝑁𝐸𝐸 = 𝑅ℎ − 𝑁𝑃𝑃 + 𝑅𝑎𝑛𝑖𝑚𝑎𝑙
(Eq. S5)
(Eq. S6)
85
that can be simulated in ORCHIDEE-GM. A negative NEE value indicates the grassland is a
86
net sink of CO2 from the atmosphere.
87
The CH4 emission in field from enteric fermentation
88
FCH4-eco is CH4 emission from grazing livestock that can be simulated by ORCHIDEE-GM
89
(Vuichard et al., 2007a; Chang et al., 2013).
90
N2O emission from managed soil
91
The N2O emissions from managed soil are mainly due to the addition of nitrogen to grassland
92
as synthetic or organic fertilizer and by excreta from grazing livestock. It includes direct N2O
93
emissions from managed soil (FdirectN-N2Oeco), indirect N2O emissions from atmospheric
94
deposition of nitrogen volatilized from managed soils (FvolN-N2Oeco), and indirect N2O
95
emissions from nitrogen leaching/runoff from managed soils in regions where leaching/runoff
96
occurs (FleachN-N2Oeco):
97
𝐹𝑁2 𝑂𝑒𝑐𝑜 = 𝐹𝑑𝑖𝑟𝑒𝑐𝑡𝑁−𝑁2 𝑂𝑒𝑐𝑜 + 𝐹𝑣𝑜𝑙𝑁−𝑁2 𝑂𝑒𝑐𝑜 + 𝐹𝑙𝑒𝑎𝑐ℎ𝑁−𝑁2 𝑂𝑒𝑐𝑜
98
All the N2O emissions were calculated following the IPCC Tier 1 algorithms (IPCC, 2006
99
Vol 4, Chapter 11, Eqns 11.1, 11.9 and 11.10). Direct N2O emissions from managed soil were
(Eq. S7)
100
calculated as:
101
𝐹𝑑𝑖𝑟𝑒𝑐𝑡𝑁−𝑁2 𝑂𝑒𝑐𝑜 = (𝐹𝑆𝑁 + 𝐹𝑂𝑁 ) × 𝐸𝐹1 + 𝐹𝑃𝑅𝑃 × 𝐸𝐹2
102
Indirect N2O emissions from atmospheric deposition of nitrogen volatilized from managed
103
soils are calculated as:
104
𝐹𝑣𝑜𝑙𝑁−𝑁2 𝑂𝑒𝑐𝑜 = (𝐹𝑆𝑁 × 𝐹𝑟𝑎𝑐𝐺𝐴𝑆𝐹 + (𝐹𝑂𝑁 + 𝐹𝑃𝑅𝑃 ) × 𝐹𝑟𝑎𝑐𝐺𝐴𝑆𝑀 ) × 𝐸𝐹3
105
Indirect N2O emissions from nitrogen leaching/runoff from managed soils in regions where
106
leaching/runoff occurs are calculated as:
(Eq. S8)
(Eq. S9)
107
𝐹𝑙𝑒𝑎𝑐ℎ𝑁−𝑁2 𝑂𝑒𝑐𝑜 = (𝐹𝑆𝑁 + 𝐹𝑂𝑁 + 𝐹𝑃𝑅𝑃 ) × 𝐹𝑟𝑎𝑐𝐿𝐸𝐴𝐶𝐻−𝐻 × 𝐸𝐹4
108
where:
109
FSN = annual amount of synthetic fertilizer nitrogen applied to soils;
110
FON = annual amount of animal manure, compost, sewage sludge and other organic nitrogen
111
additions applied to soils;
112
FPRP = annual amount of nitrogen deposited as urine and dung by grazing animals on
113
grassland;
114
FracGASF = fraction of synthetic fertilizer nitrogen that volatilizes as NH3 and NOx;
115
FracGASM = fraction of applied organic nitrogen fertilizer material (FON) and of nitrogen
116
deposited as urine and dung by grazing animals (FPRP) that volatilizes as NH3 and NOx;
117
FracLEACH-H = fraction of all nitrogen added to/mineralized in managed soils in regions where
118
leaching/runoff occurs that is lost through leaching and runoff;
119
EF1 = 0.01; EF2 = 0.02 for cattle, and 0.01 for sheep; EF3 = 0.01; EF4 = 0.0075; FracGASF =
120
0.10; FracGASM = 0.20; FracLEACH-H = 0.30. This study used the default values of these
121
parameters and emission factors from guidelines (IPCC, 2006 Vol 4, Chapter 11, Table 11.1
122
and 11.3); FSN and FON come from the gridded nitrogen addition map (see main text for
123
detail); FPRP can be calculated in ORCHIDEE-GM. It is noteworthy that Flechard et al.
124
(2007) reported a lower emission factor (0.0075 instead of 0.01 in IPCC, 2006) of direct N2O
125
emission from managed soil.
126
127
GHG fluxes at grassland ecosystem and farm scale
(Eq. S10)
128
The off-site CO2, CH4 and N2O emissions from the digestion of harvested forage by livestock
129
and from manure management contribute to the net GHG balance (NGB) at ecosystem and
130
farm scale. NGB is then calculated as:
131
𝑁𝐺𝐵 = 𝑁𝐺𝐸 − 𝐹𝐶𝑂2−𝑓𝑎𝑟𝑚 − 𝐹𝐶𝐻4 −𝑓𝑎𝑟𝑚 × 𝐺𝑊𝑃𝐶𝐻4 − 𝐹𝑁2 𝑂−𝑓𝑎𝑟𝑚 × 𝐺𝑊𝑃𝑁2 𝑂 (Eq. S11)
132
where FCO2-farm is the proportion of harvest C that is respired by ruminants or emitted as labile
133
C in CO2 fluxes; FCH4-
134
fermentation or from manure management; FN2O- farm is direct and indirect N2O emission from
135
manure management (based on IPCC, 2006). All the off-site CO2, CH4 and N2O emissions are
136
calculated at country scale rather than at the grid-cell scale at which the NGE calculations are
137
made. A positive value of NGB indicates a net GHG sink at ecosystem and farm scale.
138
Off-site CO2 fluxes
139
We assume that C exported from grasslands (as harvested biomass, Fharvest) has six
140
destinations: 1) losses during transportation then decomposed as CO2 fluxes (Fharvest-loss); 2)
141
respired by livestock as CO2 fluxes (Ranimal-farm); 3) exported as animal products (milk and
142
meat; Fmilk/LW-farm); 4) CH4 emission from enteric fermentation or from manure management
143
(FCH4-farm); 5) manure produced on-farm by the grass-fed livestock and returned to the field as
144
organic C (Freturn); 6) labile C in CO2 fluxes during manure management (Fmanure-labile). Thus
145
the off-site CO2 fluxes (FCO2-barn) can be calculated as:
146
𝐹𝐶𝑂2 −𝑓𝑎𝑟𝑚 = 𝐹ℎ𝑎𝑟𝑣𝑒𝑠𝑡−𝑙𝑜𝑠𝑠 + 𝑅𝑎𝑛𝑖𝑚𝑎𝑙−𝑓𝑎𝑟𝑚 + 𝐹𝑚𝑎𝑛𝑢𝑟𝑒−𝑙𝑎𝑏𝑖𝑙𝑒
147
or
148
𝐹𝐶𝑂2 −𝑓𝑎𝑟𝑚 = 𝐹ℎ𝑎𝑟𝑣𝑒𝑠𝑡 − 𝐹𝐶𝐻4 −𝑓𝑎𝑟𝑚 − 𝐹𝑚𝑖𝑙𝑘/𝐿𝑊−𝑓𝑎𝑟𝑚 − 𝐹𝑟𝑒𝑡𝑢𝑟𝑛
farm
is the proportion of ingested C emitted as CH4 from enteric
(Eq. S12)
(Eq. S13)
149
where Fharvest is the C in harvested biomass simulated by ORCHIDEE-GM; FCH4-farm is the
150
off-site CH4 emissions that would be calculated later; Fmilk/LW-farm is the off-site C export
151
through milk and meat production. While the C export through animal products is not
152
determined and will be neglected in the calculation of grassland ecosystem C fluxes (see main
153
text), we consider this C export all happens off-site to avoid overestimating CO2 fluxes at
154
ecosystem and farm scale. Fmilk/LW-farm is given in Eqn A10, based on the amount of animal
155
products from statistics (FAOstat, 2013):
156
𝐹𝑚𝑖𝑙𝑘/𝐿𝑊−𝑓𝑎𝑟𝑚 = 𝑌𝑚𝑖𝑙𝑘 × 𝑅𝑔𝑟𝑎𝑠𝑠−𝑓𝑒𝑑 × (𝑓𝐹𝑎𝑡 × 𝐶𝐹𝑎𝑡 + 𝑓𝑃𝑅 × 𝐶𝑃𝑅 ) + 𝑌𝑚𝑒𝑎𝑡 × 𝑅𝑔𝑟𝑎𝑠𝑠−𝑓𝑒𝑑 ×
157
𝐶𝑐𝑎𝑟𝑐𝑎𝑠𝑠
158
where Ymilk and Ymeat are the milk and meat production from statistics (FAOstat, 2013)
159
respectively; fFat and fPR is the fat and protein content in milk, respectively (default values of
160
fFat = 4% and fPR = 3.5% are used following the IPCC, 2006 page 10.60); CFat, CPR and Ccarcass
161
is the C content in fat, protein and animal carcass respectively. Values of CFat =70% and CPR
162
= 46% and Ccarcass =5.1% are derived from Byrne et al., 2007 with original value from Wells
163
(2001) for CFat and CPR; and Hammond et al. (1990) for Ccarcass; and Rgrass-fed is the ratio of
164
manure from grass-fed animals to total manure application, given by:
165
𝑅𝑔𝑟𝑎𝑠𝑠−𝑓𝑒𝑑 = ∑(𝑁𝑖 × 𝑓𝑔𝑟𝑎𝑠𝑠𝑖 ) / ∑ 𝑁𝑖
166
where Ni is numbers (LU) of beef cattle, dairy cows, sheep, and goats calculated from
167
FAOstat for each country (Appendix A); fgrassi is the grass fractions in the diet of each type
168
of animal, which are taken from literature at the scale of Western Europe, Eastern Europe and
169
USSR (Bouwman et al., 2005). As a result, the European annual mean C exported through
170
grass-fed ruminant milk and meat products is ca. 3.2 Tg C yr-1, with much less C in meat
171
products (ca. 0.3 Tg C yr-1) than in milk products (ca. 2.9 Tg C yr-1; all values above are the
172
averages for 1961-2010 and over the EU28 plus Norway and Switzerland). If considering
(Eq. S14)
(Eq. S15)
173
only the C export through grass-fed ruminant milk and meat products during grazing, it will
174
be less than 1.3 g C m-2 yr-1 over European grassland ecosystem (with ca. 190 days in the year
175
that livestock can be grazing outside in fields simulated in ORCHIDEE-GM, and averaged
176
over European grassland).
177
Freturn is given in Eqn S16, based on the assumption that organic fertilizer input to grasslands
178
(Finput in Eqn 6) come from manure of livestock fed by grass and by non-grass forage, and
179
their fractions in Finput is the fractions of grass and non-grass forage in livestock diet
180
(Bouwman et al., 2005):
181
𝐹𝑟𝑒𝑡𝑢𝑟𝑛 = 𝐹𝑖𝑛𝑝𝑢𝑡 × 𝑅𝑔𝑟𝑎𝑠𝑠−𝑓𝑒𝑑
182
Off-site CH4 emissions
183
The off-site CH4 emissions come from enteric fermentation of livestock (FCH4-EFfarm) and from
184
manure management (FCH4-MMfarm), given by:
185
𝐹𝐶𝐻4 −𝑓𝑎𝑟𝑚 = 𝐹𝐶𝐻4 −𝐸𝐹𝑓𝑎𝑟𝑚 + 𝐹𝐶𝐻4 −𝑀𝑀𝑓𝑎𝑟𝑚
186
The CH4 emission from enteric fermentation at farm (FCH4-EFfarm) is calculated as:
187
𝐹𝐶𝐻4 −𝐸𝐹𝑓𝑎𝑟𝑚 = 𝐹𝐶𝐻4 −𝑒𝑐𝑜 × (1 − 𝑇𝑔𝑟𝑎𝑧𝑖𝑛𝑔 )⁄365
188
where Tgrazing is the number of days of livestock grazing outside in fields simulated in
189
ORCHIDEE-GM. The total CH4 emission from manure management (FCH4-MM) in each
190
country can be calculated with livestock numbers following the IPCC Tier 1 algorithms
191
(IPCC, 2006 Vol 4, Chapter 10, Eqn 10.22), but depending on livestock types as well as
192
temperature:
193
𝐹𝐶𝐻4 −𝑀𝑀 = 𝑁𝑖 × 𝐸𝐹5(𝑖)
(Eq. S16)
(Eq. S17)
(Eq. S18)
(Eq. S19)
194
where EF5(i) is the emission factor for CH4 emitted from manure management (kg CH4 head-1
195
yr-1) depending on livestock types as well as temperature (IPCC, 2006 Vol 4, Chapter 10,
196
Table 10.14); Ni is livestock numbers (head) of beef cattle, dairy cows, sheep, and goats
197
calculated from FAO statistics for each country. Then for each country, the actual CH4
198
emission from manure management that happens while animals are at-barn can be calculated
199
as:
200
𝐹𝐶𝐻4 −𝑀𝑀𝑓𝑎𝑟𝑚 = 𝐹𝐶𝐻4 −𝑀𝑀 × 𝐹𝐶𝐻4 −𝐸𝐹𝑓𝑎𝑟𝑚 ⁄(𝐹𝐶𝐻4 −𝐸𝐹𝑓𝑎𝑟𝑚 + 𝐹𝐶𝐻4 −𝑒𝑐𝑜 )(Eq. S20)
201
Usually, the CH4 emissions from manure management are only around 12% of the CH4 from
202
enteric fermentation.
203
Off-site N2O emissions from manure management
204
Following the IPCC Tier 1 algorithms (IPCC, 2006 Vol 4, Chapter 10), the N2O can be
205
directly emitted from manure management (FdirectN-N2Ofarm), indirectly emitted due to
206
volatilization of N from manure management (FvolN-N2Ofarm), and indirectly emitted due to
207
leaching/runoff from manure management in regions where leaching/runoff occurs (FleachN-
208
N2Ofarm):
209
𝐹𝑁2 𝑂−𝑓𝑎𝑟𝑚 = (𝐹𝑑𝑖𝑟𝑒𝑐𝑡𝑁−𝑁2 𝑂𝑀𝑀 + 𝐹𝑣𝑜𝑙𝑁−𝑁2 𝑂𝑀𝑀 + 𝐹𝑙𝑒𝑎𝑐ℎ𝑁−𝑁2 𝑂𝑀𝑀 ) × 𝑓𝑓𝑎𝑟𝑚 (Eq. S21)
210
Where ffarm is the weighted fraction of the number of days that livestock are fed by harvested
211
forage from grasslands (at-barn) at country scale, and is given by:
212
𝑓𝑓𝑎𝑟𝑚 = 1 − ∑(𝐷𝑝 × 𝑇𝑔𝑟𝑎𝑧𝑖𝑛𝑔(𝑝) × 𝐴𝑝 )⁄∑(𝐷𝑝 × 365 × 𝐴𝑝 )
213
Where Dp is optimal livestock density from ORCHIDE-GM in grid cell p in the given
214
country; Tgrazing(p) is the number of days of livestock grazing on pasture in grid cell p in the
(Eq. S22)
215
given country simulated in ORCHIDEE-GM; and Ap is the area of grid cell p in the given
216
country.
217
All the N2O emissions are calculated following the IPCC Tier 1 algorithms (IPCC, 2006 Vol
218
4, Chapter 10, Eqn 10.25). Direct N2O emission from manure management is calculated as:
219
𝐹𝑑𝑖𝑟𝑒𝑐𝑡𝑁−𝑁2 𝑂𝑀𝑀 = ∑(𝑁𝑒𝑥𝑖 × 𝐸𝐹6 × 𝑁𝑖 × 𝑀𝑆𝑖 )
220
where Ni is livestock numbers (head) of beef cattle, dairy cows, sheep, and goats calculated
221
from FAO statistics for each country; MSi is the fraction of total annual nitrogen excretion
222
that is managed in the manure management system; EF6 is the emission factor for direct N2O
223
emissions from the manure management system (kg N-N2O per kg N in manure management
224
system). Almost all the management system on European farms are solid storage or liquid
225
slurry with a natural crust cover, when the manure is not deposited on pasture (IPCC, 2006
226
Vol 4, Chapter 10, Table 10A.4 and 10A.5), thus the MSi =100% and EF6 = 0.005 is used in
227
this study. Nexi is annual average nitrogen excretion (kg N animal-1 yr-1), which is calculated
228
as (IPCC, 2006 Vol 4, Chapter 10, Eqn 10.30):
229
𝑁𝑒𝑥𝑖 = (𝑁𝑟𝑎𝑡𝑒(𝑖) × 𝑇𝐴𝑀𝑖 ⁄1000) × 365
230
where Nrate(i) is daily nitrogen excretion rate (kg N per 1000 kg animal mass), and TAMi is
231
typical animal mass for livestock category (kg per animal). The default value of Nrate(i) and
232
TAMi are adopted from IPCC, 2006 (Vol 4, Chapter 10, Table 10.19, and Table 10A-4 to
233
Table 10A-9). Indirect N2O emission due to volatilization from manure management is
234
calculated as (IPCC, 2006 Vol 4, Chapter 10, Eqn 10.26):
235
𝐹𝑣𝑜𝑙𝑁−𝑁2 𝑂𝑀𝑀 = ∑(𝑁𝑒𝑥𝑖 × 𝐹𝑟𝑎𝑐𝑔𝑎𝑠(𝑖) × 𝑁𝑖 × 𝐸𝐹7 )
236
where Fracgas(i) is percentage of managed manure nitrogen for livestock category that
237
volatilises as NH3 and NOx in the manure management system (default value for each type of
(Eq. S23)
(Eq. S24)
(Eq. S25)
238
livestock is derived from IPCC, 2006 Vol 4, Chapter 10, Table 10.22); EF7 is emission factor
239
for N2O emissions from atmospheric deposition of nitrogen on soils and water surfaces given
240
as the same default value of EF3 with 0.01 kg N2O-N per kg of NH3-N+NOx-N volatilised.
241
Indirect N2O emission due to leaching from manure management systems is calculated as
242
(IPCC, 2006 Vol 4, Chapter 10, Eqn 10.28):
243
𝐹𝑙𝑒𝑎𝑐ℎ𝑁−𝑁2 𝑂𝑀𝑀 = ∑(𝑁𝑒𝑥𝑖 × 𝐹𝑟𝑎𝑐𝑙𝑒𝑎𝑐ℎ(𝑖) × 𝑁𝑖 × 𝐸𝐹8 )
244
where Fracleach(i) is percent of managed manure nitrogen losses for livestock category due to
245
runoff and leaching during solid and liquid storage of manure (typical range 1%-20%; 5% is
246
used in this study); EF8 is emission factor for N2O emissions from nitrogen leaching and
247
runoff given as the same default value of EF4 with 0.0075 kg N-N2O per kg of N
248
leaching/runoff.
249
(Eq. S26)
250
Text S3. Grid point selection for NBP uncertainty analysis
251
Complete simulations (as described in the main text), including a 10,000-year spin-up
252
simulation to reach equilibrium for carbon pools and a 110-year simulation for the period of
253
1901 to 2010, with the 16 factor combinations at full geographical scale (9237 grid points)
254
would require far more computational time than was available to this study. Thus for the
255
uncertainty analysis, we have made a series of simulations for a small sub-sample of grid
256
points. Importantly, the grid points should represent the spatial distribution, magnitude and
257
interannual variability of the grasslands’ NBP. We carried out a series of selection processes
258
which picked grid points with regular latitude / longitude intervals (from 0.25° to 3°, which is
259
1 to 12 times the spatial resolution used in the main study). As a result, 12 groups of grid
260
points that were evenly spread across the study area were selected with 58 to 2309 grid points
261
over the study area (Table S2; Fig. S3 as an example). For each group, the NBP of the
262
corresponding grid points were extracted and averaged, the results were then compared with
263
those at full geographical scale of 9237 grid points (control group). Figure S4 shows the
264
differences in average NBP (with respect to the magnitude of NBP) and the correlation
265
coefficient between NBP time series from all grids (control group) and from each group of
266
grid points (with respect to interannual variability of NBP). Finally, Group 6 (195 grid points)
267
was selected for the uncertainty analysis in the main text, given the fact that its average NBP
268
is close to that of the control group and they are relatively highly correlated (Fig. D4).
269
270
Reference:
271
Bouwman AF, Van der Hoek KW, Eickhout B, Soenario I (2005) Exploring changes in world
272
273
274
ruminant production systems. Agricultural Systems, 84, 121-153.
Byrne KA, Kiely G, Leahy P (2007) Carbon sequestration determined using farm scale carbon
balance and eddy covariance. Agriculture Ecosystems & Environment. 12, 357-364.
275
Chang JF, Viovy N, Vuichard N, et al. (2013) Incorporating grassland management in
276
ORCHIDEE: model description and evaluation at 11 eddy-covariance sites in Europe.
277
Geoscientific Model Development, 6, 2165-2181.
278
FAOstat (2013) http://faostat3.fao.org
279
Flechard CR, Ambus P, Skiba U, et al. (2007) Effects of climate and management intensity on
280
nitrous oxide emissions in grassland systems across Europe. Agriculture Ecosystems
281
& Environment. 121, 135-152.
282
283
Hammond AC, Waldo DR, Rumsey TS (1990) Prediction of body composition in Holstein
steers using urea space. Journal of Dairy Science. 73, 3141–3145.
284
IPCC (2006) 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Prepared by
285
the National Greenhouse Gas Inventories Programme (eds Eggleston HS, Buendia L.,
286
Miwa K, Ngara T, Tanabe K). IGES, Japan.
287
IPCC (2013) The Physical Scientific Basis. In: Climate change 2013. Contribution of
288
Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on
289
Climate Change, 2013. Cambridge University Press, Cambridge, United Kingdom and
290
New York, NY, USA.
291
Olesen JE, Bindi M (2002) Consequences of climate change for European agricultural
292
productivity, land use and policy. European Journal of Agronomy, 16, 239-262.
293
Opio C, Gerber P, Mottet A, et al. (2013) Greenhouse gas emissions from ruminant supply
294
chains – A global life cycle assessment. Food and Agriculture Organization of the
295
United Nations (FAO), Rome.
296
Soussana JF, Allard V, Pilegaard K, et al. (2007) Full accounting of the greenhouse gas (CO2,
297
N2O, CH4) budget of nine European grassland sites. Agriculture Ecosystems &
298
Environment, 121, 121-134.
299
Vuichard N, Soussana J-F, Ciais P, et al. (2007) Estimating the greenhouse gas fluxes of
300
European grasslands with a process-based model: 1. Model evaluation from in situ
301
measurements. Global Biogeochemical Cycles. 21, doi:10.1029/2005GB002611.
302
Wells C (2001) Total energy indicators of sustainability: dairy farming case study, MAF
303
Technical Paper 2001/3, Ministry of Agriculture and Fisheries, Wellington, New
304
Zealand.
305
308