Carbon Footprint Assessment of Spanish Dairy Cattle Farms: Effectiveness of Dietary and Farm Management Practices as a Mitigation Strategy

Simple Summary Livestock production has been identified as an important source of greenhouse gas emissions. The current study was conducted to quantify the carbon footprint of Spanish dairy farms and to evaluate the potential of nutritional and management practices for mitigating methane emissions at farm level. The carbon footprint ranged from 0.67 to 0.98 kg CO2-eq/kg of energy corrected milk. Simulation scenarios showed that methane emissions and the carbon footprint of milk could be reduced more through management practices rather than dietary strategies. Modelling may provide policy makers, farmers and stakeholders valuable information for planning and developing strategies to reduce the carbon footprint associated with milk production. Abstract Greenhouse gas emissions and the carbon footprint (CF) were estimated in twelve Spanish dairy farms selected from three regions (Mediterranean, MED; Cantabric, CAN; and Central, CEN) using a partial life cycle assessment through the Integrated Farm System Model (IFSM). The functional unit was 1 kg of energy corrected milk (ECM). Methane emissions accounted for the largest contribution to the total greenhouse gas (GHG) emissions. The average CF (kg CO2-eq/kg of ECM) was 0.84, being the highest in MED (0.98), intermediate in CEN (0.84), and the lowest in CAN (0.67). Two extreme farms were selected for further simulations: one with the highest non-enteric methane (MED1), and another with the highest enteric methane (CAN2). Changes in management scenarios (increase milk production, change manure collection systems, change manure-type storage method, change bedding type and installation of an anaerobic digester) in MED1 were evaluated with the IFSM model. Changes in feeding strategies (reduce the forage: concentrate ratio, improve forage quality, use of ionophores) in CAN2 were evaluated with the Cornell Net Carbohydrate and Protein System model. Results indicate that changes in management (up to 27.5% reduction) were more efficient than changes in dietary practices (up to 3.5% reduction) in reducing the carbon footprint.


Modeling Procedure
Farm emissions were simulated using the IFSM, a mechanistic model developed and validated by Chianese et al. [12] to calculate CH 4 , N 2 O, CO 2 and total GHG emissions from dairy cattle production systems through a partial LCA. The IFSM was further refined into a process-based whole farm simulation including major components for soil processes, crop growth, tillage, planting, harvesting, feed storage, feeding, herd production, manure storage and economics in dairy farms ( Figure 1) [24]. The functional unit considered in this study was 1 kg of ECM at the farm gate. The system boundaries contained all relevant activities from cradle to farm gate including on-farm activities, the production of farm inputs and carbon sequestration. Emissions attributable to imported feeds and its transportation were excluded. In addition, the model was adapted to be used in Spanish conditions such as soil type, nutritional characteristics of forages and concentrates, and the 10-year average daily local weather. Weather variables included annual temperature, relative humidity, wind velocity, mean precipitation and solar radiation from three meteorological stations corresponding to each region. requested (type and amounts of forages and concentrates) together with nutritional management strategies (feeding groups for each herd, feeding system, grazing activity and mineral supplementation). The general characteristics of the selected dairy farms from each region are presented in Table 1.

Modeling Procedure
Farm emissions were simulated using the IFSM, a mechanistic model developed and validated by Chianese et al. [12] to calculate CH4, N2O, CO2 and total GHG emissions from dairy cattle production systems through a partial LCA. The IFSM was further refined into a process-based whole farm simulation including major components for soil processes, crop growth, tillage, planting, harvesting, feed storage, feeding, herd production, manure storage and economics in dairy farms ( Figure 1) [24]. The functional unit considered in this study was 1 kg of ECM at the farm gate. The system boundaries contained all relevant activities from cradle to farm gate including on-farm activities, the production of farm inputs and carbon sequestration. Emissions attributable to imported feeds and its transportation were excluded. In addition, the model was adapted to be used in Spanish conditions such as soil type, nutritional characteristics of forages and concentrates, and the 10-year average daily local weather. Weather variables included annual temperature, relative humidity, wind velocity, mean precipitation and solar radiation from three meteorological stations corresponding to each region. Figure 1. System boundaries, primary and secondary greenhouse gas emissions sources and sinks for a partial life cycle assessment of the carbon footprint (adopted from Rotz et al. [13]). Figure 1. System boundaries, primary and secondary greenhouse gas emissions sources and sinks for a partial life cycle assessment of the carbon footprint (adopted from Rotz et al. [13]).
In addition, the Cornell Net Carbohydrate and Protein System Model (CNCPS version 6.1; [25]) was used to evaluate the contribution of diets to enteric CH 4 emissions of lactating cows on each farm and explore the impact of dietary modifications on CH 4 emissions. The CNCPS is a mechanistic mathematical model that estimates cattle requirements and nutrient supply based on animal, environment, and feed composition information and predicts nitrogen (N), phosphorus (P) and CH 4 emissions, enabling its integration in a whole-farm nutrient management plans [25,26].

Methane Mitigation Scenarios Simulation
To assess how changing dietary and management practices affects the enteric, non-enteric CH 4 emissions and milk CF, different scenarios were compared. Two extreme dairy farms were selected for further simulations: one with the largest enteric CH 4 emissions, and another with the largest non-enteric CH 4 emissions. The CNCPS model [25] was used in the farm with the largest share of enteric CH 4 emissions to evaluate three changes in the diet composition of milking cows to determine how these changes would affect enteric CH 4 emissions: (i) Reducing by 10% the forage to concentrate ratio; (ii) Improving forages quality: the original ryegrass silage (9% crude protein, CP; 65% neural detergent fiber, NDF; and 8% lignin, ADL) was changed by an improved ryegrass silage (21% CP, 50% NDF and 7% ADL); (iii) Supplying 330 mg/day of monensin.

Greenhouse Gas Emissions Profile and the Carbon Footprint of the Selected Dairy Farms
The estimated annual averages for CH 4 , N 2 O, CO 2 , total GHG emissions and CF of milk (kg of CO 2 -eq/kg of ECM) from dairy cattle farms by source of emissions (animal housing, manure storage, manure field application, grazing, fuel combustion and secondary emissions) from MED, CAN and CEN regions are presented in Table 2. The annual average CH 4 emissions were the highest in MED (328 ± 25.5 kg/cow), intermediate in CEN (273 ± 52.0 kg/cow) and the lowest in CAN (251 ± 48.0 kg/cow). The highest annual average N 2 O emissions were in MED (6.5 ± 1.81 kg/cow), intermediate in CEN (4.0 ± 1.52 kg/cow) and the lowest in CAN (3.2 ± 1.66 kg/cow). Negative values were obtained for the annual average net CO 2 emissions on all farms as the positive contributions from animal and housing, manure storage, grazing and fuel combustion were offset by the net feed production. The average CF of milk in selected farms was 0.87 kg CO 2 -eq/kg of ECM, being the highest in MED (1.0 ± 0.12), intermediate in CEN (0.8 ± 0.08) and the lowest in CAN (0.7 ± 0.07 kg of CO 2 -eq/kg of ECM). The breakdown of total GHG emissions into component gases (enteric CH 4 , non-enteric CH 4 , N 2 O and CO 2 ) was examined in absolute terms and relative to each other for all farms. CH 4 emissions from enteric fermentation and manure handling (non-enteric fermentation) represented the largest share of GHG emissions, accounting for more than 70% of total farm emissions.

Diet Evaluation with the CNCPS Model and Its Contribution to Enteric Methane Emissions
Diet evaluation of each farm with the CNCPS model and its contribution to enteric CH 4 emissions per unit of milk produced are presented in Table 3. The highest enteric CH 4 emissions (g/kg of milk) were recorded in CAN farms with an average of 13.6 and a range from 12.2 in the CAN1 to 16.2 in the CAN2 farms. Intermediate CH 4 emissions (g/kg of milk) were recorded in MED with an average of 12.5 and a range from 11.2 to 13.3 in the MED3 and MED4 farms, respectively. The lowest CH 4 emissions (g/kg of milk) were recorded in CEN farms with an average of 12.4 and a range from 10.7 to 14.4 in CEN4 and CEN1 farms, respectively. Table 3. Enteric methane emissions evaluation from feed intake in the selected farms using the CNCPS 1 model.

Characteristics of the Two Selected Extreme Farms
Two extreme dairy farms in term of CH 4 emissions were selected for further simulations (Table 4): one with the largest non-enteric CH 4 emissions (MED1), and the other with the largest enteric CH 4 emissions (CAN2). The first farm was used to simulate the effects of changes in management on non-enteric CH 4 emissions using the IFSM model. Farm MED1 is located on a 330 ha clay loam soil in Girona that is used to provide most of the forage required for feeding cows (ryegrass and corn silage). It has 82 milking Holstein cows plus replacement heifers housed in a free stall barn and bedded pack barns, respectively. Average intake was 26.5 kg DM of a 55:45 forage to concentrate ratio, and average production was 32.0 L/cow/day. The second farm was used to simulate the effects of changes in diet composition on enteric CH 4 emissions using the CNCPS model. The CAN2 farm is on a 25 ha of loam soil in Lugo (northern Spain) that produces all required ryegrass and forage for pasture. It has 42 milking Holstein cows housed in a free stall barn with access to pasture. Average estimated intake and production of milking cows was 21.3 kg DM (68:32 forage to concentrate ratio) and 28.5 L/cow/day of milk, respectively.

Dietary Change Scenarios
The CNCPS model was used to analyze three potential changes in diet composition and quality in the CAN2 farm to determine how these changes affected their enteric CH 4 emissions. Results of the simulated scenarios are shown in Table 5. The 10% reduction in the forage: concentrate ratio reduced CH 4 emissions from 16.4 to 15.6 g/kg milk, representing a 5% reduction in CH 4 emissions, but only 3.5% of the milk CF. In contrast, change in the nutritional value of the ryegrass silage had a small impact on CH 4 emissions which decreased from 16.4 to 16.3 g/kg milk. Finally, the addition of 330 mg/day monensin also had small effects on CH 4 emissions from 16.4 to 16.3 g/kg milk. The impact of these two dietary factors on the milk CF was negligible.

Management Change Scenarios
The effects of management changes on CH 4 emissions and the CF in the MED1 farm are presented in Table 6. Simulation with the IFSM revealed that the increased milk production by 13% reduced slightly CH 4 emissions (−0.2%), but the impact on the milk CF was moderate (−6.4%). Change in bedding types from straw to sand had a larger effect, reducing CH 4 emissions by 10% and the milk CF by 13.8%. Farm MED1 has already a semi-solid manure management system that it is rather efficient, reducing CH 4 emissions around 32% and the milk CF by 7.4% compared with slurry or liquid-slurry systems. The utilization of a solid manure management system would further reduce the milk CF by 17.4%. The construction and use of an anaerobic digester would reduce CH 4 emissions by 29.9% and the milk CF by 24.8%. The highest reduction potential would be obtained if the top-loaded lined earthen basin is changed into an enclosed tank, with a 37.5% and 27.5% reduction in CH 4 emissions and the milk CF, respectively.

The Carbon Footprint
The milk CF among all farms ranged from 0.6 to 1.1 kg CO 2 -eq/kg ECM and was within the range reported in earlier LCA studies using the same model for conventional dairy farms [13,15]. Del Prado et al. [17] reported a slightly higher CF (0.84 to 2.07 kg CO 2 -eq/kg ECM) in 17 commercial confined dairy farms in the Basque Country (northern Spain) using a combination of sub-models approaches. Laca et al. [18] also reported a higher CF of milk in semi-confined dairy cows in northern Spain (1.22 kg CO 2 -eq/kg ECM). Differences compared with our results are explained by the different system boundaries delimitation used, where emissions from transport and purchased feeds were included in these previous studies but not in ours, which justifies higher values. In contrast, the milk CF in Pennsylvania and California farms were lower (0.60, 0.55 and 0.61 kg CO 2-eq/kg ECM, respectively) [13] than results reported herein. Belflower et al. [27] found that the CF in conventional United States (USA) farms was about 0.87 kg CO 2 -eq/kg ECM, closer to our estimates, but variation was much larger in Holstein cows in Australia (range from 0.54 to 1.35 kg CO 2 -eq/kg ECM; [15]). Results obtained in our simulations are within the range of those reported elsewhere in similar conditions.

Methane and Carbon Footprint Mitigation Strategies
CH 4 is one the main GHG produced at farm gate from enteric and non-enteric fermentation in dairy farms which mitigation may help reduce the CF [28][29][30]. It is very important to identify which strategies may have the highest impact on the reduction of CH 4 emissions and the milk CF, so policy-makers, farmers and stakeholders may direct their policies toward the most efficient processes. In the current study, we evaluated the effects of dietary and management changes on CH 4 emissions and milk CF using CNCPS and IFSM, respectively.

Dietary Change Scenarios
Dietary changes in the CAN2 farm were simulated with the CNCPS model, and reformulated to optimize production using the ingredients available in the farm. Our results suggest that CH 4 emissions originated from enteric fermentation and the CF of milk could be reduced up to 5.0% and 3.5%, respectively, when the proportion of forage was reduced by 10%. Feeding grain increases ruminal propionate while lowering acetate levels from microbial fermentation [29][30][31]. While fibrolytic bacteria ferment toward acetate, which releases a carbon that methanogenic bacteria reduce with metabolic hydrogens to produce CH 4 , amylolytic bacteria ferment towards propionate that consumes metabolic hydrogens. Therefore, when the fermentation is shifted towards a lower acetate to propionate ratio, CH 4 production is reduced and justifies the results observed. Also, the increase in concentrate in the diet resulted in an increase in milk production that contributed to the dilution of the amount of CH 4 per kg of milk. In contrast, improving silage quality had negligible impact on reducing CH 4 emissions and the milk CF. The small impact of an improved silage quality was expected. Most dairy diets in Spain contain under 55% forage in their diets, and quality is already reasonably high. Increasing forage quality implies a better digestibility of the fiber components, but probably has a small impact on the fermentation pathways, producing similar proportions of CH 4 . There are few in vivo data exploring the impact of forage quality on CH 4 emissions and the milk CF. Boadi and Wittenberg [32] reported that CH 4 emissions in dairy heifers were reduced by 23.4 and 42.5% when forage quality improved from low to medium and high quality silage, respectively. However, the high proportional impact is counterbalance with the small overall contribution of young non-producing animals to the overall farm GHG production. Finally, the impact of monensin on CH 4 emissions observed in our report was small. Monensin decreased the acetate-to-propionate ratio in rumen fluid through increasing the flow of reducing equivalents to form propionate [33]. In fact, results from a meta-analysis conducted based on 11 experiments on lactating dairy cows showed that monensin supplementation reduced CH 4 production by about 2−3% [34]. It is possible that the small changes reported herein result from an underestimation of the effects of monensin on CH 4 emissions in the CNCPS model. However, the effects of monensin, even considering in vivo evaluations, are relatively small. All dietary changes proposed herein are within reasonable changes to be implemented in a dairy farm. Therefore, the overall potential of dietary changes appears to have a relatively small impact of up to 5% in the reduction of CH 4 emissions and up to 3.5% in the CF.

Management Change Scenarios
On the other hand, the IFSM model was used to simulate management scenarios in the MED1 farm with the highest non-enteric CH 4 emissions. Three major strategies should be considered: increase productivity to dilute fixed GHG emissions; use bedding material that produce less emissions; and modify manure management practices.
Increasing productivity of milking cows is an effective strategy to mitigate GHG emissions, which may allow a reduction in animal numbers providing the same edible product output at a reduced CF. We estimated that an increase of around 1.500 L/cow/year would reduce the CF of milk by 6.4%. Boadi and Wittenberg [32] also reported that increasing milk production of dairy cows from 5000 to 10,000 L/cow/year in the European Union would increase total CH 4 production per animal by 23%, but the production/ECM would be reduced up to 40%. In fact, the sustained increase in productivity was the single most influential mitigation factor which reduced the CF of milk in the United States dairy industry from 1944 to 2007 [35].
Bedding material may also have a relevant impact on CH 4 emissions and the milk CF. The MED1 farm had a straw-type bedding. While differences in CF compared with other organic alternatives were small, the use of sand would have a moderate, yet relevant, impact, reducing CH 4 emissions by 9.8% and the milk CF by 13.8%. The potential impact of bedding type on the CF reported herein is relevant and within the ranges observed by Aguirre-Villegas and Larson [36].
Manure is the second largest source of GHG emissions from dairy farms [36]. Therefore, manure handling strategies can be effective strategies to reduce non-enteric CH 4 emissions and the milk CF. In the current study, the MED1 farm had a semi-solid manure collection system. The solid collection systems are the least contaminating in terms of GHG emissions compared with semi-solid, multiple and liquid systems [36,37]. Solid manure has less available water and is usually stored in stock piles that promote aeration, which reduces CH 4 emissions. In contrast, liquid manure is stored on pits that promote anaerobic conditions, increasing CH 4 emissions. Compared with slurry or liquid-slurry systems, MED1 farm has already reduced CH 4 emissions and the CF by around 32 and 7%, respectively. However, moving MED1 semi-solid manure into a solid system would have an additional important impact on the CF (17.4% reduction). The installation of an anaerobic digestor could reduce enteric CH 4 and the milk CF by 29.9 and 24.8%, respectively. Aguirre-Villegas and Larson [36] reported that anaerobic digestion can reduce GHG emissions related to manure management by more than 50%, mostly in the form of CH 4 during storage. However, the largest contribution to the reduction of the milk CF would come from shifting an uncovered storage tank to an enclosed one that would reduce non-enteric CH 4 and the CF of milk up to 35.7 and 27.5%, respectively. This large impact of an enclosed manure handling system agrees with Rotz et al. [13], who reported that enclosing manure storage with a flare to burn the escaping biogas would reduce by 39% the milk CF. Overall, manure handling has the largest potential to effectively contribute to the reduction of CH 4 and the milk CF compared with those obtained through changes in dietary strategies.
This study has several limitations. For example, the costs of the proposed mitigating strategies were not considered. Therefore, future studies aimed at estimating the cost of implementation of each strategy will help clarify the cost effectiveness of these mitigation strategies. The uncertainties and assumption in primary and secondary input data such as the average body weight, feed dry matter intake, ingredient and chemical composition of diets, fertilizers and manure application to croplands and pasture emissions factors may affect the accuracy of the results of this analysis using the IFSM. Rotz et al. [37] reported that uncertainties for IFSM predictions of GHG emissions were ±10 and ±20% for enteric and manure handling CH 4 , respectively, ±50% for N 2 O, and ±20% for fuel combustion and upstream emissions. In addition, the current analysis considered all relevant activities from cradle to farm gate, but excluded emissions attributable to imported feeds and its transportation. Finally, strategies to reduce CH 4 emissions could be associated with other environmental impact such as water contamination, denitrification and eutrophication. Therefore, the milk CF analysis should be integrated with other footprint indicators (water, land and energy) in order to cover other environmental impacts spectrum of dairy farming.

Conclusions
The IFSM was used in this study to quantify the profile of GHG emissions in twelve dairy cattle farms from three Spanish regions. This study revealed that Cantabric farms are more efficient in term of GHG emissions compared with Mediterranean and Central dairy farms. CH 4 was found to be the major contributor to the milk CF. Mitigation efforts are frequently targeted to changes in dietary strategies while management efforts are often ignored. However, simulation scenarios showed that CH 4 emissions and the CF of milk could be reduced more through management practices compared with dietary strategies. Modelling may provide policy makers, farmers and stakeholders valuable information for planning and developing strategies to reduce the CF associated with milk production.
Funding: This research received no external funding.