The Carbon Footprint of Milk Production on a Farm

Abstract

The environmental impact of milk production, particularly its share of greenhouse gas (GHG) emissions, is a topic under investigation in various parts of the world. This paper presents an overview of current knowledge on the carbon footprint (CF) of milk production at the farm level, with a particular focus on technological, environmental and organisational factors affecting emission levels. The analysis is based on a review of, inter alia, 46 peer-reviewed publications and 11 environmental reports, legal acts and databases concerning the CF in different regions and under various production systems. This study identifies the main sources of emissions, including enteric fermentation, manure management, and the production and use of feed and fertiliser. It also demonstrates the significant variability of the CF values, which range, on average, from 0.78 to 3.20 kg CO₂ eq kg⁻¹ of milk, determined by the farm scale, nutritional strategies, local environmental and economic determinants, and the methodology applied. Moreover, this study stresses that higher production efficiency and integrated farm management could reduce the CF per milk unit, with further intensification having, however, diminishing effects. The application of life cycle assessment (LCA) methods is essential for a reliable assessment and comparison of the CF between systems. Ultimately, an effective CF reduction requires a comprehensive approach that combines improved nutritional practices, efficient use of resources, and implementation of technological innovations adjusted to regional and farm-specific determinants. The solutions presented in this paper may serve as guidelines for practitioners and decision-makers with regard to reducing GHG emissions.

1. Introduction

The environmental impact of milk production is a subject of growing global concern due to the sector’s share of anthropogenic greenhouse gas (GHG) emissions. One of the key indicators in an environmental impact assessment is the carbon footprint (CF), which determines the total greenhouse gas emissions attributed to a particular product or process, expressed in terms of the carbon dioxide equivalent (CO₂ e or CO₂ eq). As far as milk production is concerned, the carbon footprint includes emissions from, e.g., enteric fermentation, fertiliser management, feed production, use of outside inputs, and energy consumption [1,2,3]. The dominant greenhouse gas is methane (CH₄), produced during digestion in ruminants, followed by nitrous oxide (N₂O) originating from fertilisation and fertiliser management, and carbon dioxide (CO₂) associated with land use and fuel combustion [4,5].

The carbon footprint is sometimes defined as the weight of CO₂ eq attributed to a functional unit of a product (most frequently, 1 kg of fat- and protein-corrected milk (FPCM), calculated based on the GHG emissions (CO₂, CH₄, N₂O) generated throughout the entire life cycle of the product [6,7]. Various stages of milk production, such as fertiliser and feed production, manure management, fuel consumption, and the animals’ digestive processes themselves, contribute, to varying degrees, to the total CF. For example, enteric fermentation in dairy cows can account for up to 40–60% of the total CH₄ emissions [8].

Currently, life cycle assessment (LCA) is a recognised tool for assessing the environmental impact of milk production systems. LCA enables analysis of both direct and indirect emissions, considering, inter alia, the production of feed, energy and fertiliser, animal husbandry, manure management, and even milk cooling [5,9]. This methodology allows different systems, technologies and management practices to be compared, and the key emission areas and possibilities for their reduction to be identified. LCA analysis, conducted in accordance with standards ISO 14040 and 14044 [10,11] and the GHG Protocol, provides a basis for reliable carbon footprint reporting in the context of the requirements of the Environmental, Social and Governance (ESG) Directive and the Corporate Sustainability Reporting Directive (CSRD) [12,13].

Recent years have seen increasing pressure on agriculture and the dairy sector, forcing them to adapt to climate regulations and EU strategies, such as the Green Deal, the “From Farm to Fork” strategy, and the new European Sustainability Reporting Standards (ESRS) under the CSRD. Within their framework, the CF is among the key indicators used to reveal the impact of business activities on the climate and GHG emissions [13]. For the dairy sector, this implies the need for the systematic measurement and reporting of the CF, especially in the context of cooperation with large customers such as processing plants, retail chains, and financial institutions.

Over the past few years, many LCA-based studies have shown significant variability in the carbon footprint of milk production depending on the region, scale of production, and the feed, maintenance, and animal husbandry systems used. Flysjö et al. [1] noted CF values ranging from 0.9 to 1.4 kg CO₂ eq kg⁻¹ FPCM in northern Europe, whereas Gollnow et al. [14] demonstrated an average CF value at a level of 1.11 kg CO₂ eq kg⁻¹ FPCM in Australia. In contrast, considerably higher values were noted in small-scale systems: in India, 2.2 kg CO₂ eq kg⁻¹ FPCM [15]; in Ethiopia, up to 3.2 kg [16]; and in Colombia, from 1.8 to 3.9 kg, depending on the system [17]. These discrepancies result, e.g., from differences in the milk yield, feed conversion efficiency, fertiliser management and dependence on imported feed components.

Feeding systems also have a major impact on emission levels. Studies have demonstrated that intensive total mixed ration (TMR) systems, although providing high productivity, are often associated with a higher CF due to the use of concentrated feed with a high land use change (LUC) index value, e.g., soybean meal [18,19]. In contrast, systems based on, e.g., pastures achieve a lower CF value despite lower productivity, which was confirmed, among others, in Portugal (0.83 kg CO₂ eq kg⁻¹ of milk) [20] and New Zealand (0.75–0.81 kg, depending on the season of 2010/11–2017/18; an average of 0.78 kg) [21].

The scale of a farm also affects the CF levels. Smaller farms often have no access to the right technology to optimise resource consumption, which translates into a higher CF per product unit [3,4]. Nevertheless, certain studies highlighted the advantages of multifunctional farms. Garg et al. [15] demonstrated that after considering the additional functions, the CF of Indian farms decreased by over 20%, which suggests that the CF should be analysed in a context broader than milk production efficiency alone. Another challenge is the lack of methodological standardisation in research. The differences relate to system boundaries, functional units (e.g., raw milk vs. fat- and protein-corrected milk (FPCM)), allocation methods (e.g., economic or mass), and the global warming potential (GWP) values adopted (e.g., GWP100 vs. GWP20), which affects the interpretation of results [1,3,22]. The consideration of carbon (C) sequestration and other ecosystem services is also playing an increasingly important role in LCA analyses. Studies conducted in Colombia and Brazil showed that properly managed pastures and silvopastoral systems can partially offset emissions [17,23].

In the European and Polish contexts, the significance of integrated and transparent environmental indicators is increasing. These will play a crucial role in the implementation of strategies such as the European Green Deal and the CSRD Directive. The carbon footprint is becoming part of ESG reporting and is increasingly appearing as a criterion in product labelling systems, agricultural advisory systems, and investment financing [24,25]. Therefore, in response to these challenges, tools such as LatteGHG [26] or DairyGHG [9], which support producers in the independent assessment of the carbon footprint, are being developed. In practical terms, the CF can indicate directions for action to reduce emissions [27,28,29]. For example, the use of high sugar grass (HSG) varieties allows N₂O emissions to be reduced by as much as 11% [27], and feed additives can reduce CH₄ emissions [30].

Despite advances in research, gaps continue to exist, especially in low- and middle-income countries, where dairy systems are often dispersed and difficult to inventory. It is also necessary to better integrate the CF with other sustainable development indicators, such as biodiversity, soil health, animal welfare, and social justice. Bell [25] pointed out that a single CF indicator, although useful, can blur the trade-offs between various environmental and social aspects and should be part of broader multi-criterial analyses.

The aim of the present study was to comprehensively analyse the CF of milk production at the farm level, taking into account technological, environmental and systemic factors. Based on a review of scientific publications and environmental reports, databases and legal acts [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57], the stages of calculating the product carbon footprint, the application of LCA methodology, and the calculation approaches were identified (Section 3.1.). The main sources of GHG emissions, including the sources in milk production, were presented (Section 3.2). The CF range in milk production (Section 3.3), the effect of feed, its origin and feeding system on the CF (Section 3.4), and the production efficiency and CF (Section 3.5) were also presented. This paper also includes the farm scale and its characteristics (Section 3.6), as well as methods for reducing GHG emissions and the CF, technological innovations and future development directions (Section 3.7). A particular focus was placed on the relationships between the CF and production intensity, local environmental conditions, and possible decarbonisation pathways for the dairy sector. The aim of this study was not only to identify the emission factors but also to provide useful conclusions for farmers, advisors, decision-makers and institutions implementing climate policy in agriculture, also at the local/regional level.

2. Materials and Methods

2.1. Data Sources and Scope of Analysis

Specific guidelines were followed to ensure quality and obtain a transparent review. Firstly, the search strategy used several bibliometric databases: Taylor & Francis; Wiley Online Library; ScienceDirect (Elsevier); MDPI; and Google Scholar. Advanced search techniques for research and review papers (only the articles to which the authors had full access) were applied for the time period until 25 May 2025. The following words were adopted to be used in the search of titles and keywords: carbon footprint, milk farm, milk production, dairy cattle, cattle, dairy cows, cows. Publications were then selected, which, by their content, referred to the aspect of the carbon footprint of milk production on farms. All the selected articles are cited in the text and are available in English. A total of 46 papers and 11 environmental reports, databases and legal acts were analysed. Most of the articles were discussed, and scientific evidence of possible cause-and-effect mechanisms/sequences was presented based on them. They constitute specific guidelines or tips for managing the carbon footprint on dairy farms, which can be used not only by breeders/farmers but also by management institutions.

The CF values in individual countries, derived from the literature, were statistically analysed; a statistical description was prepared, which helped determine the N-valid, mean confidence interval 95%, median, minimum, maximum, lower quartile, upper quartile, range, interquartile range, variance, standard deviation, coefficient of variation, standard error of mean, skewness, and kurtosis. Moreover, the mean values, medians, standard deviations and standard errors of the mean were determined for each continent. All the statistical analyses were conducted using STATISTICA 13 software (TIBCO Software Inc., Palo Alto, CA, USA).

Thanks to gathering the collected literature, also in the Research Information Systems (RIS) format, simultaneous analysis of all the titles and abstracts was conducted (co-word analysis) (Section 2.2). VOSviewer, version 1.6.20 (Centre for Science and Technology Studies, Leiden University, The Netherlands), was used. The selected counting method was binary, with a minimum of six occurrences of each term. The number of terms to select was 60% of the most relevant terms. Finally, the terms were verified and selected manually. This resulted in a breakdown into the current main research areas (transparent co-word network, clear emphases, and close and stable correlations between co-words in a particular area) as well as the main trends over time and the issues that have received the most attention.

All the graphics are original and were created using Canva’s Content License Agreement.

2.2. Main Research Areas

The collected research can be attributed to three main areas (Figure 1). The first (pink) research area concerns CH₄, N₂O, manure, feed, production system, dairy production system, meat, animal, and environmental impact. The second (purple) research area combines all the issues related to the milk carbon footprint, kg FPCM, reduction, cattle, impact, greenhouse gas emission, pasture, milk yield, and allocation method. The third (turquoise) research area contains the remaining strongly interlinked issues: energy, GHG, greenhouse gas, case study, farm gate, dairy sector, time, and global warming.

Among the collected manuscripts, the most common research subject or source of data for the research was Italy (six collected studies), followed by India (five) (Figure 2). The least common were Romania, Turkey, Costa Rica, Sweden, Australia, the Netherlands, Northern Ireland, France, Mongolia and Kenya (one study each).

3. Results and Discussion

3.1. Stages of Calculating the Product Carbon Footprint, the Application of LCA Methodology, and the Calculation Approaches

The calculation of the product carbon footprint (PCF), including that for milk production, comprises several key stages in accordance with standards such as the GHG Protocol Product Life Cycle Accounting and Reporting Standard [30,31]. This is closely related to the life cycle assessment (LCA) methodology outlined in standards ISO 14040 [10] and 14044 [11]. The stages of the analysis include the following [32].

(I)

Determining the objective and scope of the study, which involves the following. (a) Identifying the purpose of the analysis, which may be to determine the carbon footprint of milk production on a farm or along the entire production chain, milk consumption, and packaging handling. It may also involve a comparison of the CF values from different groups of farms (e.g., farm size) or two products, e.g., organic and conventionally produced milk. (b) Selecting and determining the system boundaries, i.e., which treatments and processes will be considered in the analyses. Does the research end with milk production on the farm, or perhaps at the entrance or exit gate of the dairy, or does it also take into account further distribution of milk, preparation and consumption, and disposal/recycling of packaging? (c) Determining the functional unit for which greenhouse gas emissions are calculated, e.g., 1 litre of milk, 1 kg of milk, 1 kg of FPCM, 1 kg of cheese, 1 dairy cow, etc.

(II)

Collecting input data based on specific system boundaries, e.g., acquisition of raw materials (fertilisers, feed materials, feed); production (energy use, fuel consumption); distribution and transport (fuels and energy in transport and storage); use (storage, processing at the consumer’s premises); or waste management (recovery, recycling or landfilling). Depending on the LCA accuracy, the data may originate from actual measurements (e.g., detailed, simplified LCA) but also from existing databases or estimates (simplified, screening LCA).

(III)

Calculating greenhouse gas emissions for all the treatments and processes taken into account in the analysis, based on relevant emission indicators for different greenhouse gases (e.g., CO₂, N₂O, CH₄) to carbon dioxide equivalents (CO_{2 eq}), most often using commercially available software dedicated to performing life cycle assessment or carbon footprint analysis. By using the software, various methodologies for environmental impact assessment that consider greenhouse gas emissions can be easily applied.

(IV)

Interpreting the results, enabling the quantitative determination of greenhouse gas emissions per selected functional unit, both for the product and individual stages of its production or logistics. At this stage, it is possible to identify stages with high CF levels and propose solutions for reducing emissions, such as using green energy or more efficient transportation.

(V)

Reporting emissions considering greenhouse gas emissions in terms of the CO₂ eq. The documentation should include all the assumptions, restrictions, data sources and methodology for each of the four stages of the carbon footprint calculation outlined above. It is essential to note that for more detailed analyses (e.g., detailed LCA), ISO 14044 standards [11] require the use of an external critical review to ensure the transparency and impartiality of the study, as well as compliance with ISO standards.

The choice of LCA methodology has a significant impact on the carbon footprint calculation results. Pirlo [33] emphasised that the comparability of CF results between studies is limited by the strong heterogeneity of the LCA methods applied, including different system boundaries, allocation methods and the selection of a functional unit. Flysjö et al. [1] demonstrated that the application of system expansion significantly reduces the share of emissions attributable to milk, as compared with the allocation methods. In addition, Froldi et al. [3] documented the impact of the uncertainty of input data, particularly regarding the origin of feed, on the final CF result, which emphasises the significance of accurate data collection at the farm level.

According to Kiefer et al. [34], the inclusion of ecosystem services, such as C sequestration or biodiversity protection, into the life cycle assessment can significantly affect the final CF value. One of the studies [22] examined the differences resulting from applying various levels of the global warming potential (GWP). In certain cases, selecting GWP20 instead of GWP100 changed the share of methane in the total CF from 50% to over 70%. Hospers et al. [35] noted that taking into account direct LUC indices and the C balance in the soil can decrease the CF reduction rate by up to 6 percentage points, which also highlights the importance of methodological assumptions.

The analysed studies used different approaches to LCA, which hampered the comparison of the results. The differences included, inter alia, system boundaries (field-to-farm-gate, cradle-to-farm-gate), functional unit type (kg of milk, kg of FPCM, or ha), and method for emission allocation between milk and co-products (meat, natural fertiliser). For this purpose, allocation methods such as mass, economic and bioeconomic allocation were applied [5,34]. The application of Intergovernmental Panel on Climate Change (IPCC) Tier 2 and Tier 3 indicators enables a more accurate estimation of emissions than simplified methods (Tier 1), yet they require detailed data from farms. Some studies also took into account GWP20 rather than GWP100, which changed the importance of CH₄ as a climate impact factor.

3.2. Main Sources of GHG Emissions, Including Sources in Milk Production

The total GHG emissions, excluding the land use, land use change and forestry (LULUCF) categories worldwide, amounted to approximately 53.4 billion tonnes of CO₂ eq, of which as much as 72.8% was attributable to the energy sector (Figure 3) [36]. Industrial processes and product use, as well as waste management, accounted for 8.4% and 4.8% of GHG emissions, respectively. In contrast, global agriculture was responsible for 11.4% of GHG emissions, with significant variations observed on individual continents. The largest share of GHG emissions came from agriculture in South America (38.1%) and in Africa, Australia and New Zealand (26.3–29.9%), whereas the smallest share came from North America, Asia and Europe (6.4–8.7%). It should also be emphasised that in Africa and South America, the LULUCF sector accounted for a very large share of emissions (29.9–36.8%), which is primarily related to deforestation. In contrast, the lowest figure of this indicator (–9.8%) was noted in Europe. The negative value of GHG emissions implies carbon sequestration in ecosystems, which is primarily associated with effective forestry and agriculture management.

The main sources of GHG emissions from the agricultural sector included animal production. Globally, enteric fermentation in livestock accounted for nearly 48% of GHG emissions, while manure management, manure left on pasture and manure applied to soil accounted for a combined 22% of GHG emissions from agriculture (Figure 4). The European Union observed a higher total share of manure management, manure left on pasture and manure applied to soil (30%) compared with the global average value, while the values for enteric fermentation were at a similar level (46%). In addition, rice cultivation and the use of synthetic fertilisers (a total of 20% of GHG) had a significant impact on the Earth’s climate on a global scale, while at the EU level, the primary concern was the reliance on synthetic fertilisers (11.6%).

In the EU, the proportion of GHG emissions attributed to agriculture varied between Member States. The lowest shares of GHG emissions from agriculture were noted in Malta, Slovakia and Cyprus (3.7–6.2%), while the highest shares were found in Denmark and Ireland (28.6–36.2% excluding LULUCF) (Figure 5). The sector with the largest share of GHG emissions was energy in each of the EU Member States. The lowest share was noted in Ireland (57.1%), and the highest shares were in Poland, Germany and Luxembourg (83.2–84.8% excluding LULUCF). It should be stressed that in 19 out of 27 EU countries, the LULUCF sector showed negative GHG emissions (positive carbon sequestration), which largely offset the GHG emissions from other sectors, including agriculture. In this respect, Sweden and Romania stand out, as in these countries, the value of this parameter amounted to −70.3 and −44.7%, respectively (a % value of total GHG emissions from other sectors).

Eurostat data indicate that (dairy and beef) cattle farming accounted for approximately 50% of GHG emissions from agriculture in the EU, with enteric fermentation accounting for 42% and manure management for 8% (Figure 6) [8]. For other animals, these indicators were 7% and 9%, respectively. Significant differences were observed in these indicators between EU Member States. The highest shares of cattle farming in agricultural GHG emissions were noted in Luxembourg, Ireland and Slovenia (68–74%), and the lowest shares were in Greece and Bulgaria (16–27%).

The main sources of GHG emissions on dairy farms can be divided into three major categories: enteric fermentation, manure/slurry management, and the production and use of feed, energy and fertiliser (Figure 7). The external system boundaries encompass the production and transport of external feed, water, animals, and energy, which are used at the farm level for breeding and maintaining animals on dairy farms (Figure 8). It also encompasses the production and transport of external fertilisers, pesticides, machines, energy, and ancillary products used for growing crops or maintaining pastures on these farms (for animal feeding). This system boundary therefore refers to indirect emissions. Direct GHG and CF emissions (enteric fermentation; manure and soil management; on-farm/internal feed production and pasture management for cows) within dairy farms represent the internal boundaries of this system. Although enteric fermentation remains the most important part of the CF, the literature data indicate that emissions from other sources can contribute to more than half of the emissions in intensive production systems [22,27,37,38]. The share of individual categories is determined by several factors, including the production intensification level, herd structure, technologies applied, and local determinants [23,39].

According to Singaravadivelan et al. [41], the main CF components include emissions of methane (CH₄) from enteric fermentation (resulting from microbial processes occurring in the rumen), nitrous oxide (N₂O) from fertilisation and manure, and CO₂ from energy generation, feed and fertiliser production, and transport. It is enteric fermentation that remains the largest GHG source in milk production, accounting for as much as 40–80% of the CF [3,12,13,19,28,42,43]. In Brazil, enteric fermentation accounted for 60% [5], in India for 66.8% [4], while in Canada and the USA, for 44% [30,44]. An analysis by Verge et al. [37] found that CH₄ emissions in Canada accounted for approximately 47% of the CF. In comparison, a study by Froldi et al. [3], which examined 61 farms in the Po Valley, reported that methane emissions accounted for 43.3% of the CF.

Velarde-Guillen et al. [43] noted that under grazing systems, the dominant emissions originated from fermentation and fertiliser management, whereas under more intensive systems, they came from the production of feed and mineral fertilisers. Regarding extensive systems, a study conducted in Peru found that the share of enteric fermentation was, on average, 1.81 kg CO₂ eq kg⁻¹ FPCM of the total CF at a level of 2.26 kg CO₂ eq kg⁻¹ FPCM (i.e., approximately 80%) [42]. On low-productivity farms in Kenya, the emissions ranged from 2.19 to 3.13 kg CO₂ eq kg⁻¹ FPCM, of which approximately 54% originated from fermentation [39]. In contrast, in Mongolia (an intensive husbandry system), fermentation accounted for 18% of emissions, with a greater share noted for feed production (27%) and fertiliser management (28%) [38]. These data confirm that the emission source structure is strongly determined by local conditions, production intensity, and feeding systems.

Emissions from fertiliser management (especially manure storage and spreading) remain significant and strongly dependent on the technologies adopted [45,46]. Manure management represents the second key source of emissions on farms. It includes CH₄ and N₂O emissions generated during the storage and application of natural fertilisers. A study by Sorley et al. [22] showed that farms using slurry in uncovered tanks emit significantly more CH₄ and N₂O than those that use roofed systems or process waste/by-products in biogas plants. An analysis by Vida and Tedesco [47] demonstrated that the implementation of anaerobic fermentation reduced emissions from manure by 0.26 kg CO₂ eq kg⁻¹ milk, which translated into a noticeable improvement in environmental performance.

The third dominant emission source includes processes related to the production and supply of feed, especially feed purchased from external sources, and emissions associated with crop fertilisation. In India, the share of emissions related to feed production was 23% [4]. In systems based on a large proportion of purchased feed, indirect (upstream) emissions can account for as much as 30% of the CF [29]. A study by Soteriades et al. [27] showed that the use of HSG can reduce nitrogen (N) emissions from urine and, consequently, N₂O emissions from the soil (which is important in the context of reducing emissions from feeding systems). In Brazil, where intensive feed production is common, the emissions associated with the purchase and transportation of feed ingredients significantly increased the CF [23].

Although enteric fermentation remains the most important emission source in milk production, emissions from manure management and intermediate processes are becoming increasingly important, especially under more complex and intensive production systems [20,21,25]. The optimisation of these three main components, namely fermentation, fertiliser and feed, provides a key direction for an effective reduction in greenhouse gas emissions from the dairy sector [30,38,39]. Many LCA analyses emphasised the significance of indirect emissions, e.g., those related to the production of artificial fertilisers or imported feed components, especially on farms that rely on purchased feed. In tropical countries, land use and landscape transformation are significant sources of emissions.

Indirect GHG emissions, often referred to as off-farm or secondary emissions, constitute a significant component of the total CF of milk production. These emissions originate from processes associated with the manufacture and transport of farm inputs such as synthetic fertilisers, concentrated feed, pesticides, energy, machinery, and replacement animals [9]. LCA studies emphasised that although direct emissions (e.g., enteric methane) dominate, indirect emissions can account for up to 25–30% of the total CF depending on the production system and input intensity [41]. For instance, Sorley et al. [22] observed that housed dairy farms relying heavily on imported concentrates exhibited significantly higher indirect emissions than grazing-based systems, mainly due to the embodied emissions in purchased feed and energy inputs [22]. Similarly, O’Brien et al. [48] highlighted that increased concentrate feeding to boost the milk yield may backfire by raising off-farm emissions and production costs, thereby worsening environmental and economic performance [48]. Studies in mountain and tropical regions, such as those by Sabia et al. [28] and Ruiz-Llontop et al. [42], underscored the variability in indirect emission contributions due to differences in feed production, land use intensity, and the carbon sequestration capacity of grasslands [28,42]. Tools like DairyGHG and simplified calculators (e.g., LatteGHG) allow for estimating these emissions even in data-limited contexts by considering emissions from fertiliser and feed manufacturing [9,24]. Moreover, Silva et al. [5] emphasised the importance of accounting for carbon captured during photosynthesis of forage crops, which may offset a portion of indirect emissions, especially in smallholder and pasture-based systems [5]. Despite methodological challenges and regional heterogeneity, the inclusion of indirect emissions in LCA remains essential to capture the full environmental burden of milk production and to guide mitigation strategies tailored to farm structure and resource dependency [33,41].

3.3. The CF Range in Milk Production

The studies under analysis used various units to express the CF of milk production at the farm level, which, in some ways, complicated direct comparisons. The most commonly used unit represents emissions expressed as kg CO₂ eq kg⁻¹ FPCM, which takes into account the fat and protein contents of milk, thus enabling a better comparison between systems with different milk qualities [5,22]. Many studies, however, used other units, such as kg CO₂ eq kg⁻¹ raw milk [49], kg CO₂ eq liter⁻¹ milk [39], kg CO₂ eq liter⁻¹ FPCM [42], or even kg CO₂ eq 1000 kg⁻¹ FPCM [5]. In addition, the emissions per area unit, such as t CO₂ eq ha⁻¹, were used, which enables the assessment of environmental performance in the context of land use [50]. In certain publications, such as analyses of farms in South America and Africa, emissions were also referred to as livestock units or total net emissions from the farm [2,51]. The heterogeneity of functional units, in terms of both milk density and quality, is one of the main limitations of a comparative analysis. This hinders a direct comparison of the results and requires the methodological context to be taken into account when interpreting the data. Some authors did not use a standardised FPCM conversion factor, resulting in the need to retain the original units in this review.

Based on the analysis of the collected literature, it was determined that the average value of the CF per 1 kg of milk produced on a farm was 1.60 kg CO₂ eq, whereas the 95% confidence interval for the average value ranged from 1.33 to 1.87 kg CO₂ eq (

Table 1. Descriptive statistics for the carbon footprint for 1 kg of milk production.

Variable	Value
N-Valid (pcs.)	24
Mean (kg CO2 eq)	1.60
Confidence Interval −95% (kg CO2 eq)	1.33
Confidence Interval +95% (kg CO2 eq)	1.87
Median (kg CO2 eq)	1.29
Minimum (kg CO2 eq)	0.78
Maximum (kg CO2 eq)	3.20
Lower Quartile (kg CO2 eq)	1.16
Upper Quartile (kg CO2 eq)	2.09
Range (kg CO2 eq)	2.42
Interquartile Range (kg CO2 eq)	0.93
Variance (kg2 CO2 eq)	0.41
Standard Deviations (kg CO2 eq)	0.64
Coefficient of Variation (%)	40.00
Standard Error of Mean (kg CO2 eq)	0.13
Skewness	0.88
Kurtosis	0.01

). Depending on the country, the CF values ranged from 0.78 kg CO₂ eq to 3.20 kg CO₂ eq. The values were characterised by a right-skewed distribution (skewness of 0.88), with the median being lower than the average value and amounting to 1.29 CO₂ eq. A kurtosis close to zero indicated the lack of outliers. The variation coefficient of 40% was at a moderate level; the interquartile range was 0.93 kg CO₂ eq, while the range of variation expressed by the interquartile range was 2.42 kg CO₂ eq.

A study by Camargo et al. [50] estimated the average total CF of milk production on small-scale farms in the Colombian department of Boyacá at a level of 1.8 kg CO₂ eq⁻¹. This value included emissions associated with enteric fermentation (an average of 0.06 kg CO₂ eq⁻¹), manure management (0.21 kg CO₂ eq⁻¹), and soil nitrification (1.53 kg CO₂ eq⁻¹), with the latter component accounting for the largest share of the total CF. The high values were linked to intensive fertilisation and soil management practices. A study by Wei et al. [38] showed that the CF of milk production in Inner Mongolia ranged from 1.5 to 3.4 kg CO₂ eq kg⁻¹ FPCM, with the main emission-differentiating factors including the nutritional strategy, forage availability, and intensification level. A high CF value was noted on farms using mainly dry feed and grains, whereas a lower CF value was observed under systems using silage and grazing. A study by O’Brien et al. [52] determined the CF to be at a level of 1.23 kg CO₂ eq kg⁻¹ FPCM under an intensive system in Ireland, whereas in another study by O’Brien et al. [53], this value for farms with lower productivity was 1.49 kg CO₂ eq kg⁻¹ FPCM. In contrast, Abdallah et al. [54] presented an analysis of 53 farms in Rwanda, where the average emissions amounted to 2.42 kg CO₂ eq litre⁻¹ FPCM, with the main impact factors including local feed availability and slurry management.

The CF of milk production varied significantly depending on the country, production system, farm scale and calculation methodology. In a study conducted on 71 farms in Ireland, Spain, Portugal, the UK and France, the CF values for 1 Mg of FPCM milk ranged from 884 to 2494 kg CO₂ eq (i.e., from 0.88 to 2.49 kg CO₂ eq kg FPCM), depending on the feeding system (grazing, mixed, non-grazing) [22]. The average value for grazing systems was 1.13 kg CO₂ eq kg⁻¹ FPCM. For mixed systems, the value was 1.24 kg CO₂ eq kg⁻¹ FPCM, and for non-grazing systems, it was as high as 1.52 kg CO₂ eq kg⁻¹ FPCM. It is worth noting that the scatter of the results was significant, with both the lowest and highest CF values observed in the zero-grazing system. In Latin America, the CF range was between 1.54 and 3.57 kg CO₂ eq kg⁻¹ FPCM. The emission values were significantly influenced by, e.g., the region of production (tropical vs. temperate), feeding system (zero-grazing, semi-confinement, pasture), and farm profile (dual-purpose vs. specialised) [43]. In addition, the analysis considered the differences in the daily milk yield per cow, which ranged from 8.4 to 20.1 kg day⁻¹, significantly affecting the emission intensity.

In India, for farms producing buffalo milk, the CF was estimated at 2.24 kg CO₂ eq kg⁻¹ FPCM [55], whereas, for cows in the Hisar region in India, the result was 2.13 kg CO₂ eq kg⁻¹ FPCM [49]. In Kenya, the values noted ranged from 2.19 to 3.13 kg CO₂ eq kg⁻¹ FPCM, with the highest CF value observed for systems based solely on grazing [39]. In Brazil, according to a study conducted in São Paulo, the CF values without bioeconomic allocation amounted to 2.41 kg CO₂ eq kg⁻¹, while after allocation, to 2.19 kg CO₂ eq kg⁻¹ FPCM [5]. Comparisons between farms revealed that the integration of net emissions with carbon sequestration balancing can significantly reduce the carbon footprint intensity.

Analyses conducted in various countries and climate zones revealed significant differences in the CF levels of milk production. For example, in Australia, the average CF was 1.11 kg CO₂ eq kg⁻¹ FPCM [14], while according to a study by Froldi et al. [3], in Italy, this value amounted to 1.19 kg CO₂ eq kg⁻¹ FPCM. In tropical countries, such as Ethiopia, the CF can be as high as 3.2 kg CO₂ eq kg⁻¹ FPCM, mainly due to the low productivity and lack of access to modern technologies [14]. In tropical countries (e.g., Peru, Kenya), the CF was typically higher, also due to less intensive production systems, feed seasonality and a higher share of methane fermentation in total emissions [39,42]. In contrast, in the temperate zone (e.g., Europe), higher production efficiency and lower CF values per milk unit were observed, which was associated with better herd management and higher productivity [22]. These differences must be taken into account when developing climate policies tailored to local conditions.

A study by Verge et al. [37], focusing on Canada, showed that the regional diversity of environmental conditions and production practices had a significant impact on the CF value of milk production. The averaged emissions for the entire country were 1.02 kg CO₂ eq kg⁻¹ milk, with the values in individual provinces varying by more than 20%. These differences were influenced by factors such as the length of the grazing season, feed structure/composition, local cows’ milk yield, and management of waste/milk production by-products. In contrast, Hospers et al. [35] demonstrated that in the Netherlands, in the years 1990–2019, the CF for milk dropped by 35%, from 1.52 to 0.99 kg CO₂ eq kg⁻¹ FPCM. This decrease was achieved through increased milk yield, better feed efficiency, and reduced nitrogen fertilisation. In contrast, in the Azores [20], grazing systems were characterised by a low CF value (0.83 kg CO₂ eq kg⁻¹).

The map shows the average CF values for milk production on farms in individual countries around the world (Figure 9), with the highest value noted in Ethiopia and the lowest in New Zealand and the Azores. The CF in other countries ranged from 1 to 3 kg CO₂ eq kg⁻¹ milk.

Analysis of the CF values by continent (Figure 10) showed that the highest average values were noted in Africa (2.81 kg CO₂ eq kg⁻¹), followed by Asia (2.29 kg CO₂ eq kg⁻¹) and South America (2.05 kg CO₂ eq kg⁻¹), whereas the lowest values were noted in Europe (1.33 kg CO₂ eq kg⁻¹), North America (1.27 kg CO₂ eq kg⁻¹) and Oceania (0.98 kg CO₂ eq kg⁻¹). Single results were obtained for Latin America (2.56 kg CO₂ eq kg⁻¹) and Australia (1.1 kg CO₂ eq kg⁻¹). The highest variability was observed in North America (37%), Oceania (29%) and Europe (24%), and the lowest was in Asia (10%).

3.4. Effect of Feed, Its Origin and the Feeding System on the CF

The feed type and origin have a significant effect on the total CF of milk production. Locally produced feed (e.g., grasses, alfalfa and hay) tends to exhibit a lower carbon footprint value than commercial feed, especially imported soybean or ground grain [50]. Transport, agrotechnical measures, feed processing and fertilisation affect, to varying degrees, the total GHG emissions, including potentially increasing them. Under zero-grazing systems, the purchase of high-protein feed can account for a considerable proportion of the total GHG emissions from a farm. Optimisation models have demonstrated that producing grass silage on the farm can reduce the CF by as much as 15% [2]. In the LCA for milk, the above-mentioned purchase of feed, especially protein feed (e.g., soybean meal), is among the major indirect emission sources. In addition, Froldi et al. [3] and Garcia-Souto et al. [19] emphasised that feed imports from outside the UK, especially from areas undergoing deforestation, can significantly increase the CF value through the so-called LUC-related emissions. These factors were responsible for the significant difference between grazing systems and intensive systems [19]. In a study by Sabia et al. [28], the higher share of concentrated feed was associated with a lower CF but also with a greater adverse impact on biodiversity. In contrast, pasture-based systems had a higher CF but exerted a more beneficial effect on ecosystems. Therefore, the choice of a feed source requires a compromise between reducing emissions and protecting the environment/nature.

In addition, the choice of a feeding system clearly has an effect on the GHG emissions per milk unit. A study by Garcia-Souto et al. [19] demonstrated that pasture-based feeding systems generated a significantly lower CF value (507–528 g CO₂ eq kg⁻¹ FPCM), as compared with intensive systems involving total mixed ration (TMR) feeding (724–764 g CO₂ eq kg⁻¹ FPCM). Importantly, the authors emphasised the impact of imported soybean meal, whose high LUC index significantly increases the emissions in intensive systems. In another study, a grazing system was also characterised by the lowest CF per kg of milk (1.13 kg CO₂ eq kg⁻¹ FPCM), whereas non-grazing systems achieved the highest values (1.52 kg CO₂ eq kg⁻¹ FPCM) [22]. Under these systems, cows were fed forage and concentrated feed with higher emission intensity, e.g., from silage maize, soybean meal and other feed raw materials requiring transport and storage. Under intensive systems in India, mixtures of concentrated feed and hay silage were dominant, which was associated with a higher CF value [55].

A study by Uddin et al. [29] demonstrated that a high-forage diet resulted in a lower CF value for milk than a low-forage diet, by 1.35 vs. 1.49 kg CO₂ eq kg⁻¹ FPCM, respectively. These differences were mainly due to the higher share of emissions from manure management noted in systems based on a higher proportion of maize silage. O’Brien et al. [53] demonstrated that farms in Ireland, which used grazing systems, obtained a lower CF value (0.84 kg CO₂ eq kg⁻¹ energy-corrected milk (ECM)) than that on farms based on closed systems in the United Kingdom (0.88 kg) and the USA (0.90 kg). These differences were due, among other things, to the limited use of fertilisers and concentrated feed, as well as the C sequestration capacity of permanent grassland. Rotz et al. [9] emphasised that the transition from a closed system to seasonal grazing reduced the total GHG emissions by 9%, primarily due to reduced emissions from manure storage as well as reduced energy consumption. It should be noted, however, that N₂O emissions can increase as a result of direct deposition of nitrogen from urine on the pasture. A study by Soteriades et al. [25] demonstrated that the introduction of pastures with HSG reduced the acidification and eutrophication potential by 11% and 6%, respectively, as compared with traditional grassland. This indirectly translates into a reduction in the CF, especially in the longer term and with improved manure management.

In contrast, in Latin America, Velarde-Guillen et al. [43] observed no apparent differences in the CF between grazing and intensive systems, despite zero-grazing systems being more productive in terms of the milk yield. This indicates that feed efficiency and local conditions play a role that is equally important to that of the type of feed itself. A study conducted in the Peruvian Amazon by Ruiz-Llontop et al. [42] demonstrated that the silvopastoral system, despite the limited proportion of external feed, was characterised by a relatively high CF value (2.26 kg CO₂ eq kg⁻¹ FPCM), of which as much as 1.81 kg CO₂ eq kg⁻¹ FPCM originated from enteric fermentation. This means that even under grazing systems, feed quality and herd management are of crucial importance.

Under alpine conditions, Sabia et al. [28] observed that systems with a higher proportion of BSHC concentrated feed (Brown Swiss high concentrate) showed a lower CF value (1.14 kg CO₂ eq kg⁻¹ FPCM) than under systems with a low proportion of AGLC concentrated feed (Alpine Grey low concentrate, 1.55 kg CO₂ eq kg⁻¹ FPCM), with the latter systems, however, having a considerably lower impact on biodiversity. This suggests that the intensity of feeding alone is not the only factor affecting the environmental balance. Silva et al. [5] noted that farms with lower feed consumption and good milk yield achieved lower CF values, even under tropical conditions. In that study, CH₄ from enteric fermentation accounted for 60% of emissions, and the variability in the amount of dry feed offered per kg of FPCM milk was a significant CF-differentiating factor.

In summary, the feeding system affects the CF of milk production in many ways: through the diet composition, feed and fertiliser use intensity, feed quality, and manure management method. Although grazing systems often generate a lower CF, their efficiency is determined by the quality of available feed and production efficiency [9,29,53]. In contrast, intensive systems, despite higher emissions associated with feed and fertiliser, can exhibit a lower unit CF at high productivity if accompanied by efficient management and proper feed ration planning [5,28]. Therefore, the assessment of the impact of the feeding system on GHG emissions should be integrated into the analysis of the entire product life cycle, taking into account local conditions, the quality of feed raw materials, and their origin [27,41,43].

3.5. Production Efficiency and CF

Milk production efficiency is among the crucial factors affecting the levels of GHG emissions per product unit. A study by Camargo et al. [50] demonstrated a strong correlation between the gross livestock unit (GLU) ha⁻¹ and GHG emissions, especially N₂O emissions originating from soil nitrification. Importantly, farms with a higher proportion of lactating cows and a higher milk yield showed a significantly lower CF value. The average milk production in the group of farms under study was 14,903 kg year⁻¹, ranging from 5040 to 69,960 kg year⁻¹, and the average milk yield per lactating cow reached 2588 kg year⁻¹. The stocking density was, on average, 2.37 GLU ha⁻¹, which also affected the levels of emissions from an area unit.

According to the results obtained by O’Brien et al. [48], Irish farms with the highest production efficiency, understood as the amount of milk per cow, not only achieved the lowest CF value but also the highest net income per unit of CO₂ eq emissions. Jayasundara et al. [44] also confirmed that a higher milk yield was significantly correlated with a lower CF value (correlation = −0.43) and higher production profitability (0.83). On Canadian farms, the average CF was 1.02 kg CO₂ eq kg⁻¹ FPCM. However, farms with higher productivity, shorter calving intervals and improved herd management achieved significantly lower emission levels.

Analyses conducted in Mongolia by Wei et al. [38] showed that more intensive farms with higher management efficiency, characterised by a lower age at first calving, better feed quality and a higher energy efficiency index, exhibited lower CF values. There was a strong negative correlation between the CF and the FPCM 100 kg⁻¹ cow’s body weight. Mech et al. [4], when analysing production systems in India, confirmed that farms with a lower milk yield (below 3500 kg FPCM year⁻¹ per cow) showed a CF value over two-fold higher than farms with a higher milk yield. The importance of the first calving, shortening of sterile periods, and optimisation of feeding was also noted in terms of emission-reducing measures.

In New Zealand, Ledgard et al. [21] analysed data from over 7000 dairy farms and demonstrated that the CF was, on average, 15% higher on farms with a milk yield exceeding 6000 kg FPCM cow⁻¹ year⁻¹ as compared with farms with a milk yield below 3540 kg. The impact of the milk yield enabled the “dilution” of the GHG emissions and the CF in relation to the product unit. Similar conclusions were drawn in Ethiopia, where, as indicated by Feyissa et al. [51], over 50% of the CF variation between farms was directly attributable to differences in the milk yield. Wilkes et al. [39], when investigating systems in Kenya, stressed that farms with feed efficiency higher than 0.34 kg⁻¹ FPCM kg⁻¹ dry feed matter achieved a significantly lower CF value than farms with lower efficiency. It should be noted, however, that increasing the supply of concentrated feed without its optimal use can lead to an increase in emissions, both direct and indirect.

Singaravadivelan et al. [41] showed that an increase in the milk yield from 2000 to 5000 kg FPCM cow⁻¹ year⁻¹ resulted in a significant drop in the CF value. This effect, however, flattened out after exceeding the threshold of 6000 kg, indicating the existence of a limit beyond which further intensification did not bring proportional environmental benefits. It was pointed out that, with a high milk yield, the benefits of diluting the unit emissions were limited by indirect emissions from the production of feed and fertiliser and energy consumption. Mech et al. [4] also demonstrated that farms with a milk yield below 3500 kg cow⁻¹ year⁻¹ showed a CF value over two-fold higher than that on farms achieving over 5000 kg. This difference resulted from low feed digestibility, poor feed ration balancing, and high nitrogen losses. The authors indicated the significance of integrating crop production with animal production as a way to reduce emissions associated with the purchase of external feed components. Morais et al. [20] concluded that on grazing farms in the Azores, despite being less intensified, the CF value was relatively low (0.83 kg CO₂ eq kg⁻¹ of milk), which was explained by good feed quality and a high proportion of grazing. A study by Hospers et al. [35] proved that farms which increased their productivity while reducing mineral fertilisation and the use of feed from external sources were able to achieve better environmental and economic results at the same time.

Numerous data from various regions of the world showed that increasing the milk yield while maintaining or improving the standards of herd, feed, and reproductive cycle management is one of the most effective tools for reducing the CF [4,21,38,44,48,51]. However, an increase in productivity does not always translate into further reductions in the CF, especially when accompanied by the intensification of resource-intensive practices. Sustainable intensification, taking into account indirect emissions and local production conditions, appears to be crucial for the future of low-carbon dairy farming [35,41,50].

3.6. Farm Scale and Its Characteristics

The scale of a farm and its organisation and integration levels are significant determinants of the GHG emission levels per milk unit. It has been observed in many regions of the world that larger, better-organised production units achieve a lower CF value as compared with small, low-productivity farms. In Turkey, according to Aydin and Koknaroglu [56], the CF for small dairy farms was, on average, 1.35 kg CO₂ eq kg⁻¹ FPCM, whereas for large mechanised farms it was lower, amounting to 1.13 kg CO₂ eq kg⁻¹ FPCM. Similar trends were observed in Brazil, where larger farms utilising modern technologies and improved management achieved lower CF values than smaller traditional farms [23].

In Central America, Gonzalez-Quintero et al. [57] demonstrated that farms with a greater production scale, which used technical advice and applied more efficient fertilisation and feed management systems, achieved noticeably lower CF values as compared with farms unsupported by advisory systems. Moreover, access to expert knowledge enables the adaptation of emission-reducing practices. In a subsequent publication, the same authors [17] noted that smaller farms were characterised by higher sustainability levels when their activities included additional environmental functions, e.g., agroecological cultivation, water retention or local production of organic fertilisers.

A study by Garg et al. [15] conducted in India indicated the complexity of assessing the impact of small farms on the climate. The average herd size in that study was only four head of cattle, and the initially measured CF for milk was 2.2 kg CO₂ eq kg⁻¹ FPCM. However, after considering manure production and its role in closing the nutrient cycle, the corrected CF value dropped to 1.7 kg CO₂ eq kg⁻¹ FPCM. This shows that the environmental assessment of multifunctional farms requires a systemic approach, taking into account ecosystem services and non-production benefits.

The reason for the higher CF of small farms is, usually, lower feed conversion efficiency and a higher share of emissions from purchases of high-protein feed, often imported and affected by additional emissions associated with transport and the LUC index [3]. Additionally, these farms are often characterised by lower fertilisation management levels and higher N losses into the environment. In contrast, Gollnow et al. [14] emphasised that the feed conversion efficiency is one of the most important emission-reducing factors, irrespective of the production scale. Better feed utilisation results in lower CH₄ emissions and lower amounts of manure requiring further treatment.

Another significant factor differentiating farms is the level of integration of the production system, which is understood as feed self-sufficiency, production of organic fertilisers, and internal closure of the nutrient cycle. Studies by Gross et al. [45] and Eisert et al. [46] demonstrated that farms applying manure composting, using renewable energy sources (RES), e.g., from biogas plants, and using pastures in a rotational manner showed lower total CF values compared with farms relying on external energy and feed sources. An interesting observation was also presented by Guerci et al. [18], who highlighted the importance of social factors, such as the level of farmers’ education, access to advisory services, and the level of work organisation, as indirect CF determinants. An analysis of Italian systems showed that family farms, despite their smaller scale, often achieve better environmental results thanks to their greater commitment and flexibility in terms of decision-making. In turn, Feyissa et al. [51] pointed out that smaller farms in Ethiopia were characterised by a high CF variability determined by the level of local resource utilisation and labour efficiency. Farms capable of producing their own feed and organic fertiliser and managing local water resources showed lower emissions than farms fully dependent on external markets. This was also confirmed by the data provided by Garcia-Souto et al. [19], indicating that a highly organised grazing system achieved lower CF levels than intensive TMR systems in Ireland and Spain.

Economic factors are also not insignificant. Flysjö et al. [1] documented that farms which are better managed, in economic terms, are more inclined to implement emission-reducing solutions, such as feed additives, efficient herd management, and energy-saving technologies. Although the farm scale correlates with the CF, it is often not the size but the management quality and the system integration level that prove to be the decisive factors.

In summary, larger farms often achieve lower CF values thanks to the specific scale effect, improved technology, and higher productivity [4,19,23,56]. This does not mean, however, that smaller farms are, by definition, less sustainable. For multifunctional farms, locally integrated and based on expert knowledge, it is possible to achieve a very good environmental profile [15,17,18]. Not only intensification but also its quality and direction, including the pursuit of a closed loop, local sourcing and resource efficiency, remain key challenges [1,45,46].

3.7. GHG Emission and CF Reduction Methods, Technological Innovations, and Future Directions

The results of a study by Camargo et al. [50] indicated the particular importance of proper soil and manure management as a strategy for reducing the CF. The same study also demonstrated that improving the proportion of lactating cows and reducing excessive stocking densities can effectively reduce emissions. The potential for implementing silvopastoral systems as a form of adaptation to climate change, improving carbon (C) sequestration in the soil, and reducing CH₄ emissions was also noted. A study by Gonzales Quintero [57] showed that the implementation of a silvopastoral system can reduce the total CF for milk by up to 40%, primarily due to C sequestration and improved feeding efficiency. In addition, Gonzales-Quintero [17] analysed three types of farms in Colombia and documented that the farms most integrated with the silvopastoral system had a 20–39% lower CF value than conventional farms. However, a study conducted in Costa Rica demonstrated that the use of trees in the form of hedges can reduce the CF of milk production by 21–37% through C sequestration [2]. LCA models showed that considering CO₂ absorption by plant biomass can significantly affect the ultimate environmental assessment of milk production. Increasing attention is also being paid to sustainable organic fertilisation and the optimisation of mineral fertiliser rates and water and waste management at the farm level.

Technological advances and growing environmental requirements are driving the development of new solutions aimed at reducing emissions in milk production. Among the studies analysed, the following methods were most frequently indicated as ways of reducing the GHG emissions and the CF: improving feed digestibility and quality; reducing the age at first calving; shortening the calving interval; automating the feeding and milking process; selecting cows for environmental performance (e.g., lower CH₄ emissions); using digital systems for production and emissions management; implementation of the silvopastoral system; utilisation of RES (e.g., biogas plants or photovoltaics); and the use of feed additives (e.g., essential oils—EOs) [2,24,47]. It was demonstrated that the addition of EO to feed can reduce the CH₄ emissions from enteric fermentation by as much as 8.8% while increasing the milk yield [24]. A study by Eisert et al. [46] noted that future CF reduction strategies should focus on precise adjustment of the feeding system to local conditions and cow genotypes. Gross et al. [45] showed that the transition from conventional production to organic production can reduce the CF by 9%, yet the increase in CH₄ emissions from enteric fermentation requires compensatory effects to be taken into account. This indicates the need to implement technologies that reduce CH₄ emissions (e.g., feed additives) and provide more accurate emission data for organic systems.

Technologies are being developed to enable CF assessment in real time and integrate environmental data into farms’ economic accounts. In the longer term, financial support policies for the implementation of green technologies will also be important. The development of IT tools for emission modelling is an important direction for supporting environmental decisions in dairy farms. The DairyGHG model, presented by Rotz et al. [9], integrates greenhouse gas sources and sinks at the entire farm level, thereby enabling the estimation of the CF of milk production while taking into account different nutritional strategies, stocking densities and fertiliser management. In turn, Pirlo and Care [26] proposed a simplified calculation tool, LatteGHG, which enables quick estimation of the effect of various factors (productivity, stocking density, manure management system) in typical Italian dairy farms. The development of intuitive digital models adapted to local conditions can facilitate the implementation of sustainable practices not only in large farms but also in small dairy farms, including family farms.

4. Conclusions

Determining the carbon footprint of milk production in agricultural farms is a new challenge for the entire dairy sector, which is arousing a variety of emotions. This concerns all participants in the process, from farms through dairies and shops selling dairy products to consumers of dairy products and plants disposing of post-production residues and waste. Many countries have already taken the initial steps in this direction, while others are still in the early stages of this process. It is worth noting, however, that EU guidelines and regulations in this area set specific directions and deadlines for action. Moreover, customer expectations appear to be consistent in this regard. Therefore, there is an urgent need to investigate and verify this phenomenon under various environmental and production conditions. This is important for the reliable assessment of GHG emissions associated with milk production on dairy farms while taking into account carbon sequestration in soils, especially on grassland.

The content presented in this review paper clearly shows that the CF of milk production at the farm level is the result of a complex interaction of many technological, environmental and organisational factors. Not only direct CH₄ emissions from enteric fermentation and fertiliser management but also indirect emissions associated with the production and transport of feed and fertiliser, as well as energy consumption, are of key importance here. It was demonstrated that the choice of feeding system, feed origin (especially the proportion of imported high-protein components), feed conversion efficiency, and farm organisation have a fundamental impact on the CF levels per unit of milk produced at the farm level. At the same time, data from various countries and regions indicate very high variability in the CF values, including in favour of larger, integrated farms and grazing systems relying on local feed resources. On the other hand, intensifying milk production enables, to a certain extent, a reduction in the unit carbon footprint thanks to the emission dilution effect. However, once specified intensification thresholds are exceeded, further increases in productivity no longer lead to proportional environmental benefits. Another important conclusion is that both the calculation methodology (the choice of a functional unit, the scope of the system, and the emission allocation method) and the local climatic and organisational determinants strongly influence the final CF assessment result. This study also stresses that measures aimed at reducing the CF must always be tailored to the specific nature of a particular farm and region, with examples including improving the feed conversion efficiency, increasing the proportion of home-grown and green fodder, investing in biogas plants, and implementing innovations in fertiliser management. Ultimately, effective decarbonisation or GHG emission reduction in the dairy sector requires an integrated approach that combines LCA tools, local environmental and economic determinants, and the active implementation of good agricultural practices, such as those presented in the present manuscript.