Environmental Sustainability of Dairy Cattle in Pasture-Based Systems vs. Confined Systems

Abstract

The aim of this paper is to review the literature on the environmental impacts of pasture-based dairy cattle systems, focusing on the factors affecting the main impact categories. This paper also aimed at comparing data of the literature on environmental impacts in pasture-based vs. confined systems. The environmental impact of pasture-based dairy cattle systems appears to be highly influenced by several input factors. Life cycle assessments have shown significant variability in methodological approaches, which complicates the comparison of results across studies. The different variables affecting environmental impacts make it challenging to draw universally valid conclusions regarding the comparison of pasture-based and confined dairy systems on a global scale. In addition, the analysis of the variables highlights the considerable potential to reduce the environmental impact of milk production in both systems by adopting productivity-enhancing activities, low inputs and best management practices. Further aspects such as geographical factors, carbon sequestration, animal health and welfare, toxicological aspects due to the use of drugs and antimicrobials in animals and the maintenance of local animal breeds should be incorporated into LCAs for a full comprehensive understanding of the environmental impacts of dairy farms.

1. Introduction

Livestock play a key role in providing the essential nutrition and livelihoods of many communities worldwide. Today, the livestock sector is facing the huge challenge of meeting the ever-increasing demand for food, due to the growing global population which has been estimated to reach 9.7 billion people by the year 2050. Consequently, the global demand for animal protein in 2050 is expected to increase by 21% compared to 2020 [1].

Simultaneously, the livestock sector has to deal with the adverse effects of climate change, the overexploitation of natural and fossil resources and the loss of biodiversity, which means that pursuing global sustainability of the agrifood sector is essential for safeguarding the planet and its inhabitants.

To be sustainable, livestock farming has to be able to meet the needs of current and future generations, ensuring economic and social equity and environmental health. However, if not managed properly, livestock systems can contribute to global warming because of the greenhouse gas (GHG) emissions generated throughout the production chain [2].

With a value of approximately 3.8 Gt CO₂ eq per year, cattle are among the main livestock contributors to GHG emissions, accounting for approximately 62% of overall livestock emissions [2]. In fact, the dairy cow sector has been found to be responsible for 30% of these emissions [3].

However, milk represents a vital food resource for millions of people globally, and it is considered indispensable for human health due to its irreplaceable contribution to a nutritious and healthy diet [1,2].

An improvement in the efficiency of the bovine milk production chain is therefore needed to reduce the environmental impacts. Feeding systems and manure management are among the most important factors that are crucial for dairy farm efficiency and sustainability [4]. Feeding systems and manure management are responsible for the wide diversification of dairy cattle farming systems worldwide: from nomadic farming and small herds to intensive confined systems with thousands of animals [5]. Pasture-based dairy production is one of the most common farming systems and is important in the agricultural industry in middle- and high-income countries [6], but also in low-income regions of the planet [4].

Pasture-based dairy cow systems vary from extensive annual grazing systems to rotational grazing types, and transhumance in mountain pastures for the summer period only. Grazing is characterized by different traits such as duration, stocking rate (number of heads per ha), animal breeds, inputs and productivity, and the different traits are based on the pedo-climatic characteristics of the particular geographical area which may differ substantially.

A life cycle assessment (LCA) is the most accredited and reliable ISO standardized methodology used to evaluate the environmental impacts associated with all stages of the life cycle of a product or service, from the production of raw materials to the final disposal. This method is divided into four main phases: (1) definition of the goal and scope—establishing the purpose of the analysis, system boundaries and functional unit, (2) life cycle inventory—collection of all the inputs (materials, water and energy) and outputs (emissions and waste) involved in the production process, (3) life cycle impact assessment—translation of the inventory data into potential environmental impacts and (4) interpretation—analysis of the results and conclusions drawn [7].

The system boundary identifies which processes and life cycle stages are included in the analysis [8]. The functional unit (FU) is a quantitative measure of the function of the system and serves as a reference for comparing results [8]. The recommended FU for the analysis of milk production systems is 1 kg of fat and protein corrected milk (FPCM), which provides a more homogeneous comparison between different types of milk and farming practices, focusing on production efficiency [8,9]. However, some authors consider the environmental impact per kg of energy corrected milk (ECM) or per area of land used, taking hectares (ha) as the FU. In the latter case, the analysis focuses on land use efficiency, which may be useful for understanding how different land management practices and stocking densities affect the environment [10,11,12].

LCAs are increasingly used for estimating and comparing the environmental impacts of agricultural products [13]. Despite some limitations [14,15,16], in 2003, the European Commission’s Integrated Product Policy Communication identified the LCA as the “best framework for assessing the potential environmental impacts of products”.

The aim of this paper is to review the literature on the environmental impacts of pasture-based dairy cattle systems, focusing on the factors affecting the main impact categories. This paper is also aimed at comparing data of the literature on environmental impacts in pasture-based vs. confined systems.

2. LCA Methodology and Related Methodological Issues

The literature search of scientific international papers was performed using Web of Science (https://clarivate.com/webofsciencegroup/solutions/web-of-science/ accessed on 4 February 2023), Science Direct (https://www.sciencedirect.com/) and Scopus (https://scopus.com/) databases. The search was carried out using all combinations of the following keywords: pasture-based; life cycle assessment; LCA; dairy cattle; environmental sustainability and milk.

In addition, a search on the same search engines using the impact category name was also performed, i.e., climate change (CC), eutrophication (EP), acidification (AP), land use (LU), water use (WU) and non-renewable energy use (NRE). The latter is the most relevant impact used for the livestock sector [8]. The time range chosen for the published articles was the last 16 years (2009–2024). Older publications may have been used as references for definitions and concepts, but not for LCA data. Papers published after December 2024 are not included in this review.

The selection criteria used to select the literature are the following:

• Studies concerning the environmental impacts of cow’s milk production in pasture-based systems;
• Studies comparing cow’s milk production in pasture-based and a confined/semi-confined system.
• Studies using a LCA based approach to define the environmental impact of the farms. For the paragraph on biodiversity, the selection criteria were slightly different, as there was a lack of sufficient LCA studies to formulate a relevant hypothesis. We therefore used only selection criteria (1) and (2).

Since, in LCA studies on milk production, the choice of system boundary can influence the results, we have selected only the literature that use an approach from the cradle-to-farm gate.

Furthermore, the studies were organized in tables by impact category. To improve the comparability of the results and make the results more readable, we have grouped the LCA studies according to the functional units. For each paper, the following characteristics were reported in the tables: values of impact category in pasture-based vs. confined system, functional unit, number of farms analyzed, breed, country, year and authors. For climate change impact, an additional column was added to specify if ecosystem services and carbon sequestration were taken into account.

Some limitations of the LCA approach are the several different guidelines [8,9,14,16,17] which have been used by several authors to estimate the environmental impacts of dairy cattle farming systems.

Different functional units are often used in different studies, which can significantly influence the results obtained, particularly with regard to the allocation methods [18,19]. This is particularly evident for CC, in which the choice of the functional unit led to variations that can reach as much as 1.6% [20,21], but it can also apply to other impact categories.

CC can be defined as GHGs, expressed in terms of CO₂ eq, emitted into the atmosphere by an individual, organization, process, product or event from within a specified boundary [22].

Regarding the allocation methods, Kiefer et al. [18] reported that the CC results varied depending on the type of allocation (coefficient of variation 0.13–0.16% for intensive farms; between 0.16% and 0.44% in extensive systems) (calculated values from [18]).

Different LCA studies are also heterogeneous with regard to data acquisition, with some authors collecting primary data in the field, and others using national databases. In terms of climate change, not all papers include carbon sequestration in the LCA [6,23,24,25].

Carbon sequestration is the process of adsorbing and storing atmospheric CO₂ in the soil. Pasturelands have a significant impact on the overall carbon cycle fluxes, and good grazing management can lead to higher soil carbon concentrations than non-grazing systems [26,27]. In the literature, there is a poorly defined methodology to quantify carbon sequestration and not considering carbon sequestration in the LCA would overestimate the CC values of grazing farms by 10% [23] to 28–48% [25,28].

The provision of ecosystem services (ESs) is another decisive factor in the definition of CC [18]. ESs are defined as the benefits provided by ecosystems to humans, which typically include both market and non-market goods, such as the management of renewable natural resources, socioeconomic viability of many natural areas and the preservation of landscapes [29]; however, there is a lack of a single methodology to integrate ecosystem services in LCA.

When comparing different allocation methods, Kiefer at al. [18] considered ESs in their LCA study using a specific economic allocation method and demonstrated CC reductions between 20 and 46% in pasture-based (PB) farms and between 19 and 27% in intensive farms.

Other methodological variability factors were highlighted by Lorenz et al. [22]. These regarded the choice of climate gas metrics, equations for enteric methane prediction, emission factors for nitrous oxide emissions, variability in manure management and feed ration-related factors. For example, pasture and concentrate intake have been considered in different studies as percentages of total DMI, whereas the different DMI capacities of small and large dairy cows were not taken into account. According to Lorenz et al. [22], the feed intake per unit of metabolic weight should be considered since in pasture-based systems, the farmer’s choice is to use lighter cows such as small-framed Holsteins and Jerseys.

Some authors have found that the IPCC equation underestimates the N emissions associated with manure management compared to the emissions actually measured, leading to a 10–42% lower global warming potential [30]. The differences in the characterization factors used in the LCA calculation methods may also be responsible for the different values resulting from the calculation of eutrophication [12], thus increasing the complexity of harmonizing the results obtained.

The lack of a standard approach for determining the water footprint has led to the use of multiple impact assessments in the studies reviewed, considering different parameters. A limited number of studies have evaluated the effects of different methods on the quantification of water footprints of PB dairy farming.

As highlighted by Damiani et al. [31], several methods for biodiversity impact assessments have been developed; however, no method currently considers the variety of pressures on biodiversity, ecosystems, taxonomic groups and essential biodiversity classes.

3. Climate Change

Although many different factors can influence climate change (CC) in PB dairy farms, the main contributors are animal enteric fermentation and manure management [32,33]. The combined contribution of enteric and storage emissions to CC (mainly in the form of CH₄) ranges from 44.1% to 65.9% [34].

The CC values of PB dairy cow systems range between a minimum reported of 0.55 to a maximum of 1.43 kg CO₂ eq per kg of energy corrected milk (ECM) [34].

When considering fat and protein corrected milk (FPCM), the CC values of PB dairy cow systems in the literature vary between 0.60 and 2.38 kg CO₂ eq per kg of FPCM [18,25].

In PB farms, Guerci et al. [34] found the lowest impact in an organic Danish farm and the highest impact in a German farm where the cows spend around 260 days/year on pasture. The two PB farms had similar production levels, similar feed self-sufficiency, and differed, above all, in the percentage of grazing: 71% of cow DMI was pasture in the German farm vs. 22% DMI in the Danish farm. In addition, feed efficiency was higher in the Danish farm (1.43 vs. 1.34 Kg ECM/kg DMI per cow). The lower incidence of concentrates in the ration may have influenced the poorer feed efficiency and environmental performance of the German farm.

Other authors have found that better feed efficiency is an important factor related to a lower CC value [33,35], as well as milk yield [4,21] and pasture quality [25,32].

Regarding pasture quality, most studies agree that the poor quality of the grass decreases the feed conversion efficiency, thus decreasing the milk yield with negative repercussions on the CC values [4,32].

Several studies have also shown that the intensification of the pasture [4,36] and the length of the grazing season [12,32] have the potential to influence the CC values of grazing farms.

The effect of the length of the grazing season can contribute to lower CC values as observed in New Zealand farms where animals graze 350 days a year (0.8 kg CO₂ eq per kg FPCM) [37]. On the other hand, higher values were recorded in Irish pasture-based dairy systems [6] (0.9–1.72 kg CO₂ eq per kg FPCM) where during periods of adverse weather, dairy cows are kept indoors.

The effect of grazing length may depend on the latitude and is evident in some studies. For example, Guerci et al. [38] compared traditional farms in the Italian Alps with and without traditional summer grazing. They observed that although summer grazing has some positive effects (such as reducing requirements for concentrates for manure storage facilities, silage panels and welfare devices in the barns), it is related to lower feed efficiencies and milk yields which made it disadvantageous in terms of CC. In fact, a summer grazing season led to higher CC values compared to non-grazing farms (1.72 vs. 1.55 kg CO₂ eq per kg FPCM) [38].

This finding is in line with Aguirre-Villegas et al. [39] and suggests that the environmental benefits from grazing are small in regions that cannot support continuous grazing.

Other farm characteristics such as stocking rate, on-farm feed use and farm N efficiency [6,24] seem to be negatively correlated with CC.

In some marginal contexts, some authors suggest that using well adaptable dual-purpose rustic breeds [36,40] can reduce the use of extra farm concentrates.

Regarding milk yield in PB systems, the increase in milk yield has been found to reduce CC values [21,41], the extent of which depends on the level of productivity and breed. Christie et al. [42] and Lorenz et al. [22] reported more consistent reductions in CC values up to 6000 kg FPCM per cow at lactation (by 0.12 kg CO₂ eq per kg FPCM), and for subsequent increases in milk yield, the decreases in CC values slowed down.

In contrast, age at first calving has been positively correlated to CC [35], this could be due to the lengthening of the cow’s unproductive period.

Indeed, another factor which has been shown to affect farm emissions is the length of the productive life of dairy cows, which has gained interest as a potential GHG mitigation option [43]. In fact, emission intensity and profitability have been described as the most favorable in cows with a long productive life. On the other hand, cows that had not finished their first lactation performed poorly regarding their emissions per unit of product, and rearing costs were rarely recovered [43].

Some authors have reported that the replacement rate contributes to decreasing the impacts of milk [41,44], while others found no significant effects [21].

According to Chobtang et al. [37], an increase in the intensification of PB dairy farming systems can lead to increased milk production per cow and per ha. However, this can also result in increased CC values (from 0.73 to 0.86 kg CO₂ eq per kg of FPCM) because of the major use of inputs, such as the greater use of chemical fertilizers and concentrates, which are among the main off-farm emission sources in dairy cattle livestock systems [45].

When PB systems are compared to confined or semi-confined (CoSC) systems, inconsistent results from the different studies have been reported (

Table 1. Climate change, potential expressed as kg CO₂ eq in pasture-based compared to confined or semi-confined dairy cattle systems.

CC of PB	CC of CoSC	FU	Inclusion of CS or ES in the Assessment	No. of Analyzed Farms	Cattle Breeds	Country	Reference
1.00 kg CO2 eq	1.16 kg CO2 eq	kg ECM	Not included	na	na	New Zealand and Sweden	[46]
0.88 kg CO2 eq	0.87 kg CO2 eq	kg ECM	Not included CS	2	Holstein; Friesian; Holstein × Jersey and Swedish red cross-breeds	Georgia	[23]
0.79 kg CO2 eq	0.87 kg CO2 eq
0.55–1.43 kg CO2 eq	1.11–1.91 kg CO2 eq	kg ECM	Not included	32	na	Italy, Denmark and New Zealand	[34]
0.91 kg CO2 eq	0.90 kg CO2 eq	kg ECM	Not included CS	3	Holstein–Friesian	Ireland, United Kingdom and United States	[6]
0.84 kg CO2 eq	0.88–0.90 kg CO2 eq
0.91 kg CO2 eq	1.09 kg CO2 eq	kg ECM	Not included	2	Holstein–Friesian	United Kingdom	[47]
1.01 kg CO2 eq	0.93 kg CO2 eq	kg ECM	Not included	3	Holstein; Jersey	Brazil	[48]
0.87 kg CO2 eq	1.03 kg CO2 eq	kg FPCM	Not included	2	Holstein–Friesian	Ireland	[10]
1.47–2.38 kg CO2 eq	1.51–2.06 kg CO2 eq	kg FPCM	Not included ES	113	Holstein; Fleckvieh; Vorderwälder	Germany	[18]
1.35–1.70 kg CO2 eq	1.49–2.03 kg CO2 eq
0.99 kg CO2 eq	1.22 kg CO2 eq	kg FPCM	Not included	2	Holstein	Spain	[32]
2.02 kg CO2 eq	1.91–2.39 kg CO2 eq	kg FPCM	Not included	70	na	Italy	[49]
1.13 kg CO2 eq	1.24–1.52 kg CO2 eq	kg FPCM	-	71	na	Western Europe	[35]

CC: Climate change potential; FU: functional unit; PB: pasture-based dairy cattle farm systems; CoSC: semi-confined dairy cattle farm systems; ECM: energy corrected milk; FPCM: fat and protein corrected milk; CS: Carbon sequestration; ES: ecosystem services; na: not available.

).

The non-standardized data collected in this review are very heterogeneous in terms of CC values. For example, CoSC systems vary from 0.87 to 1.91 kg CO₂ eq per kg ECM [23,34] and from 0.96 to 2.39 kg CO₂ eq per kg FPCM [49].

The authors, who directly compared the two systems, found that the CC values of CoSC systems were between 4 and 50% higher than PB systems [34,47] considering kg ECM, and 3–34% higher, using kg FPCM as the FU [18,35].

Regarding the ha as a functional unit, the CC values will depend on the concentration of animals per hectare which are generally higher in CoSC farms, as observed by Sorley et al. [35]. Intensification invariably led to increased emissions when expressed on an area basis [50].

The LCA study by O’Brien et al. [10] which directly compared the environmental impact of a PB based and a confinement dairy farm reported higher CC values in the confinement dairy system due to its greater use of concentrated feed and longer manure storage period. Similarly, Guerci et al. [34] found greater emissions in intensive farms compared to PB systems due to the inclusion of conventional soymeal in the feed concentrate. Confined systems can therefore have a high feed efficiency but a higher mineral N fertilization rate and high concentrate supplementation. O’Brien et al. [6] compared intensive confinement systems in the USA and the UK with a PB system in Ireland and found that the CC value of milk from the Irish grass-based system (0.837 kg CO₂ eq per kg of ECM) was 5% lower than the UK confinement system (0.884 kg CO₂ eq per kg of ECM) and 7% lower than the US confinement system (0.898 kg CO₂ eq per kg of ECM).

Laca et al. [32] reported that although cow productivity in a PB farm was almost half that on a semi-confined farm, the environmental impact per kg FPCM in the PB system was notably lower, especially when considering farm co-products. In fact, the live weight of animals sold for slaughter as a milk co-product was higher in the PB system than in the CoSC system.

These findings confirm that mass or economic allocation can improve LCA because of a better partition of the impacts between functional units and co-products.

In addition, the purchased feed, electricity, cleaning products and transport required in PB farms [32] are lower, which all compensate for the lower milk production.

If PB systems use low inputs (such as no silage, protein ingredients and artificial fertilizers), they could have lower environmental impacts than confinement dairy farms that are similar in terms of herd size and milk productivity [49]. Irrespective of how the pasture was used, the farms with the worst production performance (low productivity and feed efficiency) also have the worst CC values [10].

In contrast, other investigations have reported from 5 to 18% higher CC values in exclusively grass-fed cows compared to CoSC dairy cow systems [44,48,49]. Grass-based dairy producers can mitigate the CC value of milk by adopting management practices that improve efficiency and performance [24], optimizing fertilizer and concentrate use [51].

The application of good grazing practices, such as rotational pastures, can also increase the carbon storage capacity of the soil by maintaining the pasture in a vegetative state and ensuring the continuous release of biomass in the soil, with a positive effect on farm CO₂-eq emissions (estimated average reduction from 9–17%) [24,52].

However, some studies that include carbon sequestration have found little or no differences [23,34] between the two systems. Similarly, the meta-analysis of Lorenz et al. [22] found no significant differences in the CC values between the two systems due to the wide range of productivity and the fact that the milk production values partially overlapped (6330-11,650 FPCM/cow in confined systems vs. 4118-6740 FPCM/cow in pasture vs. 3393-10,427 FPCM/cow in semi-confined systems). In addition, the studies including carbon sequestration did not take into account soil type and peculiarities.

Farm management is therefore fundamental and top-performing herds in different countries have between 27 to 32% lower carbon footprints than the average dairy systems in the respective countries [24].

Although farm management, production level and feed efficiency can impact climate change, according to Lorenz et al. [22], productivity limits to the PB system should also be considered.

4. Acidification

The acidification potential (AP) is the extent to which lowering the pH in soil and water negatively affects ecosystems [53].

The AP of natural habitats is caused by the atmospheric deposition of acidifying substances which is a major problem worldwide, varying greatly at regional and local levels [54].

The main acidifying substances from the agricultural sector are ammonia (NH₃), nitrogen oxides (NOx) and sulfur oxides (SOx) which lead to the release of hydrogen ions (H⁺).

The AP is strongly influenced by on-farm activities and when the degree of feed self-sufficiency of the farm increases, the AP decreases [55].

The main sources of NH₃ derive from on-farm activities [11,37,56]. Manure management (housing, storage, distribution of dung and urine) [11,49,56], crop production and the application of synthetic fertilizers [56] contribute to the total AP by between 37% to over 70% [32,57]. SO₂ emissions are mainly due to the production of electricity and the manufacturing of chemical fertilizers, while NOx is emitted from the combustion of fossil fuels [58].

Another large component of AP also derives from off-farm activities such as the production and transport of purchased feed [11,37,56] which contribute to about 34% of the total AP [55].

The AP is negatively correlated to feed efficiency and pasture use efficiency [34,37].

Regarding the influence of the pasture stocking rate on AP, there are inconsistencies among different studies. According to Penati et al. [55], decreasing the stocking rate decreases the AP and farms with a lower stocking rate than 2.3 cow/ha have shown the best results in terms of acidification per kg of FPCM. In contrast, Guerci et al. [34] reported no significant correlations between the AP and stocking rate. However, these differences could be explained by the different geographical area and by the different management of the farms examined by the two studies.

Finally, the length of the grazing season can be an important driver for improving the AP [11,12,57].

In fact, extended grazing has led to increased NH₃ emissions during grazing, and reduced NH₃ emissions from manure management during housing [11].

Reducing or eliminating manure storage and the cattle housing period, results in a lower AP per unit of milk and per unit area [12]. In an extended grazing scenario (7 months grazing), O’Brien et al. [12] observed a reduction in total AP of 10–20%.

With respect to the definition of AP in PB dairy systems, local and regional aspects are fundamental, as well as the management of grasslands [56].

In the studies comparing AP in the two systems (

Table 2. Acidification and eutrophication potential in pasture-based compared to confined or semi-confined dairy cattle systems.

Acidification of PB	Acidification of CoSC	FU	No. of Analyzed Farms	Cattle Breeds	Country	Reference
7.44–16.75 g SO2 eq	15.22–25.64 g SO2 eq	kg ECM	32	na	Italy, Denmark and Germany	[34]
6.90 g SO2 eq	11.90 g SO2 eq	kg FPCM	2	Holstein–Friesian	Ireland	[10]
1.05 × 10−2 mol H+ eq	5.51–9.97 × 10−3 mol H+ eq	kg FPCM	70	na	Italy	[49]
95.4 kg SO2 eq	126.7 kg SO2 eq	ha	2	Holstein–Friesian	Ireland	[10]
Eutrophication of PB	Eutrophication of CoSC	FU	N of Analyzed Farms	Cattle Breeds	Country	Reference
EP: 4.61–7.56 g PO43− eq	EP: 5.85–11.12 g PO43− eq	kg ECM	32	na	Italy, Denmark and New Zealand	[34]
EP: 3.40 g PO43− eq	EP: 4.60 kg PO43− eq.	kg FPCM	2	Holstein–Friesian	Ireland	[10]
FEP: 4.14 × 10−4 kg P eq	FEP: 1.22–3.16 × 10−4 kg P eq	kg FPCM	70	na	Italy	[49]
MEP: 1.05 × 10−2 kg N eq	MEP: 5.10–9.82 × 10−3 kg N eq
TEP: 1.20 × 10−1 mol N eq	TEP: 8.27 × 10−2–1.28 × 10−1 mol N eq
EP: 46 g PO43− eq	49 g PO43− eq	ha	2	Holstein–Friesian	Ireland	[10]

FU: functional unit; PB: pasture-based dairy cattle farm systems; CoSC: semi-confined dairy cattle farm systems; ECM: energy corrected milk; FPCM: fat and protein corrected milk; ha: hectare; EP: eutrophication potential; FEP: freshwater eutrophication potential; MEP: marine eutrophication potential; TEP: terrestrial eutrophication potential; na: not available.

), the PB values were approximately 72–104% lower than in CoSC when expressed in SO₂ eq kg of FPCM [10,34] and −32% when expressed per ha [10].

The lower AP impact on PB dairy systems is also partly due to the lower fertilizer inputs in this type of farming [34]. Generally, the use of grassland, which needs less fertilization, has positive effects on the AP [59]. In addition, the AP is influenced by the level of intensification in crop management because of the consistent use of N fertilizers [37].

In contrast, higher values have been highlighted by Rencricca et al. [49] in herds grazed for four months during the summer season compared to farms where cattle were not allowed to graze (1.05 × 10⁻² vs. 0.551–0.997 × 10⁻² mol H⁺ eq). This could be related to the fact that the CoSC systems were often characterized by high production levels and high dairy efficiency as also observed by Bava et al. [50].

However, other authors reported no substantial differences between PB and CoSC systems in terms of AP per kg of FPCM [32,50].

5. Eutrophication

The eutrophication potential (EP) is defined as the sum of the effects of the excessive growth of phytoplankton caused by nutrient enrichment which leads to an environmental imbalance [60]. This impact category can be divided into three eutrophication potentials: marine (MEP), freshwater (FEP) and terrestrial (TEP), depending on whether the affected geographical area is aquatic, marine or freshwater, or land-based. The EP, particularly of waters, can be accelerated by human activities such as the use of fertilizers in agriculture and industrial human waste [60]. The EP pertains directly to the leaching and run-off of nitrate and phosphate into the ground and surface water, which is influenced by local factors [55,61]. Inorganic nitrogen and phosphorus seem to be major elements affecting the propagation of algae [55,60].

Some papers report that on-farm stages contribute the most to the EP in PB systems [37,58]. Manure excretion and storage are the greatest on-farm sources of nitrate affecting MEP and FEP [11,12].

On the other hand, other papers have found a greater contribution of off-farm stages on the EP [55], especially because of the higher contribution of imported feed.

Feed efficiency is negatively correlated to the EP [32,34], while an increase in the stocking rate corresponds to an increase in the EP [34,44,55]. Farms with a stocking rate lower than 2.3 cows/ha have shown the best results in terms of EP per kg of FPCM [55].

Conversely, Herron et al. [11] report that an increase in the stocking rate, if accompanied by a good management of resources and an increase in milk productivity, leads to a decrease in the EP.

An increase in milk production is often not negatively correlated with the EP, although a higher milk production generally implies a greater quantity of arable land and use of artificial fertilizers [55].

A higher milk yield might also result in a greater demand for replacement animals due to a reduction in herd fertility [44,55]. This thus implies an increase in EP because the rearing of young animals for replacement is one of the greatest sources of both MEP, TEP and FEP [58].

The use of concentrates produced off-farm and their transport are among the highest sources of emissions regarding MEP [11,12,32] and FEP [12,37,58]. This thus suggests that greater quantities of concentrates also have higher eutrophication values [44,57].

Intensive livestock farming and intensively managed pastures increase the EP potential also due to a greater use of N, P or combined fertilizers [11,15,55].

The level of intensification of the grazing system can influence the EP, and P and N losses can be mitigated by grazing management, for example by decreasing grazing pressure [62] and lengthening the grazing period [11].

The EP is increased by steeper slopes causing greater erosion and high precipitation leading to greater leaching, while cooler and wetter climates reduce ammonia volatilization and the larger proportion of grassland, which is less prone to nitrate leaching [15].

Some studies comparing PB and CoSC (Table 2) have reported a lower EP in PB ranging from −31% to −47% [10,34]. In CoSC systems, emissions from manure storage (mainly ammonia) and from crop production (especially nitrogen losses due to leaching) contribute to the total EP. Compared to PB systems, off-farm eutrophic emissions from purchased feed resulted in greater total EP per unit of milk for CoSC. This suggests that feeding less concentrate reduces N and P losses from feed production [10]. The EP per ha was reported to be 6% lower for the PB compared to the CoSC, and total eutrophic emissions per unit of milk were greater for the CoSC compared to the PB; however, the CoSC system produced more milk per on-farm area [10].

Conversely, Rencricca et al. [49] found that mountain farms where the herd grazed for four months during the summer season showed a higher EP (from +6 to +70% higher) compared to farms where the animals did not graze, which was linked to low food self-sufficiency and low productivity of the PB system.

The findings of Rencricca et al. [49] are limited to the specific environment, and to the adverse climate conditions of the alpine area. Local geographical and climatic factors are crucial aspects that cannot be overlooked when analyzing the EP.

6. Land Use

Land is an important and limited resource, which is why land use (LU) is often considered in the LCA of agricultural and forestry production [63]. Land-use activities have transformed most of the planet’s land surface, with natural landscapes often being converted for human use [64] with impacts on the loss of biodiversity and an increase in CO₂ emissions.

However, land use (LU) is also connected to other impact categories, and a greater use of grassland could also have positive effects on increasing farm self-sufficiency and reducing the AP and CC [34], also by virtue of its capacity to sequester carbon [23].

LU includes the on-farm and off-farm land required for a farm to operate.

The greatest percentage of LU in PB dairy farms is distributed on-farm [4,12,47]. Grazing contributes to an on-farm LU from 54% up to 100% in extensive farms [4,12], while smaller percentages may be allocated to feed production and other indoor operations [44].

In any case, LU in PB dairy cow systems varies depending on the climate and the location, which also directly influence the yield and nutritional composition of the pasture and the yield of concentrates used [10,65].

Several investigations have reported a lower milk yield in PB systems compared to CoSC systems, which is associated with a greater LU per kg of FPCM⁻¹ [10] (

Table 3. Land use and water use in pasture-based compared to confined or semi-confined dairy cattle farm systems.

LU of PB	LU of CoSC	FU	No. of Analyzed Farms	Cattle Breeds	Country	Reference
1.62–1.87 m2	0.68–1.43 m2	kg ECM	32	na	Italy, Denmark and New Zealand	[34]
0.73 m2	0.93 m2	kg FPCM	2	Holstein–Friesian	Ireland	[10]
3.60 × 102 Pt	1.76–3.17 × 102 Pt	kg FPCM	70	-	Italy	[49]
0.9 m2	1.0 m2	kg FPCM	na	na	Zealand	[65]
					Europe	[19]
WU of PB	WU of CoSC	FU	No. of Analyzed Farms	Cattle Breeds	Country	Reference
1290 L	4160–9030 L	kg FPCM	70	na	Italy	[49]

LU: land use; WU: water use; FU: functional units; PB: pasture-based dairy cattle farm systems; CoSC: semi-confined dairy cattle farm systems; ECM: energy corrected milk; FPCM: fat and protein corrected milk; na: not available.

).

LU values are lower on farms with high stocking rates and high milk production per animal and per ha [4,34,55], and a noticeable change in LU may also depend on the breed reared and its productivity [44].

Some studies comparing PB and CoSC dairy farms show that PB farms require a greater amount of LU [34,49] to adequately satisfy the nutritional needs of the animals. Conversely, O’Brien et al. [10] highlights how, on average, if off-farm LU is also considered, PB farms have approximately 21% less LU compared to CoSC, because of the large portion of land used for the off-farm production of concentrate components in the CoSC system (almost 60% of the total LU) [10]. Other authors have shown a similar LU between the two systems (0.9 vs. 1.0 LU/kg FPCM) ([65] data estimated from [19,66]).

Wedderburn et al. [65] highlighted that although LU in PB farms tends to be higher, most of the land used for grazing would not be suitable for crop production for human food, therefore there would be no competition for resources and exploiting marginal areas.

7. Non-Renewable Energy Use

The sources of energy used by farms include non-renewable sources such as fossil fuels, electricity and renewable sources. The cumulative consumption of non-renewable energy (NRE) is frequently included in the LCA of dairy farms [67].

NRE measurements can be used to evaluate the strategies aimed at minimizing this impact category and therefore could be important for protecting the natural environment and improving the sustainability of farms [67,68].

From a practical point of view, in the literature, NRE is mostly expressed as the amount of fossil fuels consumed during on-farm and off-farm activities expressed in MJ [11] and the resulting comparison of PB and CoSC farms are shown in

Table 4. Non-renewable energy use in pasture-based compared to confined or semi-confined dairy cattle farm systems.

NRE of PB	NRE of CoSC	FU	No. of Analyzed Farms	Cattle Breeds	Country	Reference
0.92–2.87 MJ	2.40–5.29 MJ	kg ECM	32	na	Italy, Denmark and New Zealand	[34]
2.3 MJ	3.9 MJ	kg FPCM	2	Holstein–Friesian	Ireland	[10]
5.35 MJ	3.50–6.15 MJ	kg FPCM	70	na	Italy	[49]
31,200 MJ	41,600 MJ	ha	2	Holstein–Friesian	Ireland	[10]

NRE: Non-renewable energy use PB: pasture-based dairy cattle farm systems; CoSC: semi-confined dairy cattle farm systems; ECM: energy corrected milk; FPCM: fat and protein corrected milk; ha: hectare; na: not available.

.

The greatest influence on NRE use is off-farm activities, which contribute up to 90% of the total NRE [10,12]. On the other hand, on-farm consumption has a lower contribution, with the main source represented by fossil fuels in the operation of machinery for forage production and electricity generation [10,11,57].

Different factors contribute to NRE for PB and CoSC systems. For example, the main contributors of off-farm NRE in PB farms are the production of concentrated feed and production of synthetic fertilizers and pesticides [10,11], while for on-farm NRE, electricity can contribute 47% [69].

Shine et al. [70] highlighted that milk cooling was the largest consumer of dairy farm electricity (29%). Consumption can further be attributed to water heating (20%), milk harvesting (18%), lighting (3%), wash down pumping (3%), air compressors (1%, if applicable) and effluent water pumping (1%). The remaining 24% can be attributed to other miscellaneous consumption on the farm.

Some authors have found that PB systems have lower NRE than CoSC systems [10,34]. In particular, NRE was 41% lower when considering kg FPCM, 47–62% when considering kg ECM and 14% lower when considering ha as functional units [10].

These findings are linked to the fact that if well managed, PB systems are generally more self-sufficient in terms of feed and require fewer fertilizer inputs and other external raw materials than CoSC farms, which translates into a lower NRE use [34,55,65]. In addition, an intensification of pasture management could lead to a higher NRE consumption because of increased fertilizer use [4].

The variability factors that affect NRE include the location, productivity and stocking rate. For example, PB flatland farms require less NRE than PB farms located in mountainous areas due to the more challenging climatic and geographical conditions, such as height and fragmentation, which require higher fuel consumption [15,40,55].

The productivity of farms also affects the NRE per kg FPCM, which decreases as the amount of milk produced per cow increases [12,71]. In fact, large quantities of milk seem to compensate for the greater use of NRE due to the increased use of fertilizers on farms [4].

Studies present diverging views on the influence of the stocking rate on NRE, probably due to the different amounts of concentrate imported in the different cases studied. O’Brien et al. [12] report that the stocking rate of PB dairy systems has a significant impact on NRE, increasing the NRE as the stocking rate increases. Conversely, Guerci et al. [34] reported no relationship between the two parameters.

8. Water Use and Footprint

Water is a limited resource and from a global perspective, the dairy sector accounts for 18.1% of the overall water consumption [72]. Several water footprint methods have been developed. The LCA includes indicators based on water stress or scarcity characterization factors in order to differentiate water use in areas with different water availability [73]. The LCA considers green, blue and grey waters. Green waters are defined as rainwater and evapotranspiration of soil moisture, while blue waters are surface or deep waters, consumed for irrigation and drinking [74,75] and grey water is polluted by the production processes linked to farming systems. Water use (WU) in dairy livestock farming systems includes green waters and blue waters [74,75]. In fact, irrigated crops mostly use blue water, which could be a source of potential competition for humans.

Water is essential for the majority of dairy farm processes, and the on-farm total WU is typically divided into the following: stock drinking, milking parlor and irrigation [76]. WU can be measured as liters per FPCM, which indicates how much the production of one liter of milk contributes to the loss of freshwater available for the global population, expressed in liters [74].

The main factors affecting WU are as follows:

(a)

Feed factors: Dry matter intake (DMI) and the quality of the forage ingested affect voluntary drinking water intake (WI) [70,77,78]. Dry matter intake can increase the water intake due to increased water losses from feces and urine, increased water used for nutrient oxidation and heat dissipation from nutrient metabolism [79]. Ash intake from the diet also affects WI, probably because of its high blood and urine osmolarity, resulting in increased urinary excretion [80]. Some authors have reported the direct association of some cations (Na, Cl and K) with blood and urine osmolarity [80] and with WI. Other studies found no significant correlations [79]. Palhares et al. [72] also reported that high crude protein content in the diet leads to greater WI and lower water efficiency. In PB dairy farms, the intake of grass may reduce the overall WI. In fact, grass, in general, contains a larger amount of water than hay and concentrates (70–90% vs. 10–30% or more) [65]. However, the dry matter in fresh pasture varies considerably, and it is difficult to accurately measure the DMI of pasture. The adoption of precision nutrition can improve water efficiency and reduce the environmental impact. Palhares et al. [72] showed that animal nutrition can mitigate and help reduce the cost of water, natural resource consumption and the polluting potential of livestock.

(b)

Factors related to climate and farm location such as maximum daily temperature, evaporation potential, sunshine hours and rainfall [15,70,75]. Under high temperatures, cattle exhibit increased WI as a strategy to reduce heat and regulate body temperature [78]. Palhares et al. [77] found higher WU in PB and semi-confined systems compared to intensive systems due to the increased exposure to climate variations. Confined systems exhibited the highest average WI for animal drinking (on average, 84 L per cow a day vs. 58 L in semi-confined and 66 L in PB systems). However, when WI is calculated per kg milk per day, the intensive systems showed the best performance (3.65 L WI per kg milk per day vs. 4.0 L and 3.85 per kg milk per day in semi-confined and PB systems, respectively) [77]. The average daily WI for cattle in a shaded pasture is lower [78] and increases in the summer season [76].

(c)

Animal factors: The wide range of factors that determine the intake of water by a particular animal was recently reviewed by Singh et al. [81]. Metabolic body weight, milk yield [76], physiological phase and herd characteristics [70,77,81] were the main animal factors found to affect WI. A higher metabolic body weight increases WI due to increased water evaporation losses in the body, while milk yield increases WI due to the direct water loss from milk [82]. Palhares et al. [77] identified a direct relationship between the number of lactating cows, milk yield and animal drinking water consumption in dairy farms. However, according to the study by Shine et al. [70], WU is largely correlated with milk production and moderately correlated with herd size and the number of lactating cows.

(d)

Managerial processes: The import of feed, production of concentrates, working practices and irrigation all affect WU [65]. The water footprint therefore varies based on the quantity of forage produced on-farm. The amount of forage imported comprises 10% of the total WU. In PB farms, the water required for pasture production was found to contribute 85% to the WU, which was mostly green water, with only 1% of blue water being used [74].

(e)

Other factors affecting WU include farm tools and equipment. It is important to improve the efficiency of water distribution and prevent leakage in order to minimize water usage and losses. It has been reported that a 26% water loss derives from poorly managed pipes for water distribution to animals [76]. In addition, milking, parlor cleaning and the cooling of milk have been reported to contribute most to the water consumption [70,76,77]. In fact, on average, 33% of WU occurred within the milking parlors [70]. The main drivers of WU in milking parlors were average milk yield per cow and milking frequency [76].

Confined systems have shown the highest consumption for washing the milking parlor [77]. The reported WU for washing was approximately 4 L water per kg milk a day for confined systems vs. 3.75 and 3.9 for PB and semi-confined farms, respectively. However, Palhares et al. [77] highlighted some limitations due to several factors that influence the WU in washing milking parlors, such as the architecture of the milking parlor and waiting corral, the type of milking machine, the waiting time for animals, floor conditions, floor scraping practices before washing, use of pressurized water, washing systems (hose or flushing) and staff training. In this regard, published studies have shown considerable variability in WU in washing milking parlors from 25 L to 924 L water per cow a day [83].

The parlor type affects the amount of water used, as rotary milking parlors involve a greater WU compared with herringbone parlors [76].

Regarding milk cooling, open loop pre-cooling systems use 41% more water than no pre-cool and 25% more water than closed loop pre-cool systems [70].

Little is known about the WU in PB systems since they are difficult to monitor. In the literature, the reported average values of water consumption footprints for PB farms range from 376 to 754 L per kg FPCM [74,75], also including evapotranspiration for pasture and feed production (Table 3).

According to some authors [49], the WU is between 64% and 80.9% lower in PB than in CoSC systems, with values varying depending on the season [77].

9. Biodiversity

Biodiversity was defined by the ONU Convention on Biological Diversity [84] as “the variety and variability of living organisms and the ecological systems in which they live” and includes genetic, species and ecosystem diversity.

Environmental sustainability is also related to the maintenance of animal and plant biodiversity [40].

In recent decades, there has been a global decline in biodiversity [40,85]. This holds true for all biodiversity levels: ecosystems, species and genetic diversity within species.

Extensive pastures contain natural elements which, in addition to providing a physical barrier, attract insects and other small animals, consequently improving pollination services [85] and making the pasture itself an ecosystem [34].

In the soil beneath permanent pastures, there are more earthworms and microbial masses than in cultivated soils [86], thus contributing to the functional diversity of soils and ecosystems [87,88].

In extensively managed pastures, the entomological and plant biodiversity are maximized [89], although good results can be obtained by adopting rotational grazing systems in more intensively managed farms [90].

However, the benefits of each system are variable during the grazing season and are period dependent [89].

Semi-natural grasslands, on which dairy cattle graze, are also foraging and breeding habitats for birds, insects and wild plant communities, and are also used for the conservation of protected species [91].

The biodiversity impact category estimates the biodiversity losses resulting from the different land uses with values that can be expressed in the damage score (DS) per kg of ECM, which explains how the production of 1 kg ECM affects the relative change in species richness compared with a control system [34].

Apart from changes in land use, the other main direct drivers of biodiversity loss include resource overexploitation, invasive species, climate change and pollution [92]. In addition, one of the weaknesses of the characterization factors is the common focus on plant species richness and only capturing data on arthropods, birds, etc., indirectly. Conversely, local dairy breed maintenance is not considered as a characterization factor.

Although PB farms generally have a higher land use, they use more grass in the feed rations and less imported feed, which improves both soil carbon sequestration and biodiversity and reduces CC and freshwater ecotoxicity [26]. Guerci et al. [34] also found that biodiversity losses seemed to be more influenced by low input use than land occupation and other impact categories.

Some authors have reported that PB dairy farms have a lower environmental impact on biodiversity than CoSC systems [91]. When the PB farming intensification gradient (extensive–intermediate–intensive) changes, the area of semi-natural habitat decreases from 42% to 6.1% [93]. In PB farms, increases in milk yield and livestock density negatively affect biodiversity due to changes in the plant composition of pastures, whereas in small-scale mountain dairy farms, large biodiverse areas are maintained [94]. PB systems located in marginal and alpine areas also require the use of rustic and dual-purpose cow breeds, enabling the genetics to be preserved and guaranteeing the survival of animal breeds that are not used in intensive agriculture [95].

The effects of PB dairy systems on biodiversity are complex and depend on the availability of water, the climate and the plant communities that compose it [89,94].

The biodiversity losses in CoSC dairy farming seem to be related to feed sources. The cultivation of concentrated feeds and the protein components of the rations (soy) often result in changes in land use, in the use of fertilizers and the diffusion of monocultures which lead to a loss of biodiversity [85].

Conversely, other authors have found that biodiversity losses decreased when the level of intensification of the production system increased [96]. They suggest that land sparing (due to agricultural intensification and the separation of production) may be preferable to land sharing (diffuse, less intensive agricultural production). However, the authors [96] also highlighted some limitations of their results due to the complexity of assessing biodiversity losses.

10. Conclusions

Life cycle assessment is a commonly used method for assessing the environmental impact of milk production from cattle; however, significant variability in the methodological approaches complicates the comparison of results across studies.

The environmental impact of pasture-based systems for dairy cattle appears to be highly influenced by several input factors, including feed factors, animal-related factors, management practices, climate and geographical area. These variables make it difficult to draw universally valid conclusions regarding the comparison of pasture-based and confined dairy systems on a global scale.

In addition to a comparison of the different systems, this review highlights that there is considerable potential to reduce the environmental impact of milk production in both systems by adopting productivity-enhancing and best management practices and tailoring the mitigation options for each specific region.

The LCA approach should be standardized for the functional unit, the use of allocation method and the collection of primary data in the field in order to improve comparability across the studies.

A better understanding of the different farming systems needs to incorporate further aspects in LCA such as geographical factors, climatic zone, country peculiarities, animal characteristics (es metabolic weight), carbon sequestration, animal health and welfare and ecotoxicity due to the use of veterinary drugs and antimicrobials in animals. A broader concept of One Health and the maintenance of local animal breeds also need to be taken into account.