2.1 Data sources

2.2 Farm systems characteristics

The modal dairy systems and scaled-up experimental farmlets will be referred to as dairy farm systems. The farm systems descriptions with key productive and economic performance indicators for the modal and intensified dairy systems are given in Table 1. The national average dairy systems (2014 and 2019) present an average of 1.0 cows ha^-1 and produce 415 kg milk solids ha^-1. The intensified systems (“CRS” and “10MIL”) present, on average, a two fold increase in stocking rate (1.8 cows ha^-1) and solids productivity per ha (1,000 kg milk solids ha^-1) regarding the national average.

All the systems studied presented common features of typical dairy farming in Uruguay. Around 60% of the total farm area is occupied by long-term mixed pastures that last 3-4 years, and the rest is occupied by annual pastures (i.e., oats, ryegrass) and crops (i.e., sorghum, maize) destined for grazing and silage or grain production. While most of the silage is produced on-farm, these systems rely on purchased concentrate (82-100% bought-in). Generally, supplements are offered on a feed pad or sacrifice paddock where cows are confined during pasture shortage seasons, and when access to pasture is constrained due to weather events. Inorganic fertilizer is imported and applied to pastures and crops in these systems.

Table 1: Description of the national average dairy systems (2014 and 2019) and the “CRS” and “10MIL” intensified dairy systems 

¹The national average was calculated as the milk production-weighted of all modal dairy farm systems surveyed in 2014 and 2019.

2.3 Farm-gate balance calculation

The balance presented in this study involved calculating N inputs and N outputs at the farm-gate according to the guidance document developed by the EU N Expert Panel²², as it is widely used in science and policy. Nitrogen surplus was calculated as the difference between total N inputs and total N outputs and expressed on a per-hectare basis. Nitrogen use efficiency was calculated as the ratio of total N outputs divided by total N inputs and expressed as a percentage.

All N input and output data were collected in individual farms and expressed in kg N ha^-1 of farm area yr^-1. The unit of land considered for this study was the dairy area, defined as the area destined for the milking cows plus the area destined for the dry stock.

3.1 Farm-gate balance

The farm-gate balance of the average Uruguayan dairy systems was similar in 2014 and 2019, as no major changes were observed in inputs, outputs, and surplus between the two surveys, although NUE did improve (Table 2, Table S2 and Table S5). In these systems, inputs were similarly distributed between fertilizer, imported feeds and biological N fixation and N atmospheric deposition, which accounted for 32, 31 and 37% of total N inputs, respectively. Outputs were amply dominated by milk production, as live weight production made a minor contribution. The seven modal dairy farms identified in 2019 imported on average 103 kg N ha^-1 (ranging from 59 to 137 kg N ha^-1) and exported 32 kg N ha^-1 (ranging from 13 to 42 kg N ha^-1), thus presenting a N surplus of 71 kg N ha^-1 (44 and 97 kg N ha^-1) (Table 2, Table S5).

There was a disparity between modal dairy farms (2019) in their contribution to the national dairy N surplus, depending on how their contribution is assessed. When considered in terms of produced milk, all modal systems contributed rather proportionally, and thus, the cumulative milk production scaled almost equally with cumulative N surplus (Gini coefficient = 0.21: Figure 1A). However, modal dairy systems differed greatly in size, and therefore just 21% of the number of farms (i.e., those grouped in the M5B, M5A and M6 modal systems) accounted for 74% of the total national dairy N surplus (Gini coefficient = 0.69; Figure 1B).

The farm-gate balance of the “CRS” and “10MIL” intensified systems had a N surplus 79 to 280% higher than the average 2019 modal dairy system, respectively (Table 2). In the “CRS” systems, NUE remained similar to the 2019 average dairy system, but NUE decreased to 22% in the “10MIL” systems. In the intensified systems, fertilizer and imported feeds became the main N inputs, accounting for 37 and 43% of total N inputs, respectively. Although intensification changed little absolute biological N fixation, this source became relatively less important.

Farm-gate N surplus increased with intensification (Table 2, Figure 2 and Figure 3). Close positive relationships were observed between N surplus and N input (R²=0.998) and N output (R²=0.965). However, while N surplus increased proportionally more as N input increased (b=1.26; Figure 2A, Figure 2D), N output increased proportionally less (b=0.33; Figure 2B). Therefore, N surplus and N output had a markedly nonlinear link (b=2.14). As a direct consequence of these allometric relationships, NUE was not associated with N inputs in a simple way (R²=0.00; Figure 2C). Instead, NUE showed a complex relationship with N inputs, increasing up to 100 kg N ha^-1 and decreasing above 200 kg N ha^-1.

The link between N surplus and all indicators of intensification was also nonlinear. It increased at an accelerating rate as milk solids productivity increased (b=1.77; Figure 3A). This pattern was also clear for stocking rate (b=1.93), total feed intake (b=2.33), as well as for operating profit (b=1.42; Figure 3B-D). This means that increments in all these variables resulted in progressively more significant N surpluses.

Table 2: Farm-gate N balance of the national average dairy systems (2014 and 2019) and the “CRS” and “10MIL” intensified systems 

¹The national average was calculated as the milk production-weighted of all modal dairy farm systems surveyed in 2014 and 2019.

Figure 1: Gini plot of cumulative proportional total milk production (A) for total number of dairy farms vs. (B) cumulative proportional N surplus for Uruguayan modal dairy systems in 2019 

Figure 2: Relationships between the components of farm-gate balance: N input, N output, N surplus and N use efficiency (NUE)  

Figure 3: Relationships between N surplus and several parameters associated with intensification: milk solids productivity (A), stocking rate (B), total feed intake (C) and operating profit (D)  

3.2 Mapping results onto the EU Nitrogen Expert Panel input-output framework

When the relationship between N inputs and outputs for each farm system evaluated was mapped onto the EU Nitrogen Expert Panel input-output framework, NUE and N output targets were achieved by some, but not all, assessed systems (Figure 4). On one hand, the most extensive modal systems had adequate NUE but too low N outputs (i.e., N output below the minimum required to be profitable; 19 and 21 kg N ha^-1). On the other hand, “10MIL” systems presented high N output and NUE within the desired range, but their N surplus was substantially greater than 160 kg N ha^-1. The same pattern was observed in pastoral dairy systems from New Zealand, Australia, Europe, and the US ^{₂₎₍₇₎ 9}.

Figure 4: Annual N input and N output values of dairy systems mapped onto the framework proposed by the EU N Expert Panel²². The two dashed lines encompass the 20-60% range of desired NUE, as lower values risk increasing N pollution and higher values risk mining soil N. The horizontal dotted lines represent the N output associated with the minimum productivity to reach an operating profitable to either (i) remunerate the land and capital factors for current Uruguayan dairy systems or (ii) equal the average operating profit of current agricultural systems that compete for land with dairy systems. Finally, the shaded grey gradient represents the current uncertainty regarding what N surplus would lead to substantial losses of harmful reactive N species. 

4.1 Current and future dairy farming systems: the impact of intensification

Current pastoral dairy systems in Uruguay have lower stocking rates and productivity than pasture-based dairy systems from New Zealand, Australia, Northern Europe, or the US (c.f.Table 1vs. McDowell and others²⁹; Luo & Ledgard³⁰; Ros and others³¹). Likewise, the N surplus of current Uruguayan dairy systems, at 71 kg N ha^-1, is at the lower end of the values reported for pastoral dairy systems worldwide⁹, less than half the national average dairy N surplus of New Zealand³⁰ (186-281 kg N ha^-1), Australia³² (156 kg N ha^-1), Ireland³³ (155 kg N ha^-1), or the Netherlands³⁴ (174-208kg N ha^-1). However, the NUE of current Uruguayan dairy farms (20-35%) is within the 20-40% range reported for dairy systems worldwide⁹. Therefore, the performance of the Uruguayan dairy in terms of farm-gate N balance, as measured in 2019, is relatively better than comparable systems. Furthermore, NUE increased by two per cent points between 2014 and 2019 (Table 1).

Intensification of pastoral dairying in Uruguay, via enhanced forage consumption and increased stocking rate, has been shown to improve profitability in commercial¹¹ and experimental dairy farms ^_16)(17) . For instance, the intensified “CRS” and “10MIL” systems achieved two- to three-fold rises in operating profit compared to the national dairy systems. However, economic performance alone does not guarantee the sustainable intensification of dairy ^{₁₎₍₃₅₎} . Increasing milk production per hectare while maintaining low production costs to remain internationally competitive may entail greater environmental risks.

Farm-gate N balance shows that the “CRS” and “10MIL” intensified systems imported on average 187 and 356 kg N ha^-1 and exported 59 and 85 kg N ha^-1. This resulted in an average N surplus of 128 and 271 kg N ha^-1, respectively (Table 2, Figure 4), comparable to those of intensified pasture-based dairy systems in leading exporting countries ^{_29)(30) 31}.

These results confirm the general trend reported in the literature that intensification of dairy systems increases stocking rate, milk production and operating profit, but also farm-gate N surplus ^{_2)(7)9)(22) 36}. Therefore, it is safe to assume that environmental problems associated with intensified dairy systems in more developed countries may also emerge at some stage due to intensification in Uruguay.

As the average dairy farm intensifies by adopting “CRS” and “10MIL” strategies, N inputs to the farm system increased due to higher use of fertilizer and imported feeds required to support larger feed demand of higher stocking rates. Notably, the NUE of the intensified systems was neither consistently lower nor related to the composition of N inputs. Opposite to negative trends observed in other countries⁹, in the present study, NUE peaked at intermediate values of N input, a consequence of the shape of the allometric relationship between N outputs and inputs (Figure 2C).

4.2 Mapping Uruguayan dairy farms onto the EU Nitrogen Expert Panel framework

Optimal N performance targets would ideally combine the efficient use of high N inputs to sustain animal productivity while minimising N loss⁹. Plotting systems in terms of their N input and output allows us to assess their compliance (or not) with three concurrent targets²². First, a N output target is defined by the minimum profitability required to be competitive (e.g., against other land uses). Second, NUE targets set by achievable efficiency under local agroecological conditions. Third, a maximum N surplus target to avoid losses to the environment of harmful reactive N species. This is the most uncertain target and should be defined for the specific conditions according to environmental sensitivity and policy³⁷.

Mapping Uruguayan farms in the N input-output framework shows that current dairy systems present a lower N surplus than intensified systems, and, thus, less potential risk of harmful N losses. However, 40% of these systems (i.e., M1, M2B, M3B, M4B), which account for 20% of national milk production, had N outputs below the minimum 21 kg N ha^-1 required to be competitive in Uruguay (Figure 4 ). Notably, de Klein and others⁹ placed that threshold at 80 kg N ha^-1 for other pastoral countries, perhaps due to higher values of land.

At the other end of the distribution, the most intensified systems presented the highest output per hectare and NUE above 20%, but also the highest N surplus. Finally, some current and intensified systems plot in the space comprising the desirable range of outcomes for all targets. This suggests that intensified dairy systems can, in fact, be more profitable than the current average system while complying with N performance targets.

4.3 Strategies for sustainable intensification

Different strategies can improve N performance. These strategies can be grouped into intensification, extensification, increasing efficiency and avoiding soil degradation²² (Figure S2). Our results suggest that, in Uruguay, dairy farms with low productivity could increase N inputs and improve their performance without major increases in risk of environmental impact. These systems -which comprised 50% of dairy farms in 2019- could reach the desirable range of outcomes (unshaded part of the figure) following a “traditional” intensification pathway.

For the most intensified and profitable systems (e.g., “10MIL”), the challenge is to increase their efficiency (i.e., NUE) and reduce N surplus. This requires improving, first, internal N recycling, especially N in animal excreta⁹, to reduce fertilizer inputs, and second, the amount of harvested grass per unit of N fertilizer applied³⁸. These adjustments allow either greater N output in saleable products for the same N input, or lower N inputs while maintaining N outputs.

Provided appropriate management practices are adopted to account for those two key aspects, intensified pasture-based dairy may be achieved without major increases in N surplus. In the Netherlands, N surplus remained unchanged in commercial dairy farms between 2013 and 2015 due to the implementation of environmental regulations limiting N application and making mandatory the export of manure surpluses³⁴. Similarly, intensified Irish dairy farms reported positive impacts of regulations upon N surplus and NUE due to less inorganic fertilizer N use and improved manure application timing³⁹.

4.4 Need for further research

Farm-gate N balances are the simplest indicator of a dairy farm’s environmental performance. Its simplicity, however, is also its main drawback because N surplus is not linearly or unequivocally related to harmful N losses to the environment, i.e. nitrate leaching or ammonia volatilisation ^{_{4)(40)41)(42)} 43}. Indeed, the relationship between N balances and groundwater concentrations has been found to be poor, unclear, or non-existent ^{₄₎₍₄₀₎ 43}. Moreover, part of the N surplus might not be lost to the environment but stored in soil organic matter, bringing both environmental and productive benefits⁴⁴. Therefore, N surplus is best interpreted as an indicator of potential risk ^{₂₎₍₉₎} .

Pollution from reactive N is arguably the planetary boundary surpassed to a larger extent by food production systems⁴⁵. Whether the high N surplus associated with intensive dairy systems would end up increasing water pollution -the most problematic local impact of N excess- requires research. Future studies, possibly combining field measurements with simulation models, should quantitatively assess the relationship between N surplus and the biochemical identity, spatial distribution, and temporal dynamics of N losses. Without this, there is no solid basis for designing either (i) effective N management options for the spatial and temporal hotspots that contribute disproportionally to N environmental risk, or (ii) sound targeted policies able to optimise the compromise between economic and environmental sustainability of dairy systems.