This study is based on a region of ˜1.24 million km2, which includes ˜45 % of the area of Argentina (Fig. 1a) and contains >90 % of the national herd (SENASA, 2017). The study area was divided in seven beef production regions, following MAyDS (2017): North-western Argentina (NWA, housing 9 % of the national herd), North-eastern Argentina (NEA, 27 %), Semiarid Region (SA, 6 %), Northern Pampa (NP,15 %), Western Pampa (WP, 11 %), South-eastern Pampa (SEP, 15 %) and South-western Pampa (SWP, 9 % Fig. 1). There is a clear difference in productivity between regions, considering that the average weaning rate in the central pampean regions is 73 % and drops to 66 % in the peripheric non-pampean regions (estimated from MAyDS, 2017). In addition, the average final weight in the pampean regions is 373 kg and is reached in 352 days, whereas in the non-pampean regions the average final weight is 393 kg and is reached in 497 days (estimated from MAyDS, 2017).

Fig. 1.

a) Map of the study area (in grey), showing the different beef production regions (MAyDS, 2017): North-western Argentina (NWA), North-eastern Argentina (NEA), Semiarid Region (SA), Northern Pampa (NP), Western Pampa (WP), South-eastern Pampa (SEP) and South-western Pampa (SWP). b and c) Distribution of cattle among the different production systems in each region for cow-calf (b) and finishing (c) stages. The size of the pie chart is proportional to the size of the herd in each region.

(0.35MB).

The beef production cycle in Argentina has two distinct stages: cow-calf and finishing. Cow-calf stage includes the reproduction process, producing calves of 165 kg live weight on average, which enter the finishing stage at weaning. Finishing stage can take many forms and sometimes it can be further divided in backgrounding and fattening. On average, the finishing stage produces steers/heifers of 383 kg after 425 days (MAyDS, 2017). To represent the diversity of beef production systems in the study area, the analysis followed the description of Modal Systems from the Second Biennial Update Report of the Greenhouse Gases Inventory of Argentina (MAyDS, 2017). All modal systems are defined first by the region they occur in. Then, cow-calf systems are characterized as low, medium and high productivity (one of each in each one of the seven regions, resulting in 21 systems in total), and differ among each other by diet composition, and reproductive and sanitary management, which resulted in a different characteristic weaning rate (see Table 1a, and Table I in the appendix for more details). All manure from cow-calf systems was assumed to be allowed to lie as deposited, without any management. Finishing systems in the study area are more diverse, with a total of 47 finishing systems, characterized by region, confinement, diet, length of finishing period, and final weight (MAyDS, 2017; see Table II in the appendix). For the purpose of this study, finishing systems were classified into three categories: grazing, confined and mixed (when animals spend a part of the stage confined and another part grazing; see Table 1b, and Table II in the appendix for more details). Each finishing system was assigned a specific manure management system, according to literature (MAyDS, 2017; see Table III in the appendix for more details).

There is a geographical variation in the distribution of cattle between systems. For cow-calf systems, low productivity systems tend to be more common in the northern regions (NWA, NEA, and NP). For finishing systems, intensification decreases radially from the centre of the study area — in the Pampas — with confined systems disappearing in NEA, SA, and SWP (Fig. 1b and c).

For each cow-calf system, we modelled a herd with 100 cows during a whole cycle (from joining to weaning, see Figure I of the appendix) and estimated all the animal-days of each animal category involved, from the reproductive parameters of the system (described in MAyDS, 2017). As a whole cycle is longer than one year, it includes two reproductive seasons: one that will produce calves counted for this cycle and another that will produce calves that will be part of the next cycle.

First, we estimated the number of cows in each one of three pathways: 1) Lactating at first service and got pregnant in both services, 2) Lactating at first service and got pregnant in the first, but not the second service, and 3) Not lactating at the first service, but got pregnant in both services. The number of cows in pathways 2 and 3 was estimated from the replacement rate for the system, and the number of cows in pathway 1 was the difference between the total (100) and the two other pathways. Then, the number of animal-days for each category (lactating and gestating, not lactating and gestating, lactating and not gestating) was estimated for each pathway, in order to arrive to a total number of animal-days of each category for each system (see Figure I in the appendix). The number of replacement heifers was estimated from the replacement rate, and the days they were needed for, from the fattening length (MAyDS, 2017). The number of cows sent to slaughter was calculated from the replacement rate. The number of calves to be send to the finishing stage (Ni) was calculated as:

For cow-calf systems, we first calculated the amount of beef produced per calf, by multiplying the weaning weight by the carcass-weight to live-weight ratio for calves (0.55; MAyDS, 2017). Weaning weight for each region was calculated as the mean initial weight for finishing systems in the same region (see Table II in the appendix). The amount of beef produced per calf was then multiplied by the number of calves per cow in each system, in order to obtain the amount of beef produced per cow. We also calculated the amount of beef produced by culled cows. Then, the total amount of beef produced per year under each system was calculated by multiplying the amount of beef produced per cow by the number of cows present in each region (2016 stock; SENASA, 2017) and the proportion of animals in each system (see Figure II in the appendix; MAyDS, 2017).

For finishing systems, we calculated the amount of beef produced per steer/heifer, by multiplying the weight gain (final weight minus initial weight) by the carcass-weight to live-weight ratio for steers and heifers (0.56; MAyDS, 2017).Then, the total amount of beef produced under each system per year was calculated by multiplying the amount of beef produced per steer or heifer by the number of steers and heifers present in each region (2016 stock; SENASA, 2017) and the proportion of animals in each system (see Figure II in the appendix; MAyDS, 2017).

The intensity of the environmental impact of beef production in Argentina was calculated for four dimensions (soil erosion, GHG emissions, pesticide ecotoxicity and impact on biodiversity) for each production system in both stages, as described below (and see Figure III in the appendix for an overview). Impact intensity is expressed as impact per kg of beef (carcass weight) produced. Then, the absolute impact of each stage and of all beef production in the study area was calculated on the basis of the 2016 cattle stock per department (SENASA, 2017) and the distribution of cattle among the different systems in each region (MAyDS, 2017), by multiplying the values per kg of beef by the total amount of beef produced (see section 2.2).

Descriptions of agrochemical use in crops, and the estimation of erosion factors followed the regionalization of the 2017 edition of the Applied Agricultural Technology Survey (ReTAA, after its Spanish initials; Bolsa de Cereales, 2017). This survey divides the study area in 17 regions (ReTAA regions, see Figure IV in the appendix) that follow departmental divisions. Thus, each department in the study area was assigned to one beef region and one ReTAA region. For the impacts were ReTAA regions were used, the results are expressed by beef region, by combining the values of all departments in each beef region.

Impact on biodiversity

To estimate the impact on biodiversity, we developed the Impact on Biodiversity Index (IBI).

IBI is formed by two components. The first one, Ai, is the area required to produce one kg of beef under system i (see section 2.4.2). This includes all the area required to produce the feed (grazed or harvested), and it is calculated from the diet description and the yield (for crops) or dry matter production per hectare (for grasslands and pastures) of each feed item. The result is expressed in hectares per kilogram of beef. The second component is based on the Agrobiodiversity component of the Grassland Conservation Index (developed by Parera and Carriquiry, 2014; and applied to beef production by Picasso et al., 2014), where pij is the proportion of the total area required to produce the amount of each feed item (i) required to produce one kg of beef under system j, and bj is the relative weight of each item (representing the difference in biodiversity from the natural vegetation). Thus, IBI is an expression of how much land is required to produce one kilogram of beef weighted by the relative impact of each land cover on biodiversity. For this calculation, all feed items were classified in four categories (each one with a particular coefficient bj), following the classification from Parera and Carriquiry (2014): natural vegetation (b = 1), perennial pastures (b = 0.9), annual pastures (b = 0.5), and crops (b = 0.4).

Soil Erosion

Hydric soil erosion was calculated via the RUSLE equation (Renard et al., 1991):

where A is estimated average soil loss in tons per hectare per year, R is the rainfall-runoff erosivity factor (estimated from the mean annual rainfall in each ReTAA region, data from INTA, 2018), K is the soil erodibility factor (estimated from the proportion of each soil type in each ReTAA region, data from Cruzate et al., 2007), LS is the topographic factor (slope length and steepness, the same value was used for all regions; Gaitán et al., 2017), and C is the cover factor (for crops, the average of the most common rotations was used, and for grasslands and pastures a table value; Gaitán et al., 2017). Finally, a value per hectare of crop and grazing land for each ReTAA region was obtained, and the value for each system was calculated based on the area of each land use required to produce one kilogram of beef (all values used for this estimation can be found in Table V in the appendix). The result is expressed in kg of soil lost per kg of beef.

For the cow-calf stage, we proposed two interventions in order to reduce the productivity gap among producers and thus simulate the proposed increase in weaning rate: 1) to redistribute all the cattle in the low productivity system of a certain region between the high and medium productivity systems in the same region (maintaining current proportions), and 2) to assign all cattle in each region to the higher productivity systems. Sub-scenarios (Bai et al., 2002) for cow-calf systems were created by simulating the application of each intervention, resulting in three sub-scenarios for this stage: Business as usual (BAU), Increased productivity and Maximum productivity (Figure V in the appendix).

For the finishing stage, we proposed two interventions to simulate an increase in final weight: 1) to transfer all the cattle in grazing systems to the grazing system with the higher final weight in the region, and 2) to simulate and increase in the length of confined systems in order to reach a 10 % increase in carcass weight (representing a 18 % increase in final live weight). The extension of the finishing period was calculated using a constant growth rate (Pordomingo, 2018). The sub-scenarios for the finishing stage were constructed by simulating the application of each one of the interventions, as well as the combination of both, resulting in four sub-scenarios for the finishing stage: BAU, longer confinement, heavier grazing, and both interventions (i.e., longer confinement + heavier grazing; Figure V in the appendix).

Finally, we created scenarios by assessing all combinations of sub-scenarios for both cow-calf and finishing stages (Figure V in the appendix). For each scenario, we calculated the changes in production and in each of the four impact dimensions in both relative and absolute terms. All sub-scenarios were created assuming no changes in the cattle stock in each region. However, for the combined scenarios, the stock entering the finishing stage reflected the increase caused by the corresponding cow-calf scenario, so that extra calves produced in each region were assigned to the finishing systems of the same region, following the same proportions observed in the business as usual scenarios.

Considering the many different data sources used, the multi-step calculations and multiple simplifications made in this analysis, we adopted as a decision rule that to say that either beef production or any of the impacts had changed, the difference between a certain scenario and business as usual had to be greater than |5|%.

The proposed interventions rose average weaning rate in the study area from 70 % (estimated in the business as usual scenario) to 73–81 %, and average final weight went from 383 kg (in 425 days) to 390−398 kg (in 429–458 days, see Tables IX and X in the appendix). All interventions increased beef production, even though the changes to the cow-calf stage (i.e., increasing weaning rate by a switch towards systems with higher productivity) had stronger effect increasing beef production (12–29 %) than changes to the finishing stage (i.e., increasing final weight; 3–7 %; Table 2).

The two sub-scenarios for the cow-calf stage (“Increased productivity” and “Maximum productivity”) reduced soil erosion and GHG emissions but increased pesticide ecotoxicity and the impact on biodiversity (Table 2). Considering the different patterns observed in different regions (Fig. 4), we created two new cow-calf sub-scenarios where the interventions were only applied in the non-pampean regions (NWA, NEA and SA), where there was little or no trade-off between impacts. These new sub-scenarios increased production without increasing any impact (Table 2). For finishing systems, all sub-scenarios maintained most impacts at similar levels than the BAU scenario, but the increase in production was modest (up to 7 %, Table 2).

The combination of all sub-scenarios from both stages resulted in 20 scenarios for beef production in Argentina. All combined scenarios — except those based on BAU for the cow-calf stage — increased production (13–36 %). Most of them lowered soil erosion and GHG emissions per kg of beef, around half reduced the impact on biodiversity per kg of beef, and only one reduced the pesticide ecotoxicity intensity (see Table XI in the appendix). However, most scenarios increased all absolute impacts (Table 2). No scenario produced an increase in production without raising some of the absolute impacts. The best performance (i.e., increasing beef production while minimising the increase of its impacts) was from the scenario that combined “Increased productivity in the non-pampean regions” for the cow-calf stage with “Heavier grazing” for the finishing stage, where production increased by 15 %, and only pesticide ecotoxicity increased (by 15 %), while the other three impacts where kept at the same level as BAU scenario (Table 2).

Our results show a great variability in the environmental impact of beef production systems in Argentina, where impact levels change according to the production stage, type of system, region, and impact dimension considered.

Dividing the analysis by stage allows us to plan mitigation strategies by identifying leverages and turning points. In Argentina, the cow-calf stage is responsible for most GHG emissions since this stage involves a larger number of animals (i.e., the reproductive herd), and each one of those animals contributes to GHG emissions, but they do not necessarily contribute directly to beef production. The great contribution of the reproductive herd to GHG emissions intensity has been noticed, and one of the usual mitigations strategies is to increase the efficiency of reproduction to reduce the burden of the reproductive herd (Gerber et al., 2013). On the other hand, our findings show that the finishing stage in Argentina accounts for most of the pesticide ecotoxicity and the impact on biodiversity, as crops present higher values for these impacts than grazing lands and crops represent a larger proportion of the diet in this stage. This also explains the fact that most of the soil erosion comes from the cow-calf stage: although grazing lands are associated with lower soil erosion per hectare than crops, larger extensions of land are required to produce the necessary feed, and thus the total soil erosion for cow-calf systems in Argentina is higher. It is worth noticing that cow-calf systems in other parts of the world can be much more intensive than in Argentina, with higher reproductive efficiency and more crops in the diet, and that would affect the relative contribution of each stage to each impact (Gerber et al., 2013; Robinson et al., 2011).

Our results regarding the trade-off between impacts are in agreement with studies pointing out that GHG emissions cannot be used as the sole indicator for the sustainability of beef production (Röös et al., 2013). The observed differences in the intensity of the trade-off between impacts for cow-calf systems from different regions can be explained by differences in base line productivity and input levels (MAyDS, 2017). In general, cow-calf systems in the central regions in the Pampas have higher productivity than those from the peripheric non-pampean regions (see section 2.1). In all regions, the increase in productivity for cow-calf systems is based in changes in reproductive and sanitary management and diet (Table 1). The improvements in reproductive and sanitary management are mostly the same in all regions, but diet changes are not: in the central pampean regions high productivity systems use more crops and grazing crops than in the peripheric regions (Table I in the appendix). As crops are associated with higher pesticide use, these changes in diet may cross a threshold and produce an increase in both pesticide ecotoxicity and impact on biodiversity. On the contrary, production in the peripheric non-pampean regions is less intensive (high productivity systems still rely mostly on pastures, instead of crops), so that there is room for further intensification without reaching that threshold. Mastrangelo and Gavin (2012) described a similar behaviour when looking at bird diversity along a cattle intensification gradient in North-western Argentina: intermediate-intensity silvopastoril systems yielded just 20 % less beef than high-intensity pasture systems, while maintaining ˜90 % of the bird species present in nearby forests (high-intensity systems maintained just ˜50 % of the bird species). The same applies to different parts of the world: at first, intensification increases productivity without a noticeable increase in environmental impact. However, pass a certain threshold (that would be different in each context and for each impact), any increase in productivity comes with a noticeable increase in environmental impact (Tscharntke et al., 2012).

As a general rule, the larger the gap in efficiency between producers, the larger the potential to increase production with current technology, by just adopting best practice (Gerber et al., 2013; Pacín and Oesterheld, 2015). Available reports for Argentina show that the efficiency gap in cow-calf systems from the non-pampean regions (>100 %, Giancola et al., 2013), is larger than in the pampean regions (˜70 %, Némoz et al., 2013). The larger potential to increase production in non-pampean regions and the fact that there is further room for intensification (without increasing the environmental impact) in those regions, highlights non-pampean regions as the logical focus for the future development of the beef sector.

This study deals with lands that are currently under agricultural use and does not analyse the effects of the expansion of the agricultural frontier. However, it is important to keep in mind that the Chaco region (that overlaps in part with NWA and NEA in this study) is a global deforestation hotspot, and any further intensification processes have to rely on current agricultural land and/or be based in systems that maintain the forest cover. In this regard, this study only deals with current mainstream beef production systems, but there are many alternative beef production systems under development (e.g. Forests management with integrated livestock or MBGI), that might alter or eliminate the trade-offs between production and environmental impact and between different impacts.

Many studies on beef sustainability focused just on impact intensity, and some authors have been reminding us that — as our results clearly show — impact intensity can be a useful metric when tweaking a production system, but proper scenario analysis needs to be based on absolute impacts (Garnett et al., 2015). Our results allowed us to detect one scenario that increases beef production in Argentina by ˜15 % without a significant increase in most of the absolute impacts, excepting for pesticide ecotoxicity which increased by 15 %. However, pesticide ecotoxicity impact could be reduced with changes in agrochemical use through integrated pest management (Abbas et al., 2018) and changes in crop selection (Coupe and Capel, 2016). Furthermore, improve grassland and pasture management could provide better quality feed for grazing systems, reducing the need to supplement diet with crops (Godde et al., 2018). For example, a study on the effect of rotational grazing in the South-eastern Pampa found and improvement in rangeland condition and in carrying capacity, achieving stocking rates 60 % higher than the average for the region (Jacobo et al., 2006).

Our results also clearly show that any significant change in beef production in Argentina requires interventions at the cow-calf stage, as all the scenarios that significantly increased beef production included some intervention at this stage (Table 2 ). On the other hand, changes in the finishing stage have the potential to moderate the increase in absolute impacts resulting from the increased production. The relative impact of interventions in each stage in both production and impacts might change across the globe (Pelletier et al., 2010).

Table 2.

Current absolute values for beef production and its environmental impacts in Argentina, and the percentage of change (relative to business as usual) under different scenarios. Values in bold are considered an improvement from BAU (for production, it means an increase >5 %, for impacts it means any decrease, or an increase ≤ 5 %). BAU = business as usual; NP = non-pampean region. Dotted line highlights the best performing scenario (i.e., the one that managed to increase beef production while minimising the increase of its impacts). See text (section 2.4) for a description of each scenario.

All scenarios considered in this study were based on practices that are currently available and have been proven to increase production and reduce — at least — some of the impacts, but the fact remains that not all producers have adopted them (Pacín and Oesterheld, 2015). Argentina has about 205 thousand beef farms, and a very high diversity in scale among beef producers: 52 % of beef farms have less than 100 cows and sum up ˜8 % of the national herd, whereas on the other hand 5 % of the farms have more than 1000 heads and concentrate ˜40 % of the national herd (SENASA, 2017). Furthermore, the structure of production varies greatly among regions. In Buenos Aires province (at the heart of the pampean region), 40 % of beef farms have less than 100 cows and sum up ˜5 % of cattle in the province and 8 % of the farms have more than 1000 heads and concentrate ˜51 % of the cattle. Meanwhile, in Chaco province (NWA) 84 % of beef farms have less than 100 cows and sum up ˜24 % of the cattle in the province and 1 % of the farms have more than 1000 heads and account for ˜34 % of the cattle (SENASA, 2017). Different producers face different barriers and need different solutions. In this regard, there are some local studies highlighting some possible barriers to the adoption of these practices in Argentina, and possible ways to overcome such barriers (Giancola et al., 2013; Némoz et al., 2013). Some of these barriers are due to a lack of knowledge, either about the existence of better practices or about when and how to apply them properly. For instance, producers still perform undiagnosed routine antiparasitic treatments, which could affect body condition and stimulate resistance. In these cases, extension services might help to increase adoption. In other cases, the barriers for adoption of best practices are economic: producers know how they could be more productive, but they lack the means or incentives to do it. For example, land tenure issues might deter producers from investing in infrastructure (paddocks, watering systems, etc.), and small-scale producers might not be able to cope with the fixed costs associated with more qualified workers or veterinarians. Legislation and public policies could help these situations by facilitating financing and regulating rental contracts. In this regard, Mastrangelo et al. (2019) analysed farm survey data from the Chaco region and found that the influence of access to credit and land tenure on gross revenue are higher for semi-subsistence than commercial farms, As semi-subsistence farms are more likely to have low productivity systems, both land tenure regularization and better access to credits have great potential to increase the productivity of the sector. Furthermore, in some cases barriers might be systemic, where professional or extension services, or infrastructure (e.g. electricity or roads) are missing. In these cases, it is necessary to improve rural infrastructure and extension networks. Finally, for some producers, livestock might not be their main activity. They do not have time or the inclination to allocate more resources to increase their productivity. In those cases, there might be ways to incentivise them to increase productivity, or to find alternative income sources.