The research process until obtaining the final pig database is described in Figure 1A. Articles obtained by online searches (4,237 references) were critically evaluated and successive exclusions were performed. The main exclusions (more related to methodological aspects of the original studies) were performed when assessing the full-text, when 36 references were eliminated (criterium i and ii: 15 publications; criterium iii: 2 publications; criterium iv: 13 publications; and criterium v: 6 publications). The final list of 55 selected studies is described in Table 1.

The first LCA study identified in the pig database was published in 2005. Considering the entire database, 26 journals reported publications, with 16 papers being published in Journal of Cleaner Production and 5 papers in Animal. Production scenarios located in Brazil (which was considered in eight studies), Spain (considered in six studies), France (considered in five studies), and China (considered in four studies) were assessed in the selected papers, as illustrated in Figure 2A. The frequency of studied countries is highly related to the location of the main research groups. However, it is important to highlight that the order of most studied countries is not in complete agreement with the pork production ranking (led by China). Another important aspect related to the geographical characteristics of the papers is that five studies were developed by researchers from countries different than the one (or at least one of the regions) considered in the simulations, with Brazil or South America being studied in three of them.

A scope described as cradle-to-farm gate was used in the majority of the studies, which means that all phases comprised from the crop cultivation (and its inputs/outputs) up to the animal rearing phase were considered in these projects. The impacts associated with slaughtering and processing were considered in nine publications only.

Climate change was the focus of our study. However, the LCA studies also reported other impact categories (Figure 3A). From those variables, the most prevalent were eutrophication and acidification, followed by the use of energy and land.

The main subjects under evaluation in the studies focusing on pig production are presented in Table 2. The characterization of pig production in the region or country was the main objective in 18 studies. Another important objective in the studies was the comparison of production systems (including organic or alternative housing systems), which was the main subject in nine papers.

Changes in feeding practices (diet composition or feeding programs) were studied in 25% of the papers. The relative participation of feed production (which includes each ingredient's life cycle, fabrication, and transport) varied from 31 to 76% of the overall greenhouse gas (GHG) emissions in the pig database (Figure 4A). Despite the importance of feeding to the total pig production impact, the diet composition used in the inventory was described by the minority of the papers. Only 38% of the papers described the ingredient formulas, while only 29% of the studies showed any description for dietary nutritional composition, limited sometimes to crude protein. In addition, the environmental impacts related to the production of individual ingredients were presented in only 9% of the papers. The proportion of total impact associated with feed was highlighted in most of the studies. However, the impact of feed production (considering as a functional unit; e.g., 1 ton of feed) was presented in only 15% of the publications. These data would be of great value for further investigations on feeding practices that may mitigate the potential environmental impact of pig production. In addition, more information on feeding practices would allow a better comparison among studies, as great variability exists between the final results (impact of pig production) presented by the studies even for the same functional unit.

As previously stated, crop production is a major contributor to the overall impacts of the pig production chain. The globalization of feed ingredient markets is relevant to LCA studies because it disconnects commodity production from its use/consumption. In a context in which most of the ingredients used for feed production are internationally traded, it is important to highlight that the impacts associated with a certain product are virtually shared with several countries involved in the international trade. The most frequent example of this intercontinental sharing was the use of soybean imported from South America, mainly from Brazil, in European countries. Considering the pig database, 49% of the studies mentioned the use of Brazilian soybean. For that reason, several papers also mentioned the inclusion of overseas transport during the inventory characterization.

The research process until obtaining the final poultry database is described in Figure 1B. Articles obtained by online searches (6,502 references) were critically evaluated, which resulted in several exclusions. Seventeen references were excluded when assessing the full-text (criterium i and ii: 2 publications; criterium iii: 1 publication; criterium iv: 7 publications; and criterium v: 7 publications). The final list of 30 selected studies is described in Table 3.

The first study identified in the poultry database was published in 2006. Considering the entire database, 13 journals reported publications, with 10 papers being published in Journal of Cleaner Production and 5 papers in Poultry Science. Broiler production was evaluated in 18 studies, eggs were the main product evaluated in 10 studies, and both products were assessed in two papers. Production scenarios located in the United Kingdom and Iran (which were considered in 6 studies each); followed by Argentina, Brazil, France, Netherlands, and the USA, which were considered in two studies each; as illustrated in Figure 2B.

Likewise to the pig database, the scope described as cradle-to-farm gate was used in most studies focusing on poultry production. Impacts associated with slaughtering and processing were considered in 10 publications. Besides climate change, studies presented other impact categories (Figure 3B), such as acidification, eutrophication, and the use of energy and land. The characterization of the meat or egg production in the region or country was the main objective in 15 studies (Table 4).

Three papers described the environmental impacts of replacing ingredients in feed formulas, while one paper described the impacts of dietary supplementation with protease. Feeding was highlighted as the major source of environmental impact in most studies, accounting for 28–82% of the overall impact of climate change (Figure 4B). Despite the importance of feeding to the total impact, the diet composition used in the inventory was not described in most studies. Only 13% of the papers described the ingredient formulas, while only 10% of the studies showed any description for dietary nutritional composition. In addition, the environmental impacts related to the production of individual ingredients were presented in only 13% of the papers, with the impact of feed production (considering as a functional unit; e.g., 1 ton of feed) being presented in only 20% of the publications. The use of Brazilian soybean was reported by 30% of the studies, highlighting the importance of international trade also for the environmental impact of poultry production.

Even though a comparison between organic and conventional systems will not be deeply reviewed, it is important to highlight that several studies indicated the production system as one of the important aspects determining the relative contribution of feeding to the overall environmental impact (due to the feed ingredient composition, number of feeding phases, among others). Conventional production systems were considered in most simulations (i.e., conventional feed ingredients). However, some studies evaluated the environmental impacts of adopting alternative production systems (e.g., organic, free-range, certified labels). According to Leinonen et al. (62, 63), the global warming impact necessary to obtain a given functional unit of feed (e.g., 1 ton) can be low in organic farms in comparison to conventional farms. However, a higher feed amount is generally necessary for organic farms than in conventional production systems to obtain the same functional unit. Several reasons are indicated in the papers, as the impairment in feed conversion ratio, an increase in feed consumption, or even waste of feed or products. Thus, when the total cycle is analyzed, a greater global warming potential impact may be associated with feeding animals in organic than in conventional systems (10, 62, 63).

The environmental impacts of a given rearing system are highly correlated with animal performance, especially feed efficiency (27, 87). Thus, technologies that improve animal performance usually have great potential to mitigate life cycle environmental impacts. In this particular aspect, some technologies were assessed in the reviewed studies. Immunological castration and feed additives are some of these factors (16, 31, 53, 75), but probably many more aspects still need to be evaluated in future research.

Another important factor evaluated in some studies was the impact of innovative practices during the cultivation or processing of feed ingredients. The use of maize genetically modified (59), different processing methods for soybean (48), or crop-animal integrated systems (41) were evaluated, and impacts in the final product (i.e., functional unit) were reported.

The use of alternative feed ingredients is an important strategy in livestock systems. Some studies presented environmental advantages when using co-products in the assessed feeds (28, 33, 39). Other papers indicated that these advantages may be related to the calculation method, with favorable results being reported only when the impacts were not co-allocated between the main and the co-products (35, 88). The environmental cost to obtain co-products cannot be ignored in the LCA analysis, and this should be probably further evaluated in future research. Another limitation to be considered when comparing studies are the different ingredient choices given the difficulties in data acquisition, especially for local or limited ingredients as well as the great variability among processes used to obtain co-products (64, 88).

The distance between feedstuff production location and their place of use is an important argument in favor of using ingredient choice (or replacement) as a strategy to mitigate environmental impacts. Feed ingredients are products with cross-border flows, which are a consequence of globalization. Reducing the distance from producers to consumers means fewer transportation needs, and consequently fewer costs and emissions. Using this argument, several studies were developed proposing the use of locally grown ingredients instead of products cultivated in different countries or even continents (22, 43, 64). This is particularly important for local protein-ingredients that replace imported soybean and soybean meal (64). However, the use of local ingredients must not impair feed conversion (24, 66). Otherwise, the advantages may be lost caused by increased demand for feed to reach the same final weight.

Feed composition in terms of ingredients is also a way to reduce the excretion of nutrients and, consequently, manure composition. For that reason, the choice of ingredients needs to be made always with caution, focusing on the origin, but also on the nutritional quality of the product. Nitrogen excretion in manure is highly correlated with diet formulation. If an increase in nitrogen losses in the manure is related to a given ingredient choice, it is expected that this modification will lead to higher GHG emissions and probably other major consequences too (65). In this context, strategies that mitigate nutrient excretion, such as enzyme supplementation (75), synthetic amino acid partially replacing protein crops, or the use of low-protein diets (6, 42, 57), can potentially mitigate the environmental impacts of both pig and poultry production. The modification of the feed formulation method (89) and the adoption of precision feeding techniques (46, 90) are also very important and innovative tools. Due to its relevance for future animal production, precision feeding will be further discussed in the next section, with a focus on pig production.

Feeding is a major source of environmental impacts, as previously discussed. When correctly applied, precision feeding is an efficient tool to decrease production and environmental costs (91). Pigs and poultry are usually fed according to group requirements, disregarding individual particularities. This means that all animals receive the same feed for an extended period, with part of the population receiving nutrients above their requirements (92). The animals that receive nutrients above their needs excrete this excess. An increased protein intake decreases protein efficiency utilization, resulting in larger nitrogen excretions (93). In many pig commercial systems, the nitrogen retention in conventional phase-feeding programs will rarely exceed 35%, being that the efficiency of nitrogen utilization used in many LCA studies (94). However, nitrogen efficiency varies depending on age, sanitary status, and crude protein levels (95, 96).

Precision feeding consists in providing the right amount of feed with the right balance composition to each animal at the right time. Thus, precision feeding can be defined as the technology that provides each animal the nutrients tailored to meet in real-time the animal requirements (91). Nitrogen and phosphorus excretions can be decreased by 40% and consequently reduce production costs by 10% when using an individual precision feeding program (93, 97).

In this context, precision feeding can improve the sustainability of pig production systems. Automatic feeding stations allow pigs to be fed individually with a diet whose composition is appropriate to their growth potential (91). This strategy is an important pattern shift in animal nutrition because at this point nutritional requirements are no longer consider static, but as dynamic processes that develop differently for each individual.

The use of precision feeding instead of conventional group feeding systems already demonstrated several benefits. The increased nutrient-use efficiency and the consequent reduction in the excretion of polluting substances to the environment, improving the overall sustainability of the production system, are the main advantages presented by this feeding system (91). In addition, studies have shown that it is possible to considerably reduce soybean meal and dicalcium phosphate in diet formulations compared to conventional programs. In validation studies (93, 97), individual feeding allowed a reduction in lysine intake by up to 26%, and nitrogen and phosphorus excretion by 30 and 14%, respectively, without affecting the productive pig performance.

Before applying precision feeding techniques, it is necessary to study the environmental impacts of adopting these techniques. An LCA study performed by Andretta et al. (46) intends to estimate the environmental impact of precision feeding techniques applied to pig production. Once again, in Brazilian scenarios, feeding was the largest source of environmental impact. In addition, the study showed that replacing conventional group feeding with daily group feeding (nutrient supply adjusted daily to meet the group requirements) could decrease the potential impact of eutrophication by 4% and acidification by 3%. The mitigation was even greater (up to 6% for the potential impact of climate change and 5% for eutrophication and acidification) when the program was applied to each animal individually (pigs received diets daily tailored to their requirements).

The study also highlighted a reduction over time in the potential impact of climate change associated with pig feed production related to reducing the expected dietary nutrient levels. In the simulated population, reducing the dietary standardized ileal digestible lysine level by one percentage point led to a reduction of up to 194.7 kg of CO₂-eq per ton of feed, depending on the simulated scenario. Certainly, the main advantage of this method was the improved nutrient use efficiency. In other words, the same amount of product was produced using fewer resources. Monteiro et al. (34) performed a similar study considering Brazilian and French scenarios with simulated data (the previous study used data collected in vivo). In their study, a precision feeding system that fed pigs individually was able to reduce the impact of climate change by 7%.

Animals are exposed to several conditions during their lives and these factors may impact directly their nutrient requirements (98). For example, sanitary challenges affect the way amino acids are used by the animal because the nutrients that would be used for protein deposition are directed to cope with the immune system (98). Sanitary challenges also impact the growth performance of pigs and broilers (99), reducing feed efficiency and consequently increasing the environmental impact associated with this production (27). Cadero et al. (54) reported a significant effect of impaired health status on the carbon footprint of pig production. This is only one of several topics that need to be more evaluated in the future, especially in a scenario with reduced use of antibiotics in animal production.

Several studies on the environmental impact of animal production have been published, but only a few have worked using precision feeding programs or considering sanitary challenges. More studies must be carried out to better understand their environmental impact on modern pig and poultry production. Despite all the variability found in livestock, precision systems can foster some eco-friendly solutions by the possibility of managing animals as an individual, having their diets tailored based on real-time data.

LCA is a well-known and established method to evaluate environmental impacts, particularly for complex production chains as those in the livestock sector. However, LCA has its limitation like any other scientific method. Some of these limitations have been described by Finkbeiner et al. (100). Some of these gaps may apply in the studies described in this systematic review.

One important limitation observed in the studies was the assessment of water use. Many studies did not include this impact category or they did not consider water consumption (water not returned to the system), which is very relevant for agriculture (100, 101).

The great variability in functional units is certainly another important limitation to be highlighted. The unit choice is a challenging task because it impacts directly on the results and is also related to the objective and scope (100). However, the variability among studies is a great limitation when comparing results since transformations are sometimes not possible or precise (e.g., results expressed for 1 ton of live pig are difficult to compare to those expressed for 1 ton of carcass because there are more processes included and sometimes the carcass yield is not fully known).

In addition, impacts on human health are probably insufficiently covered in LCA studies dealing with pig and poultry production. Soil contamination, noise, and odors are some of these impacts that are not commonly addressed in LCA studies. Additionally, the LCA method fails to consider other aspects, such as biodiversity, welfare, and social aspects (100). The positive impact of specific activities may be also disregarded.

Finally, the choice of a single scenario to represent the reality of an entire production chain is another important limitation of some reviewed LCA studies. The issue related to data gathering was previously highlighted (102). A single model (e.g., data collected in a single scenario) are not able to describe the pig and poultry production systems worldwide, and neither probably across regions. Even in integrated systems that are characterized by a higher level of uniformity, it is possible to observe a different performance in each producer (for the same genetic type, with the same feed, and similar management practices). Thus, variability is something that needs to be considered in future LCA studies.