Net food production of different livestock: A national analysis for Austria including relative occupation of different land categories / Netto-Lebensmittelproduktion der Nutztierhaltung: Eine nationale Analyse für Österreich inklusive relativer Flächenbeanspruchung

The contribution of different livestock categories to human food supply in Austria was estimated by contrasting the animal products produced with the feedstuffs fed to the respective livestock category on a national basis. The human-edible feed conversion efficiency (heFCE) was taken as an indicator of the net food production. The heFCE was defined as human-edible output via the animal products divided by potentially human-edible input via feed-stuffs (Wilkinson, 2011; Ertl et al., 2015), and calculated on a gross energy and a crude protein basis for the years 2011, 2012, and 2013 for the following livestock categories: cattle, dairy cows, growing-fattening bulls, swine, laying hens, broiler chickens, turkeys, sheep, and goats. The heFCEs were calculated for two different scenarios: For the “CURR” scenario, the current food consumption habits were presumed when calculating human-edible outputs and inputs (i.e., the human-edible fractions of feedstuffs were estimated based on the common achievable extraction rates for human-edible nutrients from feeds), whereas for the “MAX” scenario, the potential maximum extraction rates of human-edible protein and energy based on today’s technology were presumed for both sides.

Production data for cattle, swine, poultry, sheep, and goats were obtained from Statistics Austria for the years 2011, 2012, and 2013 (Bundesanstalt für Agrarwirtschaft, 2016; Statistics Austria, 2016). The units given were: gross domestic production in metric tons of carcass per year for meat products, production in metric tons per year for eggs, and the sum of milk sold and used on farm in metric tons per year for cow, sheep, and goat milk.

The total feed inputs (including both nationally grown and imported feeds) for each livestock category were taken from the national annual feed balance in metric tons of dry matter per year for the years 2011, 2012, and 2013 (Statistics Austria, 2015). The values in the national feed balance are estimates, derived with a software that calculates the feed demand and distribution of feeds over different livestock categories, as explained in more detail in Eurostat (2002), Steinwidder and Krimberger (2003), and Klapp and Theuvsen (2013).

The human-edible outputs were calculated as the amount of energy and protein in the human-edible portions of the livestock products. For the CURR scenario, the amount of bones, as well as losses between slaughter and consumption and the amount used as pet food were subtracted to calculate the amount of edible meat from the total bone-in carcass weight. The respective factors (% of carcass weight) are given in Table 1. For the MAX scenario, only the amount of bones was subtracted from the carcass. The compositions of animal carcasses and eggs were obtained from the USDA Nutrient Database (USDA, 2016), whereas for the composition of cow, sheep, and goat milk, national data were used (Mayer and Fiechter, 2012; ZuchtData, 2014). For goat carcass composition, the protein content of sheep carcasses was taken from USDA (2016), and the fat content was calculated as an average from the results reported by Ringdorfer et al. (2006) for Austrian conditions. The gross energy content of animal products was calculated from the respective nutrient contents, using the caloric factors 23 kJ/g (protein), 38.9 kJ/g (fat), and 17.2 kJ/g (carbohydrates) (Persson, 2011). The presumed composition of animal products and the respective references are summarized in Table 2.

Besides edible meat, the following animal by-products were considered as human-edible output: For the CURR scenario, the “fancy meats” heart, liver, kidney, and tongue, which are usually considered marketable (Ockerman and Hansen, 2000) plus 25% of the blood of cattle and swine (which is approximately the percentage of blood that is used as human food in Austria (Walter et al., 2008)). For the MAX scenario, all potential human-edible animal byproducts, as well as 100% of the blood of cattle and swine, and the edible kill fat were considered as human-edible outputs. According to Venegas Fornias (1996), the amount of human-edible animal by-products, including blood, accounts for 12, 14, and 14% of the live weight of cattle, swine, and sheep (also taken for goats). The chemical composition of animal by-products was calculated from the weighted averages using animal by-product yields given in Ockerman and Hansen (2000), and the chemical composition of each by-product given in Venegas Fornias (1996) and USDA (2016). The amounts of by-products considered in the calculations, as well as their nutrient composition, are summarized in Table 3.

The human-edible inputs comprised the potentially human-edible amount of protein and energy in the feed-stuffs, as well as in the bought-in young stock (calves as input into the growing-fattening bull system). Seventy-nine feedstuffs out of the 176 listed in the national feed balance were actually present in the years 2011–2013 (Statistics Austria, 2015). The potentially human-edible amounts of protein and energy from feedstuffs were calculated by estimating the potentially edible fractions of feedstuffs based on the available literature. Where available, potentially human-edible fractions from the low and high scenarios given by Ertl et al. (2015) were taken for the CURR and MAX scenario in the present study, respectively. For all other feedstuffs, the potentially human-edible fractions were estimated based on the methods described in detail in Ertl et al. (2015). In total, 35 feedstuffs were found to have the potential to be at least partly edible for humans. Of these feedstuffs, the presumed potentially human-edible fractions of protein and energy are summarized in Table 4. These fractions were estimated based on current processing methods and do not consider that the proportion of what is considered as human-edible might be changed via bio-refining in the future. The potentially human-edible fractions of protein and energy were then multiplied for each feedstuff with the respective protein and gross energy content. The energy and protein contents for most concentrate feeds and industrial by-products were taken from INRA et al. (2015). For concentrate feeds not listed in INRA et al. (2015) and for forages, the nutrient composition was taken from the German Society of Agriculture (Universität Hohenheim-Dokumentationsstelle, 1997), and the gross energy content was calculated using the equation provided by GfE (2001) (gross energy = 0.0239 × crude protein (g/kg DM) + 0.0398 × crude fat (g/kg DM) + 0.0201 × crude fiber (g/kg DM) + 0.0175 × nitrogen-free extracts (g/kg DM)).

Cattle, swine, laying hens, broiler chickens, turkeys, sheep, and goats were considered as being raised in closed systems; thus, all inputs and outputs were assigned to the respective livestock category, and no allocation of inputs and outputs to different categories within species was necessary. For dairy cows and growing-fattening bulls, the following allocations were considered: Besides cow milk, culled cows and calves from dairy cows not needed for replacement (with a presumed average live weight of 45 kg) were considered as outputs of the dairy system. The number of culled cows was calculated as the total number of dairy cows divided by the average productive live span for the respective year. The number of calves from dairy cows was calculated as the total number of dairy cows × 365 / average calving interval × 0.96 (considering a stillbirth rate of 4% (ZuchtData, 2014)). The number of replacement calves for the dairy systems was calculated as the number of culled cows × 1.05 (considering 5% rearing losses (Fürst-Waltl and Fürst, 2012). Accordingly, the feed required for replacement calves was also considered as an input for the dairy system. The output of the growing-fattening bull system was defined as the net weight gain between the start of the growing-fattening period and slaughter. The live weight at the beginning of the growing period was either 150 kg (about 60% of the animals) or 280 kg, depending on the system from which the bulls came (conventional calf rearing or suckler calves) (Statistics Austria, 2015). The numbers of animals entering the systems at the two live weights were taken from the national feed balance (Statistics Austria, 2015). The gross output of the growing-fattening bull system was the number of animals entering the system minus 1% (estimated losses) times the average slaughter weight for bulls derived from Statistics Austria (2016).

The protein quality of different human-edible inputs and outputs was assessed using the protein digestible indispensable amino acid score (DIAAS), which has been proposed recently as the preferred method for determining protein quality (FAO 2013). When available, the DIAAS for feedstuffs and cow milk were taken from Ertl et al. (2016). Additionally, DIAAS were calculated for oats and field beans, according to the methods described in Ertl et al. (2016), resulting in protein quality scores for the feedstuffs, which accounted together for about 95% of the total human-edible protein input (scores shown in Table 4). For all animal products except cow milk, the amino acid composition was taken from USDA (2016). Due to the lack of values on the ileal amino acid digestibility of these animal products, the protein digestibility (meat: 94%, milk: 95%, and eggs: 97% (FAO/WHO, 1991)) was taken instead for the calculation of the DIAAS, as proposed in FAO (2013). The resulting DIAAS of the animal products are shown in Table 5. To compare the differences between the protein quality of human-edible inputs and outputs, the protein quality ratios (PQR = protein score for the protein in the animal product / protein score for the protein in the human-edible feeds fed to the respective livestock category) were calculated from the weighted average amino acid scores of inputs and outputs, respectively (Ertl et al., 2016).

To compare different livestock categories with regard to their use of arable land, feedstuffs were grouped in the following four categories: 1) feedstuffs being the main crop from arable land (A-MAIN) (e.g., feed grains, alfalfa, annual forages), 2) feedstuffs being a co-product from arable land (A-Co) (e.g., oilseed cakes, feed straw), 3) feedstuffs from permanent grassland (according to its current use) (GRASS), and 4) other feedstuffs (e.g., fish meal, feed yeast) (OTHERS). In the next step, the amount of feed protein and energy derived from feedstuffs from each category was calculated as percentage of the total dietary protein and energy for each livestock category.

The percentage of human-edible protein and energy in diets of different livestock categories is shown in Table 6. Irrespective of the animal species, the amount of potentially human-edible feeds in livestock diets is a key factor when assessing the net contribution of livestock to the human food supply. Feeding less human food to livestock not only increases the total amount of food available, but it also reduces the environmental impacts of food production (Eisler et al., 2014; Schader et al., 2015). Due to their forestomach system, ruminants can digest fibrous human-inedible plant materials well, meaning that, from a nutrition ecology perspective, ruminants do not compete with humans for food (Hofmann, 1989). The digestive systems of pigs and poultry, however, are similar to that of humans, resulting in similar demands with regard to the nutrient composition of their diets. Although the composition of diets for our farm animals has been optimized over time to maximize productivity, these evolutionary differences still result in profound differences in the amount of potentially human-edible feed in the diets of different livestock categories (Table 6). With potentially human-edible energy and protein contents generally ranging from 9% for the CURR scenario to slightly under 20% for the MAX scenario, the diets of ruminants (except the growing-fattening bull system) comprise substantially less potentially human-edible food than the diets of monogastric animals, in which around 50% (CURR scenario) to about 80% (MAX scenario) of the dietary protein and energy can be considered to be human-edible. For swine and poultry, these values are similar to the proportions reported for diets in other countries. The Council for Agricultural Science and Technology (1999) reported fractions for human-edible energy of 0.73 (USA) and 0.58 (South Korea) for swine; 0.62 (California) and 0.65 (South Korea) for broilers; and 0.62 (USA) for laying hens. For typical rations in the UK, Wilkinson (2011) estimated the human-edible proportions of the diets’ dry matter at 0.64 (swine), 0.75 (broilers), and 0.65 (laying hens). While there seems to be a certain agreement on the proportion of potentially human-edible food in diets for monogastric animals, the values reported for ruminants vary strongly between different countries. For instance, Wilkinson (2011) reported human-edible proportions of 0.36 for dairy cow diets and 0.47 for beef cattle and sheep diets in the UK, whereas the Council for Agricultural Science and Technology (1999) found proportions of the human-edible energy ranging from 0.09 (South Korea) to 0.30 (USA) for dairy cow diets and 0.12 (South Korea) to 0.69 (Nebraska) for beef cattle diets. This suggests that, for ruminant systems, the proportion of potentially human-edible feed depends strongly on the (regional) preconditions (availability and price of concentrate feeds versus quality, availability, and price of forage feeds), whereas for monogastric animals, the composition of diets seems to be quite similar across countries with regard to their proportion of potentially human-edible food. This is most likely due to the limited capability of monogastric animals to digest fibrous (human-inedible) feeds and to the similar production environments for intensive pig and poultry production systems across different countries.

The heFCEs of different livestock categories under the current feeding conditions are shown in Table 7. Dairy cows had the highest heFCE both for energy and protein. Except for dairy cows and cattle in total, all livestock categories had heFCE < 1 for both the CURR and MAX scenarios. Livestock systems with heFCE > 1 are net contributors to the human food supply, whereas at heFCE < 1, livestock consume more human-edible food through their feeds than they produce. The finding that dairy cows have the highest net food production among all analyzed livestock categories is in agreement with Wilkinson (2011), and can be explained by two facts: dairy cows’ rations include only small potentially human-edible fractions (Table 6) and dairy cows are relatively efficient in converting feed dry matter and feed nutrients into milk (Cassidy et al., 2013; Peters et al., 2014). By contrast, swine and turkeys have the lowest heFCE for protein and energy, respectively. The substantial difference between the heFCE for protein and energy in turkeys is due to the relatively higher protein content as compared to the low energy (low-fat) content of turkey meat, whereas the relatively higher heFCE for energy compared to heFCE for protein in swine can be explained by the low protein content of pig meat in contrast to its high energy content (Table 2). Except for swine, under the MAX scenario, the heFCE was always higher for protein than for energy (Table 7). This is in agreement with results from Wilkinson (2011), where – except for dairy cows – the heFCE was always higher for protein than for energy.

Besides the quantitative changes, the transformation of plant protein into animal protein through livestock systems also affects the protein quality of the human-edible protein. The high values for PQR found for different livestock categories in this study (ranging from 1.11 for turkeys to 1.94 for sheep (Table 7)) clearly support the idea that these differences in the protein quality need to be included in the debate about the net food production of livestock (Ertl et al., 2016). The variations in PQR are the result of differences in the protein quality of the animal products (Table 5), as well as differences in the protein quality of the human-edible protein inputs (Table 4). However, it is questionable whether the distinctly lower (i.e., as compared with other animal products) DIAAS calculated for turkeys in this study (Table 5) truly reflects the reality. According to the procedure described in FAO (2013), the DIAAS only considers the first limiting ileal digestible indispensable amino acid. Thus, if the content of one single indispensable amino acid given in the official database is an outlier (i.e., extraordinarily low, as for example, valine in the case of turkey meat), the quality of this protein might be under-or over-estimated. Therefore, PQR for turkeys should be interpreted with care.

As a result of combining PQR with heFCE for protein (PQR × heFCE), not only cattle in total and dairy cows, but also laying hens, sheep, and goats achieved values > 1 for the CURR scenario, meaning that, these systems can be considered to be net contributors to the human protein supply (Table 7). From the perspective of the net food production, the critique on the low efficiency of livestock systems should, therefore, not be generalized. Although it is true that animals require about 6 kg of plant protein to produce 1 kg of animal protein (Pimentel and Pimentel, 2003), the conclusion that about 85% of the protein is, thus, wasted in the food chain (Aiking, 2011) is misleading, because it considers neither the human-inedible plant protein inputs, nor the differences in the protein quality between the plant and animal proteins.

The feed conversion ratios (FCR = kg feed dry matter per kg bone-in carcass weight) for pork, broiler chicken, and turkey found in this study (Table 8) are in overall agreement with the results from other countries. Peters et al. (2014) found a FCR of 3.6 (pork), 2.6 (broiler chicken meat), and 3.4 (turkey) for the US production system, whereas Wilkinson reported a FCR of 3.6 for pork and 2.0 for poultry for typical production systems in the UK. These results indicate that these systems are very similar over different countries (genetic potential of animals, feeding intensity, type of diet, and so on). The FCR for the growing-fattening bull system is not directly comparable with the results from other studies, because in this study, we only considered the last phase of the growing-fattening period, and not the whole life of the bulls. However, the broad range found for the FCR for different beef production systems in the UK (7.8–27.5 kg (Wilkinson, 2011)) indicates that the beef production systems are less uniform than the meat production systems involving monogastric animals, which could already be seen when comparing the potential human-edible content of livestock diets over different countries.

The comparison of the different meat production systems shows that the net food production of a livestock system is mainly influenced by three parameters: 1) the potentially human-edible fraction of the animals’ diet; 2) the FCR; and 3) the nutrient composition of the animal product. For protein, the PQR can be seen as a fourth influencing factor. Unfavorable results in one of these parameters can be compensated with favorable results in one or several of the others when assessing the net food production of a production system. Table 8, for example, shows that the growing-fattening bull system requires about two to four times more feed protein and feed energy to produce the human-edible protein or energy in the animal product as compared with other meat production systems, whereas the net food production of the different meat production systems is similar. The low efficiency of converting feed nutrients into animal products in beef production systems compared with other meat production systems (higher FCR) can, thus, be compensated by the lower proportion of human-edible nutrients in the ration (Table 6).

The concept of heFCE allows a quantitative comparison of the human-edible inputs and outputs of livestock systems under the current feeding conditions. However, this concept does not consider the fact that the human-inedible feeds could be grown on land suitable for the cultivation of crops for direct human consumption (van Zanten et al., 2016). In order to include the potential suitability of land currently used to grow feeds to cultivate food crops, we classified all feedstuffs based on the land category from which they were derived. The percentages of the total feed protein and energy derived from different land categories are shown in Table 9. Except for the growing-fattening bull system, about 50% of the feed protein and energy for ruminant systems was derived from permanent grassland, whereas the feed protein and energy for monogastric animals was nearly exclusively derived from arable land; either as a main crop (41–51% of the feed protein and 67–76% of the feed energy) or as a co-product. This clearly explains why reducing the amount of food-competing feeds would result in a drastic decline in the numbers of pigs and poultry (Schader et al., 2015). The fact that only about 10% of the feed protein and energy for growing-fattening bull systems is derived from permanent grasslands suggests that, in this respect, highly intensive ruminant feeding systems are relatively similar to the feeding systems for monogastric animals. The role of co-products was more important for the provision of feed protein than for feed energy over all livestock categories. This indicates that in systems where animals are mainly fed co-products from crop production to increase sustainability and land use efficiency (as proposed by Schader et al. (2015) or Van Kernebeek et al. (2016)), the energy supply, especially for monogastric animals, might be more challenging than the provision of protein.

The fact that about 50% of the feed energy and protein for cattle, sheep, and goats is derived from permanent grassland emphasizes the important role that ruminants play from a food security perspective, because this land category would otherwise be virtually unsuitable for human food production (Schader et al., 2015). The argument that the current permanent grassland might partly be suitable for growing food crops is true, but with regard to the negative effects on soil erosion and soil carbon emissions, the conversion of grassland to arable land should be avoided (Janzen, 2011; Soussana et al., 2010). In addition to permanent grasslands, growing perennial forage crops on arable land as feed for ruminants (e.g., alfalfa or clover) contributes to maintaining or improving soil fertility by replenishing soil organic matter, preventing erosion, or fixing nitrogen (Janzen, 2011).

Many different methods are available to assess and compare the specific elements of sustainability of different livestock systems (e.g., life cycle assessments (de Vries and de Boer, 2010), land use efficiency (van Zanten et al., 2016), fossil energy requirements (Pimentel and Pimentel, 2003), or the efficiency of livestock in producing food for humans (Wilkinson, 2011)). However, the results from these different methods are often contradictory, making it nearly impossible to rank different livestock systems in terms of their sustainability. Pork, chicken meat, eggs, and milk, for example, require similar areas of land to produce one kg of protein (de Vries and de Boer, 2010; Nijdam et al., 2012), but regarding the net protein production for human consumption and the use of non-arable grassland, milk is superior to meat products (Wilkinson, 2011; Table 7 and Table 9). However, irrespective of the method used to assess the sustainability of different livestock categories, reducing the amount of potentially human-edible feeds is a key factor to more sustainable livestock systems (Eisler et al., 2014; Schader et al., 2015). Estimating the potentially human-edible fractions of livestock diets and calculating the heFCE based on actual national data allows livestock categories to be rated with regard to their current contribution to the net food production, and it also allows conclusions to be drawn on the changes in the livestock numbers or livestock diets that might be needed in the future in order to improve the net food production of livestock in general.