By-products of food and bioenergy industries

Numerous food and biofuel industry side streams are already used as major components of ruminant diets such as hulls and feed meals from milling industry, distillery and brewery by-products, meals and expellers from plant oil production, molasses and pulps from sugar processing, etc. (Feedipedia, 2018; Luke, 2018). Biofuel by-products as ruminant feeds have been reviewed in detail by Makkar et al. (Reference Makkar, Cooper, Weber, Lywood and Pinkney2012). Recent attempts have aimed at utilising such side streams that have not previously been used. Wadhwa and Bakshi (Reference Wadhwa and Bakshi2013) estimated that nearly 50% of all fruits and vegetables in the European Union go to waste with losses occurring during agricultural production, processing, distribution and by consumers. Vegetable residues may be composted and used as soil amendments but with only a limited added value. One option to add value to these products is to preserve them by sun drying (Wadwha et al., Reference Wadhwa, Bakshi and Makkar2015) or ensiling (Orosz and Davies, Reference Orosz and Davies2015) and feed to livestock. Vegetable and fruit residues are challenging raw materials for ensiling as they are easily perishable and typically moist (Wadwha et al., Reference Wadhwa, Bakshi and Makkar2015; Table 1; Supplementary Table S1). Solid-state fermentation of the fruit and vegetable wastes in combination with other non-competing human food biomass could possibly (a) enrich them with proteins and other nutrients, (b) improve feed quality and (c) enhance ensilability (Wadwha et al., Reference Wadhwa, Bakshi and Makkar2015).

Table 1 Chemical composition of some alternative and common feeds for ruminants

EE=ether extract.

¹ References in Supplementary Table S1.

² Tannins g/kg dry matter (DM).

³ Crude fibre.

The production of fruit and vegetable residues is often seasonal, and in many cases they are produced by small or medium size companies, resulting in rather small batches. To be able to recycle these residues back into the food chain requires high hygienic quality of the products and good stability to allow efficient logistics. Some of the major constraints in the use of fruit wastes are the presence of antinutritional factors such as pesticides, mycotoxins, heavy metals and dioxins (Wadhwa et al., Reference Wadhwa, Bakshi and Makkar2015). There are, however, positive experiences as, for example, ensiled tomato and olive by-products have been successfully used in the diets of dairy goats (Arco-Pérez et al., Reference Arco-Pérez, Ramos-Morales, Yáñez-Ruiz, Abecia and Martín-García2017) and ensiled apple pomace up to 30% in the diets of lactating dairy cows (Wadhwa et al., Reference Wadhwa, Bakshi and Makkar2015).

By-products of oilseed crops such as soya bean and rapeseed meals and expellers are widely used as supplementary protein for dairy cows. One of the less used oilseed crops is an ancient plant camelina (Camelina sativa). Camelina has a moderate seed yield potential (Table 2) that combined with low-nutrient requirements and a good resistance to diseases, pests and drought makes it adapted also to low-input farming (Heuzé et al., Reference Heuzé, Tran and Lebas2017b). Camelinaseed oil is an economically interesting on-farm raw material for biofuel production (Keske et al., Reference Keske, Hoag, Brandess and Johnson2013) to increase farmers’ energy independence. Camelinaseed oil is also fit for human consumption (Heuzé et al., Reference Heuzé, Tran and Lebas2017b). Camelina expeller contains lipids with significant amounts of essential fatty acids 18:2n-6 and 18:3n-3 (Bayat et al., Reference Bayat, Kairenius, Stefański, Leskinen, Comtet-Marre, Forano, Chaucheyras-Durand and Shingfield2015), but it is also relatively abundant in CP and essential amino acids (AA) (Table 1). However, ruminal degradability of camelina protein in situ (76%) was higher than that of soya bean (58%) or rapeseed (52%; Lawrence and Anderson, Reference Lawrence and Anderson2015). Feeding unprocessed or processed camelinaseeds to ruminants has sometimes, but not always, decreased DM intake (Table 3; Supplementary Table S2; Table 4; Supplementary Table S3) that may be related to glucosinolates (Lawrence et al., Reference Lawrence, Anderson and Clapper2016). Nevertheless, replacing various conventional protein feeds in ruminant diets with camelina expeller has resulted in comparable milk and protein yields (Table 3) or ADG (Table 4).

Table 2 The suitability for local production of some common and alternative feeds in different production systems, potential yields in Europe, the need of land or water for feed production, and other main environmental aspects regarding crop and ruminant production

TInt=intensive temperate production; TExt=extensive temperate production; DM=dry matter; Yes=suitable; (Yes)=suitable with some restrictions such as species or cultivars (pulses, grass and wheat) or the proximity of the seaside (seaweed).

¹ Van Krimpen et al. (Reference Van Krimpen, Bikker, Van der Meer, Van der Peet-Schwering and Vereijken2013).

² Bayat et al. (Reference Bayat, Kairenius, Stefański, Leskinen, Comtet-Marre, Forano, Chaucheyras-Durand and Shingfield2015).

³ Watson et al. (Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017).

⁴ Table 5.

⁵ Yield potential in tropical areas; Heuzé et al. (Reference Heuzé, Tran, Edouard, Renaudeau, Bastianelli and Lebas2016b).

⁶ Makkar et al. (Reference Makkar, Tran, Heuzé, Giger-Reverdin, Lessire, Lebas and Ankers2016).

Table 3 The effect of some alternative protein feeds on milk production of ruminants

E=expeller; S=seed; DMI=dry matter intake; CS=cottonseeds; RSM=rapeseed meal; SBM=soya bean meal; SBS=soya bean seeds; SFM=sunflowerseed meal; WLS=white lupin seeds.

¹ Isonitrogenous substitution rate (SR) of control protein feed by alternative protein feed.

² Change (%) due to alternative protein feed compared to control protein feed.

³ Number of diet comparisons.

⁴ References shown in Supplementary Table S2.

⁵ Concentrate intake.

⁶ Not reported.

Table 4 The effect of some alternative feeds on the average daily gains (ADG) of ruminants

OM= organic matter.

¹ Substitution rate of control feed by alternative feed.

² Effect of alternative feed on dry matter intake (DMI) or ADG: Dec=decrease; –=no effect; Inc=increase; nr=not reported.

³ References shown in Supplementary Table S3.

Feeding camelina expeller results in high concentrations of trans-11 18:1 and cis-9,trans-11 18:2, unaltered or slightly decreased 18:0 and cis-9 18:1 concentrations and a significant decrease in total saturated fatty acids in dairy cow (Halmemies-Beauchet-Filleau et al., Reference Halmemies-Beauchet-Filleau, Kokkonen, Lampi, Toivonen, Shingfield and Vanhatalo2011 and Reference Halmemies-Beauchet-Filleau, Shingfield, Simpura, Kokkonen, Jaakkola, Toivonen and Vanhatalo2017), in sheep (Szumacher-Strabel et al., Reference Szumacher‐Strabel, Cieślak, Zmora, Pers‐Kamczyc, Bielińska, Stanisz and Wójtowski2011) and in goat milk (Cais-Sokolińska et al., Reference Cais‐Sokolińska, Pikul, Wójtowski, Danków, Teichert, Czyżak‐Runowska and Bagnicka2015) as well as in sheep meat (Table 4). Besides beneficially modifying lipids in ruminant milk and meat, camelina lipids at inclusion rate of 6% in the diet DM decreased ruminal methane and carbon dioxide production of dairy cows by 29% and 34%, respectively (Bayat et al., Reference Bayat, Kairenius, Stefański, Leskinen, Comtet-Marre, Forano, Chaucheyras-Durand and Shingfield2015). However, caution should be exercised in the dosage of lipids as the reduction in methane emissions due to the dietary polyunsaturates may be accompanied with lowered DM intake and milk yield (Bayat et al., Reference Bayat, Kairenius, Stefański, Leskinen, Comtet-Marre, Forano, Chaucheyras-Durand and Shingfield2015).

Grain legume seeds

Grain legumes such as faba bean (Vicia faba), pea (Pisum sativum) and lupins (Lupinus sp.) are old crops cultivated in all arable continents. There are three major modern lupine species bred to animal feed namely white (Lupinus albus), blue (Lupinus angustifolius) and yellow lupin (Lupinus luteus). In the short-term, grain legumes are presumably the most promising alternatives to soya bean (Glycine max) and rapeseed in the temperate areas because their cultivation practices are already available and implemented (Figure 1). However, grain legume seeds are edible by humans as well. Therefore, the utilisation of human-inedible feeds for ruminants and/or feeds the production of which require less or not at all arable land should be encouraged to improve further the sustainability of food production system in the longer term.

Figure 1 Rough overview of some feeds for ruminants with respect to time to enter readily on the market, extent of production today and potential to increase utilisation in ruminant nutrition sustainably in future (small red bubble=limited; medium-sized blue bubble=moderate; large green bubble=high). Data adapted in part from FAOSTAT (2016), Kruus and Hakala (Reference Kruus and Hakala2016) and USDA (2016).

The unique capacity of leguminous plants in conjunction with rhizobium symbionts to biologically fix and utilise atmospheric N enables that inorganic N-fertilisers with rising prices and high requirement of energy in manufacturing are not required. Indeed, the emissions of a potent greenhouse gas N₂O from legume cultivation are generally lower than those from N-fertilised crops (1.3 v. 3.2 kg/ha; Watson et al., Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017). The seed yield potential of grain legumes under optimal conditions is similar or exceeding that of conventional protein crops (Table 2). These advantages make legumes increasingly attractive in the intensive farming in addition to current wide spread use in the low-input and organic farming.

A prerequisite for the spread of grain legume production is the profitability relative to other crops. This is influenced, for example, by yields, volatile producer prices, incentives and production costs. Though the producer prices of grain legume seeds are on average 1.1 to 2.0 times higher than that of wheat in Europe (FAOSTAT, 2016), the competitiveness against more common crops such as wheat is uncertain mainly due to inconsistent DM yields and high seed costs. However, the incentives for protein feeds and reducing the seed costs by producing the seed on-farm can improve the competitiveness of grain legume cultivation. The cultivation of grain legumes is more challenging than that of cereals and grasses as they are sensitive to lodging and due to pests and pathogens they require efficient crop rotation (van Krimpen et al., Reference Van Krimpen, Bikker, Van der Meer, Van der Peet-Schwering and Vereijken2013). Nevertheless, the plant breeding may be able to overcome these agronomical constraints if given enough attention and resources.

Grain legume seeds differ in the chemical composition, the CP content ranging from 240 (peas) to 400 g/kg DM (soya beans). Soya beans have in general the highest ether extract (EE) content, whereas faba beans and peas contain significant amounts of starch and lupin seeds NDF (Table 1). The main storage carbohydrate of lupins is pectin instead of starch (White et al., Reference White, Staines and Staines2007). Lupin seeds contain more EE than faba beans and peas (Table 1) with cis-9 18:1 and 18:2n-6 as major fatty acids (White et al., Reference White, Staines and Staines2007). The protein in grain legume seeds, faba beans and lupin seeds in particular, is low in methionine (Table 1), which is often the limiting AA for the lactation performance of dairy cows (e.g. Pisulewski et al., Reference Pisulewski, Rulquin, Peyraud and Verite1996).

The feasibility of the use of alternative grain legumes in ruminant diets is determined not only by their chemical composition, but also by the rate and extent of degradation of nutrients in the rumen. The degradability of faba bean, pea and lupin protein in the rumen is often over 80% (Watson et al., Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017) that is significantly higher than those of soya bean or rapeseed expellers. In addition, the heat-treatment of faba beans, peas or lupin seeds to lower ruminal degradability has seldom improved animal performance (White et al., Reference White, Staines and Staines2007; Watson et al., Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017). It is plausible that the high-protein degradability in the rumen together with suboptimal AA profile in the undegraded protein of alternative grain legume seeds limit their production responses in high-yielding ruminants. Faba beans contain also antinutritional factors such as vicine and convicine (Heuzé et al., Reference Heuzé, Tran, Delagarde, Lessire and Lebas2016a), lupins quinolizidine alkaloids (Wasilewko and Buraczewska, Reference Wasilewko and Buraczewska1999) and peas lectins and tannins (Heuzé et al., Reference Heuzé, Tran, Giger-Reverdin, Noblet, Renaudeau, Lessire and Lebas2017a). However, ruminants are not susceptible to most of them because of microbial metabolism and degradation in the rumen (Watson et al., Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017).

Replacing protein in soya bean meal partially or completely with faba beans, blue lupin, white lupin or peas has resulted in rather similar bovine lactation performances (Watson et al., Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017; Table 3). Furthermore, the milk fat concentration of medium chain saturates has been lower and those of cis-9 18:1 and 18:2n-6 higher in cows fed white lupins seeds relative to soya bean meal (White et al., Reference White, Staines and Staines2007). In contrast, the milk production responses of alternative grain legumes are often inferior compared to the rapeseed meal in dairy cow nutrition (Watson et al., Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017; Table 3). Substitution of rapeseed meal with faba beans has typically decreased milk protein yield and increased milk urea concentration and the proportion of N excreted in urine suggesting less efficient use of protein in faba beans than in rapeseed (Puhakka et al., Reference Puhakka, Jaakkola, Simpura, Kokkonen and Vanhatalo2016; Table 3), thus leading to increased N emissions from animals.

Partial or total replacement of soya bean or rapeseed protein by faba beans, lupin seeds or peas has not significantly altered ADG or meat chemical composition in growing sheep or cattle (Table 4). Besides replacing protein in ruminant diets, starchy faba beans and peas (Table 1) and lupins with higher metabolisable energy content than cereals (Watson et al., Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017) have potential in replacing cereals as well. Indeed, the substitution of cereal grains by grain legumes in dairy cow diets generally increases milk production (White et al., Reference White, Staines and Staines2007; Watson et al., Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017). Furthermore, starch in peas and faba beans has lower degradability in the rumen than cereal starch (Watson et al., Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017) that lowers the risk for acidosis.

Biorefining of forage crops

Interest in using grass biomass as a raw material for green biorefineries has arisen recently (McEniry and O’Kiely, Reference McEniry and O’Kiely2014; Hermansen et al., Reference Hermansen, Jørgensen, Lærke, Manevski, Boelt, Jensen, Weisbjerg, Dalsgard, Danielsen, Asp, Amby-Jensen, Sorensen, Jensen, Gylling, Leindedam, Lübeck and Fog2017). Grass is effective in converting solar radiation into chemical forms of energy and it grows well in humid temperate areas with a capacity for higher biomass and CP production compared to most annual crops (Table 2). Further, existing technology is available for its cultivation, harvesting and ensiling (Wilkinson and Rinne, Reference Wilkinson and Rinne2018). When preserved as silage, the grass biomass can be refined all year round although losses in the protein and water soluble carbohydrates will take place during the fermentation process compared to the parent herbage.

Typically the first step in a green biorefinery process is liquid–solid separation resulting in a liquid fraction containing the soluble components of grass and a fibrous solid fraction. The yield of the fractions depends on the technical solutions of the process, but it is also greatly affected by the raw material characteristics. The ensiling process can even serve as a pretreatment for the biorefinery process, and it may be further improved by using fibrolytic enzymes at the time of harvest as it has increased the liquid yield (Rinne et al., Reference Rinne, Winquist, Pihlajaniemi, Niemi, Seppälä and Siika-aho2017). In the simplest approach, grass juice can be used as a liquid feed to enrich the diet with highly nutritive forage-based component and it is readily consumed by dairy cows and monogastric animals (Rinne et al., Reference Rinne, Keto, Siljander-Rasi, Stefanski and Winquist2018), or the fibre fraction can be used as a feed for ruminants (Savonen et al., Reference Savonen, Franco, Stefanski, Mäntysaari, Kuoppala and Rinne2018). Grass fibre is less lignified than, for example, woods and straw, and milder processes can be used to hydrolyse it (Niemi et al., Reference Niemi, Pihlajaniemi, Rinne and Siika-aho2017). The hydrolysed sugars can further be used for a variety of purposes including direct use as feeds, and as substrates for lactic acid fermentation or SCP production. Green biorefineries have potential to improve local nutrient self-sufficiency, provide new business opportunities for rural communities and to produce ecosystem services such as improved soil structure, carbon sequestration and biodiversity. The high costs related to transportation and processing have to date prevented the development of commercial green biorefineries on a large scale (Xiu and Shahbazi, Reference Xiu and Shahbai2015).

Grain legumes as forage

Harvesting grain legume stands as whole crop silage enables the utilisation of nutrients in stems and leaves as well and extending the cultivation in areas where the length of growing season may limit complete seed ripening. Although yield potential and organic matter digestibility (OMD) of grain legume stands are high (Rinne et al., Reference Rinne, Dragomir, Kuoppala, Smith and Yáñez-Ruiz2014; Table 2), data on the effects of grain legume whole crop silages on ruminant performance and product quality is limited. In milk production, white lupin silage resulted in lower total DM intakes, but almost similar bovine lactation performance to maize silage as basal forage (Kochapakdee et al., Reference Kochapakdee, Moss, Lin, Reeves, McElhenney, Mask and Santen2004). In meat production, animal performance has been similar or better when white lupin or pea silages have replaced partially or completely grass silage in cattle or sheep diets (Table 4). Due to their lower fibre concentration relative to grass silage, legume silages may lower ruminal methane emissions (Hristov et al., Reference Hristov, Oh, Firkins, Dijkstra, Kebreab, Waghorn, Makkar, Adesogan, Yang, Lee, Gerber, Henderson and Tricarico2013).

Compared to sole cropping, the bi-cropping of grain legumes and cereals may enhance and stabilise DM yields, reduce weeds and plant diseases and improve N-fixation (Hauggaard-Nielsen et al., Reference Hauggaard-Nielsen, Jørnsgaard, Kinane and Jensen2008). As a forage, grain legume–cereal crop mixtures complement the nutritive value of each other providing an appropriate balance between readily fermentable nutrients and N in the rumen (Watson et al., Reference Watson, Reckling, Preissel, Bachinger, Bergkvist, Kuhlman, Lindström, Nemecek, Topp, Vanhatalo, Zander, Murphy-Bokern and Stoddard2017). Replacing half of the grass silage DM with faba bean–wheat silage had no effect on DM intake or bovine milk, fat and protein yields or feed N conversion efficiency to milk protein (Lamminen et al., Reference Lamminen, Kokkonen, Halmemies-Beauchet-Filleau, Termonen, Vanhatalo and Jaakkola2015). Whole crop faba bean–wheat or pea–wheat silages have successfully replaced grass silage in beef production as well (Table 4). Due to the lower costs of N fertilisers and good yield potential, grain legume silages seem to provide a viable alternative for maize and grass silages both in the intensive and extensive production systems (Table 2). The feeding value and ruminal methane emissions of diets containing forage legumes (lucerne, clovers) have been reviewed elsewhere (Dewhurst, Reference Dewhurst2013).

Temperate wood-derived products

Wood is the most abundant source of carbohydrates worldwide. Principal components of wood are cellulose (400 to 450 g/kg DM) and hemicelluloses (200 to 300 g/kg DM, Sjöström, Reference Sjöström1993). Agroforestry approaches in ruminant nutrition are less common in the temperate areas compared to the tropics or the Mediterranean area. There are, however, some applications where, for example, willow (Salix sp.) production for wood chips and the grazing of ruminants are combined to provide additional benefits such as improved microclimate for the animals, self-medication and soil carbon sequestration, although the potential of the untreated wood-based materials to provide energy and nutrients to high-yielding dairy cows is limited (Smith et al., Reference Smith, Leach, Rinne, Kuoppala and Padel2012 and Reference Smith, Kuoppala, Yáñez-Ruiz, Leach and Rinne2014). Indeed, the in vitro digestibility of DM of untreated wood of various tree species was poor with a range from 0.002 to 0.035 (Millett et al., Reference Millett, Baker, Feist, Mellenberger and Satter1970).

A variety of technologies have been used over decades to improve the digestibility of wood-derived lingo-cellulosic materials. The key is to break the link between the lignin and the cell wall carbohydrates, particularly hemicelluloses, in order to improve the digestibility of ligno-cellulose by rumen microbes. Most pulping and papermaking residues have undergone at least partial delignification. Depending on the process, the residue may contain different proportions of hemicellulose and/or cellulose with or without lignin. The digestibility of pure cellulose is rather high and corresponds to the digestibility of typical ruminant feeds such as cereal grains and good quality forages. Saarinen et al. (Reference Saarinen, Jensen and Alhojärvi1959) determined the in vivo digestibility of 40 wood pulps produced by various pulping methods and reported a range in digestibility from 0.27 to 0.90 depending on the lignin content. The in vivo digestibility of bleached (lignin erased and the pulp whitened) chemical pulp fines from mixed hardwood was 0.78 for DM and 0.86 for carbohydrates (Millett et al., Reference Millett, Baker, Feist, Mellenberger and Satter1973), indicating that the materials have a high energy value for ruminants.

Although wood-derived cellulose can be used as a feed for ruminants, it has higher value as, for example, paper raw material. In contrast, hemicelluloses are a by-product of pulping that are typically burned, and interest of using them as feeds has arisen. Hemicelluloses are not homogeneous compounds but a group of mixed polysaccharides. They can be divided into four groups according to their main type of sugars: xylans, xyloglucans, mannans and β-glucans. Spruce (Picea sp.) and pine (Pinus sp.; softwood) contain somewhat less hemicelluloses than birch (Betula sp.; hardwood) and hemicellulose composition differs between species (Saarinen et al., Reference Saarinen, Jensen and Alhojärvi1959). Glucomannans and galactomannans are the principal hemicelluloses of coniferous trees (spruce and pine) and xylans in deciduous trees (birch) while β-glucans are restricted to grasses.

Hemicelluloses in a liquid form are often called wood molasses or wood sugar concentrates. They have successfully been used as diet components for ruminants at up to 10% of DM intake (Zinn et al., Reference Zinn1990 and Reference Zinn1993; Herrick et al., Reference Herrick, Hippen, Kalscheur, Anderson, Ranatunga, Patton and Abdullah2012). An in vitro gas production experiment revealed that hot water and pressure extracted galactoglucomannan and xylan were readily used as fermentation substrates by rumen microbes of dairy cows fed a grass silage and cereal based diet but arabinogalactan was not (Rinne et al., Reference Rinne, Kautto, Kuoppala, Ahvenjärvi, Willför, Kitunen, Ilvesniemi and Sormunen-Cristian2016). In an in vivo digestibility trial, the OMD of the hot water and pressure extracted galactoglucomannan was 0.591 (Rinne et al., Reference Rinne, Kautto, Kuoppala, Ahvenjärvi, Willför, Kitunen, Ilvesniemi and Sormunen-Cristian2016).

Bark is another component of wood that has limited value in the pulp and sawmill industry. Although wild ruminants consume bark voluntarily, the energy value of it is so low that incorporating it into dairy cow diets resulted in the reduction of milk production (P. Kairenius et al., unpublished results). Thus, some processing would be needed to improve the digestibility of bark. Wood-derived feeds typically have very low N and P concentrations. If the basal diet were high in these nutrients, wood-derived feeds could dilute diets and subsequently increase, for example, the N use efficiency of lactating dairy cows as it is mainly determined by N intake (Huhtanen et al., Reference Huhtanen, Nousiainen, Rinne, Kytölä and Khalili2008). Wood-derived feeds may also provide a source of feed in the case of lack of other feeds, for example, in crisis situations. In general, they may fit best in the diets of animals with low-energy requirements rather than in dairy cow diets in the intensive production systems.

Fodder trees and shrubs

Low-quality forages such as rice (Oryza sativa) straw and pangola (Digitaria eriantha) grass low in protein and high in NDF and ADF are common in ruminant nutrition in the tropics (42, 691 and 424 g/kg DM for rice straw (Heuzé and Tran, Reference Heuzé and Tran2015b) and 5 to 12, 610 to 790 and 350 to 420 g/kg DM for pangola grass (Tikam et al., Reference Tikam, Phatsara, Mikled, Vearasilp, Phunphiphat, Chobtang, Cherdthong and Südekum2013), respectively). Thus, the basal diet is typically much lower in protein and higher in fibre compared to that used in the intensive ruminant production of the temperate zones. In Asian tropics, rice straw is commonly supplemented with cassava (Manihot esculenta) chip rich in soluble carbohydrates but poor in CP (750 to 850 g/kg DM and 20 to 30 g/kg DM, respectively; Wanapat and Kang, Reference Wanapat and Kang2015) and soya bean meal. However, the high price of soya bean meal limits its use in smallholder farming.

Leaves of local fodder trees and shrubs such as cassava, leuceana (Leucaena leucocephala), moringa (Moringa oleifera) and sesbania (Sesbania sesban) often contain almost as much CP as NDF (Table 1), the concentration of former being roughly half of that in soya bean meal. Supplementing the rice straw-based diets with these alternative protein sources increases DM intake, improves microbial protein synthesis in the rumen and the efficiency of rumen fermentation with a shift towards propionate (Table 5; Supplementary Table S4), thus potentially mitigating methane production. These beneficial changes may be due to certain natural secondary compounds present in these alternative feeds, namely condensed tannins and saponins (Wanapat et al., Reference Wanapat, Kang and Polyorach2013).

Table 5 Effect of using tropical fodder tree and shrubs supplementation on feed intake, rumen volatile fatty acid production and milk yield in ruminants fed rice straw based diets

DM=dry matter; TVFA=total volatile fatty acids; C₂=acetate; C₃=propionate; C₄=butyrate; Dec=decrease; –=no effect; Inc=increase; RLS60=40% rice straw+60% leucaena silage fed ad libitum; FHM+CH=75 g flemingia hay meal+75 g cassava hay.

¹ References shown in Supplementary Table S5

Combined food–feed production system to provide a year round feeding calendar and to enrich smallholder farming environment is illustrated in Supplementary Figure S1. Under the proposed system, two grass types with (a) erect and tall growth habit and (b) semi-prostrate or prostrate growth habit are used to maximise the biomass production under zero-grazing and grazing, respectively. Roots from cassava can be utilised as a carbohydrate source while the whole top is dried to provide protein (Wanapat, Reference Wanapat2009; Wanapat et al., Reference Wanapat, Foiklang, Ampapon, Mapato and Cherdthong2017). In addition, the leaves of fodder trees and shrubs such as leguminous leucaena, flemingia (Flemingia macrophylla), and moringa are harvested in intervals and used fresh or preserved for later use. The intercropping of cassava with leguminous crops, for example, common bean (Phaseolus calcaratus) and cowpea (Vigna unguiculata), has potential to improve soil fertility and to increase biomass yield (Wanapat, Reference Wanapat2009; Wanapat et al., Reference Wanapat, Foiklang, Ampapon, Mapato and Cherdthong2017). Crop residues such as rice straw, corn stover and sugar cane top are also exploited in ruminant feeding.

Jatrophas

Jatrophas are drought-resistant shrubs or small trees native to American tropics and widely distributed in the tropical and subtropical regions around the world. Jatropha genus includes more than 175 species, Jatropha curcas being one of the most studied species in animal feeding. Jatropha is an interesting biofuel crop due to the high EE concentration of its kernels (570 to 600 g/kg DM; Makkar et al., Reference Makkar, Cooper, Weber, Lywood and Pinkney2012), and the de-fatted kernel residue, jatropha kernel meal, is a good source of nutrients with CP concentration of 620 to 770 g/kg DM (Table 1). In comparison to soya bean protein, jatropha is deficient in lysine, but richer in other essential AA (Table 1; Makkar et al., Reference Makkar, Cooper, Weber, Lywood and Pinkney2012).

The majority of jatropha species are highly toxic to both ruminants and monogastrics due to phorbol esters (1 to 3 mg/g kernel meal; Makkar et al., Reference Makkar, Cooper, Weber, Lywood and Pinkney2012), but they can successfully be detoxified. The complete detoxification is absolutely necessary to avoid animal mortality (Elangovan et al., Reference Elangovan, Gowda, Satyanarayana, Suganthi, Rao and Sridhar2013). In addition, the high concentration of antinutritional factors (trypsin inhibitors, lectin and phytate) may limit the use of jatropha especially for monogastrics unless deactivated by heat treatment and supplemented with phytase enzyme. When completely detoxified, the substitution of soya bean by jatropha has not impaired the DM intake or ADG of sheep and goats (Table 4). Though the yield potential is high (Table 2), the inconsistency of yields of current cultivars is the major restriction for the spread (Heuzé et al., Reference Heuzé, Tran, Edouard, Renaudeau, Bastianelli and Lebas2016b).

Single-cell protein

Single-cell protein consists of microbial cells from yeast, bacteria, fungi or microalgae. These micro-organisms can utilise a wide variety of inexpensive feedstocks and wastes as sources of carbon, nutrients and energy for growth to produce biomass rich in protein. The protein content of SCP varies due to culture conditions, species and strains (Lindberg et al., Reference Lindberg, Lindberg, Teräs, Poulsen, Solberg, Tybirk, Przedrzymirska, Sapota, Olsen, Karlson, Jóhannsson, Smárason, Gylling, Knudsen, Dorca-Preda, Hermansen, Kruklite and Berzina2016) but is in the same order as in soya bean expeller (Table 1). The major constraints are the risk for allergens and the accumulation of heavy metals, pesticides and toxins especially if grown on polluted and contaminated substrates, generally high-nucleic acid content (bacteria and yeasts >fungi >microalgae; 60 to 120, 70 to 100, 30 to 80 g/kg DM, respectively) and economical and efficient mass-scale production and harvesting (Nasseri et al., Reference Nasseri, Rasoul-Amini, Morowvat and Ghasemi2011; Lindberg et al., Reference Lindberg, Lindberg, Teräs, Poulsen, Solberg, Tybirk, Przedrzymirska, Sapota, Olsen, Karlson, Jóhannsson, Smárason, Gylling, Knudsen, Dorca-Preda, Hermansen, Kruklite and Berzina2016). Dietary nucleic acids and their derivatives are rapidly degraded in the rumen and certain end-products can be re-used as sources of carbon and N for bacterial growth (McAllan, Reference McAllan1982), but the N in nucleic acids is not as easily available as that of true protein or ammonia.

The basic stages of SCP production process include (a) medium preparation, (b) fermentation or photosynthesis and (c) harvesting and downstream processing like washing, cell disruption, protein extraction and purification (Ravindra, Reference Ravindra2000). The SCP concept was introduced already during the First World War primarily as a human food (Lindberg et al., Reference Lindberg, Lindberg, Teräs, Poulsen, Solberg, Tybirk, Przedrzymirska, Sapota, Olsen, Karlson, Jóhannsson, Smárason, Gylling, Knudsen, Dorca-Preda, Hermansen, Kruklite and Berzina2016). However, the higher production costs of SCP linked to challenges in efficient and economical cell recovery in relation to more conventional foods and feeds is perhaps the main reason why SCP has not reached widespread commercial use so far. Established processes include the use of yeasts Candida lipolytica and Candida tropicalis with alkanes as substrate (product called Toprina), bacterium Methylophilus methtlotrophus with methane as substrate, bacterium Pseudomonas methylotrophus (Pruteen) with methanol as substrate, filamentous fungus Peacilomyces variotii grown on sulphite spent liquor of forest industry sidestream (Pekilo) and yeast Kluveromyces marxianus grown on whey (Nasseri et al., Reference Nasseri, Rasoul-Amini, Morowvat and Ghasemi2011). The reasons why the SCP concept could become more common and economically viable in future are the rising ecoawareness and the need to intensify nutrient and resource utilisation combined with the sharp price rises caused by the prospect of protein scarcity (Lindberg et al., Reference Lindberg, Lindberg, Teräs, Poulsen, Solberg, Tybirk, Przedrzymirska, Sapota, Olsen, Karlson, Jóhannsson, Smárason, Gylling, Knudsen, Dorca-Preda, Hermansen, Kruklite and Berzina2016).

Microalgae

Microalgae are a diverse group of unicellular or simple multicellular microorganisms with widely varying nutritive composition (Table 1). As animal feed, microalgae have several potential uses. Species high in lipids, such as 22:6n-3-enriched Schizochytrium sp., can be used to modify ovine (Bichi et al., Reference Bichi, Hervás, Toral, Loor and Frutos2013) or bovine (Boeckaert et al., Reference Boeckaert, Vlaeminck, Dijkstra, Issa-Zacharia, Van Nespen, Van Straalen and Fievez2008) milk fat healthier for humans in terms of increased trans-11 18:1, cis-9,trans-11 18:2 and n-3 content. Algal 22:6n-3 supplementation has increased also the n-3 content of ruminant meat (Meale et al., Reference Meale, Chaves, He and McAllister2014), but no effects were found on methane production (Moate et al., Reference Moate, Williams, Hannah, Eckard, Auldist, Ribaux, Jacobs and Wales2013). In turn, microalgae or defatted microalgae residues high in CP (e.g. Spirulina platensis and Chlorella vulgaris), or high in carbohydrates can substitute conventional protein (Lamminen et al., Reference Lamminen, Halmemies-Beauchet-Filleau, Kokkonen, Simpura, Jaakkola and Vanhatalo2017) or energy feeds (van Emon et al., Reference Van Emon, Loy and Hansen2015), respectively.

The AA composition of microalgae generally compares favourably to soya bean meal (Becker, Reference Becker2013) and rapeseed meal (Feedipedia, 2018; Luke, 2018), but may vary significantly between species (Table 1). However, in comparison to rapeseed meal and soya bean meal, microalgae protein is often lower in histidine, which is typically the first AA limiting milk production on grass silage and cereal-based diets (e.g. Vanhatalo et al., Reference Vanhatalo, Huhtanen, Toivonen and Varvikko1999). The protein degradability of many microalgae species is suggested to be higher than that of rapeseed (Costa et al., Reference Costa, Quigley, Isherwood, McLennan and Poppi2016; Lamminen et al., Reference Lamminen, Halmemies-Beauchet-Filleau, Kokkonen, Simpura, Jaakkola and Vanhatalo2017), soya bean and cottonseed meals (Costa et al., Reference Costa, Quigley, Isherwood, McLennan and Poppi2016), but this can possibly be affected by the growing and harvesting conditions of microalgae (Lodge-Ivey et al., Reference Lodge-Ivey, Tracey and Salazar2014). Compared to the conventional protein or energy feeds, large doses of microalgae or defatted microalgae residue may impact negatively on feed intake of ruminants depending on microalgae composition (van Emon et al., Reference Van Emon, Loy and Hansen2015; Costa et al., Reference Costa, Quigley, Isherwood, McLennan and Poppi2016; Lamminen et al., Reference Lamminen, Halmemies-Beauchet-Filleau, Kokkonen, Jaakkola and Vanhatalo2016 and Reference Lamminen, Halmemies-Beauchet-Filleau, Kokkonen, Simpura, Jaakkola and Vanhatalo2017). The palatability of microalgae can possibly be improved by feed processing, for example, pelleting (Hintz et al., Reference Hintz, Heitman, Weir, Torell and Meyer1966). Compared to rapeseed meal, microalgae have not affected milk yield, but decreased the milk protein yield of dairy cows in late lactation, which together with decreasing N utilisation for milk production suggests that the protein value of microalgae is possibly slightly lower than that of rapeseed meal (Lamminen et al., Reference Lamminen, Halmemies-Beauchet-Filleau, Kokkonen, Simpura, Jaakkola and Vanhatalo2017), but similar to soya bean protein (Table 3).

The local on-farm production of microalgae in ponds or in closed photoreactors connected to animal drinking water system could lower the energy inputs of feed drying, preservation and transportation making microalgae cultivation in future a viable concept also in the extensive farming. Indeed, microalgae have successively been distributed through drinking water (Panjaitan et al., Reference Panjaitan, Quigley, McLennan and Poppi2010) to growing cattle grazing low quality grasses to improve microbial protein production in the rumen and diet digestibility (Panjaitan et al., Reference Panjaitan, Quigley, McLennan, Swain and Poppi2015). In addition, microalgal-derived renewable biofuels have high potential to replace fossil fuels of diminishing reserves in future. The cost for the biofuels production from microalgae is not yet competitive with fossil fuels, but with advancing technologies and possible government incentives it may soon become profitable (Milano et al., Reference Milano, Ong, Masjuki, Chong, Lam, Loh and Vellayan2016) thus providing defatted microalgae residues for livestock in a mass-scale.

Seaweeds

Seaweeds are complex multicellular organisms growing in salt water or a littoral zone of marine environment (van der Spiegel et al., Reference Van der Spiegel, Noordam and Fels‐Klerx2013). They can be of many different shapes, sizes, colours and composition. Fresh seaweed contains very large amounts of water (700 to 900 g/kg DM) and needs to be consumed quickly or preserved by, for example, drying or ensiling. Brown algae (Phaeophyceae) are of lesser nutritional value than red (Rhodophyceae) and green algae (Chlorophyceae) due to lower CP content (up to 140 v. up to 500 and 300 g/kg DM, respectively). The protein content of marine seaweeds varies between seasons, but in situ rumen degradable protein remains unaffected with high inherent variability between algal species (24% to 51% of CP; Tayyab et al., Reference Tayyab, Novoa-Garrido, Roleda, Lind and Weisbjerg2016). Protein in all seaweeds is typically deficient in essential AA except for methionine (Makkar et al., Reference Makkar, Tran, Heuzé, Giger-Reverdin, Lessire, Lebas and Ankers2016; Table 1).

Seaweeds are low in cellulose (about 40 g/kg DM) but rich in specific complex carbohydrates (e.g. alginate, laminarin and fucoidan). Step-wise increase in the levels of seaweeds in the diet may enable rumen microbes to adapt and utilise these compounds (Makkar et al., Reference Makkar, Tran, Heuzé, Giger-Reverdin, Lessire, Lebas and Ankers2016). Seaweeds concentrate heavy metals and minerals from seawater and contain several times the ash content of land plants that limits their gross energy value and requires regular monitoring (van der Spiegel et al., Reference Van der Spiegel, Noordam and Fels‐Klerx2013; Makkar et al., Reference Makkar, Tran, Heuzé, Giger-Reverdin, Lessire, Lebas and Ankers2016).

Makkar et al. (Reference Makkar, Tran, Heuzé, Giger-Reverdin, Lessire, Lebas and Ankers2016) have recently reviewed in detail the nutritive value of seaweed indicating that some species have the potential to contribute to the protein and energy needs of ruminants (e.g. Macrocystis pyrifera, Palmaria palmatata, Laminaria digitata, Ulva lactuca), while others contain a number of bioactive compounds, which could be used as prebiotics for enhancing production and health status of animals (e.g. Ascophyllum nodosum). Moreover, some seaweed species have shown potential to mitigate ruminal methane production in vitro depending on the basal diet (Maia et al., Reference Maia, Fonseca, Oliveira, Mendonça and Cabrita2016). The seaweeds used for animal feeding can be cultivated or harvested in the wild (Table 4; Makkar et al., Reference Makkar, Tran, Heuzé, Giger-Reverdin, Lessire, Lebas and Ankers2016; Tayyab et al., Reference Tayyab, Novoa-Garrido, Roleda, Lind and Weisbjerg2016) serving to mitigate nutrient loading and to counteract eutrophication processes (Lindberg et al., Reference Lindberg, Lindberg, Teräs, Poulsen, Solberg, Tybirk, Przedrzymirska, Sapota, Olsen, Karlson, Jóhannsson, Smárason, Gylling, Knudsen, Dorca-Preda, Hermansen, Kruklite and Berzina2016). However, high collection rates in the wild have impaired the equilibrium of coastal ecosystems (Makkar et al., Reference Makkar, Tran, Heuzé, Giger-Reverdin, Lessire, Lebas and Ankers2016). In addition, increased cultivation of seaweeds may promote increased production of bromoform, a metabolic by-product of seaweeds that causes the depletion of atmospheric ozone layer (Carpenter and Liss, Reference Carpenter and Liss2000).

Duckweeds

Duckweeds are monocotyledonous, small floating plants with no stems or true leaves of the botanical family Lemnaceae comprising of four genera (Lemna, Spirodela, Wolffia and Wolfiella). Duckweeds are found worldwide, but they grow best in stagnant water between 17.5°C and 30°C (Heuzé and Tran, Reference Heuzé and Tran2015a) and may have a 50% biomass increase every two days (van Krimpen et al., Reference Van Krimpen, Bikker, Van der Meer, Van der Peet-Schwering and Vereijken2013). Thus, duckweed is a potential novel nutrient source for herbivores worldwide. Only few studies have been performed on duckweed in ruminants (van der Spiegel et al., Reference Van der Spiegel, Noordam and Fels‐Klerx2013). Overall, duckweed is consumed well in both dried and fresh forms (Heuzé and Tran, Reference Heuzé and Tran2015a) and it can supply a significant proportion of protein and other nutrients to animals with no significant adverse effects on performance (Cheng and Stomp, Reference Cheng and Stomp2009; Zetina-Cordoba et al., Reference Zetina-Cordoba, Ortega-Cerilla, Ortega-Jimenez, Herrera-Haro, Sanchez-Torres-Esqueda, Reta-Mendiola, Vilaboa-Arroniz and Munguia-Ameca2013).

The duckweed protein is much lower in essential AA histidine, methionine and lysine compared to that of soya bean and rapeseed expeller (Table 1) that may limit duckweed’s production responses relative to them. Estimates of ruminal protein degradability vary widely between 50% and 80% (Heuzé and Tran, Reference Heuzé and Tran2015a). Duckweed contains significant amounts of ash and NDF (Table 1), but has low-lignin content (57 g/kg DM; Heuzé and Tran, Reference Heuzé and Tran2015a). It has therefore potential to substitute also forage (Zetina-Cordoba et al., Reference Zetina-Cordoba, Ortega-Cerilla, Ortega-Jimenez, Herrera-Haro, Sanchez-Torres-Esqueda, Reta-Mendiola, Vilaboa-Arroniz and Munguia-Ameca2013) and minerals (particularly P; van der Spiegel et al., Reference Van der Spiegel, Noordam and Fels‐Klerx2013) in ruminant diets. Nevertheless, high oxalic acid content may restrict the use of duckweed for livestock (van der Spiegel et al., Reference Van der Spiegel, Noordam and Fels‐Klerx2013).

Similarly to microalgae, local on-farm production of duckweed, for example, in ponds may offer a viable concept for ruminant feed production in future. Nutrient scavenging from field runoffs, manure and greywater by duckweeds has potential to reinforce circular economy practices at farm level and to decrease the environmental footprint of ruminant-based food production systems. The very high growth rate (van Krimpen et al., Reference Van Krimpen, Bikker, Van der Meer, Van der Peet-Schwering and Vereijken2013) enables that duckweed could be regularly harvested and fed to animals as fresh. Feeding fresh duckweed also limits the costs related to drying and preservation on-farm. Due to much bigger particle size relative to microalgae, simple mechanical harvesting of duckweed is feasible.