Introduction

Agroindustrial residues are the result of various physical, chemical and biological processes in the industrialization of animal or vegetable products, which usually do not have utility as a raw material for the production chain (^{Rosas, Ortiz, Herrera, & Leyva, 2016}; ^{Saval, 2012}). These residues have been receiving great attention from farmers and researchers in animal science from a couple of decades ago due to its potential use in animal feed. Many agribusinesses that use agricultural raw materials produce residues that can be used as fuel, for animal feed or as fertilizers (^{Food and Agriculture Organization [FAO], 1997}).

In several developing countries, agroindustrial and crop residues are often used as the main components of the diet in livestock, based on their local availability, the quality of the input and due to their low price (^{Mugerwa, Kabirizi, Zziwa, & Lukwago, 2012}; ^{Tingshuang, Sánchez, & Yu, 2002}). Throughout the history of livestock, agroindustrial residues such as oil flour, bran, brewery and distillery grains, beet pulp and molasses, have been widely used. However, less conventional waste has currently greater availability, such as fruit and vegetable processing residues, serums, and culinary residues (^{Mirzaei-Aghsaghali & Maheri-Sis, 2008}).

In the Peruvian Amazon, there is great biodiversity of crops that are currently undergoing industrialization processes; these generate unconventional residues that can be included in the diet of dairy and fattening cattle in the area. Among the industrialized products, fruits, vegetables, roots, seeds, leaves, tubers, and pods are included; some are sold fresh, transformed into nectars, juices, jams, salads, flours, oils, wines, powdered concentrates, and preserves, to name a few examples (^{Saval, 2012}). Currently, there is a notable worldwide trend in the growth of residue production due to the increase in the processing of products for human consumption (^{Saval, 2012}). To get an idea of the percentage of residues generated by the industries, palm oil uses only 9.0 % of the fruit components; the remaining 91.0 % is residue. The coffee industry uses 9.5 %, and the remaining 90.5 % is waste, and the cacao industry uses less than 10 %, and the rest is leftovers (^{Saval, 2012}). Based on this availability, agroindustrial residues would be a good alternative as cattle feed, considering their nutritional composition, price, and time when these are used (^{Preston & Leng, 1987}).

The chemical composition of these unconventional inputs of the tropics should be taken in a referential manner since animal response evaluations would necessarily be required (^{Preston & Leng, 1987}). However, the methods and procedures used for the description of nutritional content, such as a proximal or Weende analysis, are highly accurate and have been extensively evaluated and validated, allowing their replication in new studies (^{Association of Official Analytical Chemists [AOAC], 2005}). Likewise, knowledge of food digestibility is important for the formulation of ruminant rations (^{Bochi-Brum, Carro, Valdés, González & López, 1999}). In that sense, in vitro methods to establish the digestibility of inputs (e.g., ^{Goering & Van Soest, 1970}) have been extensively studied, modified and validated with in vivo digestibility methods, allowing their application according to the accessibility of equipment (^{Giraldo, Gutiérrez, & Rúa, 2007}).

The aim of this study was to carry out the nutritional characterization of 10 agroindustrial residues from the San Martín region in Peru, and based on this, classify them as potential inputs for cattle feeding in the tropics.

Materials and methods

Laboratory analysis

The establishment of the proximal chemical composition was carried out in Laboratorio de Evaluación Nutricional de Alimentos (LENA) [Nutritional Food Evaluation Laboratory] of Departamento Académico de Nutrición [Academic Department of Nutrition], Universidad Nacional Agraria La Molina (UNALM). The contents of dry matter (DM), crude protein (CP), fat or ethereal extract (EE), crude fiber (CF), ash, and nitrogen-free extract (NFE) were measured according to the ^{AOAC method (2005)}.

The establishment of the neutral detergent fiber (NDF) was carried out using the ^{Ankom method (2017a)}: Neutral Detergent Fiber in feed - Filter bags technique, and in the case of the in vitro apparent dry matter digestibility (IVADMD), the ^{Ankom method (2017b)}: Technology Method 3 - in vitro True Digestibility was used, employing the Daisy Incubator equipment. The rationale for this method is to establish incubation conditions similar to those that occur in vivo using solutions, such as minerals, nitrogen sources, and reducing agents that promote the anaerobiosis necessary in the process (^{Giraldo et al., 2007}). The incubation process lasted 48 hours. Both tests were carried out in Laboratorio de Nutrición de Rumiantes, Departamento Académico de Nutrición [Laboratory of Nutrition of Ruminants, of the Academic Department of Nutrition], UNALM.

In addition, protein fractionation as a percentage of total protein was determined according to the methodology published by ^{Licitra, Hernández and Van Soest (1996)}. In this method, fraction A represents non-protein nitrogen (NPN), which is soluble nitrogen in trichloroacetic acid (TCA). Fraction B1 is the rapidly degradable true protein established as a precipitable protein with TCA of the soluble protein in buffer minus the NPN. Fraction B2 is the true protein with an intermediate degradation rate, established as an insoluble protein in buffer minus the insoluble protein in neutral detergent. Fraction B3 is a slowly degradable true protein in neutral detergent determined as insoluble nitrogen (NDIP) minus fraction C. Fraction C is the protein not available or bound to the cell wall derived from insoluble nitrogen in acid detergent (^{Sniffen, O'Connor, Van Soest, Fox, & Russell, 1992}). This analysis was performed on 14 residue samples in the Laboratory of the Institute of Agricultural Sciences in the Tropics, University of Hohenheim, Germany. Finally, the usable crude protein (uCP) of these samples was determined according to the equation of ^{Zhao & Cao (2004)}.

The estimation of total digestible nutrients (TDN) and the net energy of lactation (NE_L) of each of the residues was obtained as proposed by ^{Waller (2004)}. To the resulting value, 10 % was subtracted as a safety margin (^{Waller, 2004}). TDN constitutes a unit of expression of the energy content in food, and when its value is known, other expressions of energy can be calculated using appropriate equations (^{Posada, Rosero, Rodríguez, & Costa, 2012}). The NE_L is a measure of the energy requirements of the milk produced and may vary according to its fat percentages (^{National Research Council [NRC], 1989}). It should be noted that this equation has certain limitations concerning its use in unconventional tropical foods.

Results and discussion

The results of the chemical composition and also of the in vitro apparent digestibility of the agroindustrial residues evaluated are shown in table 1. Ten agroindustrial residues were classified based on the results of the proximal analysis, neutral detergent fiber (NDF), and in vitro apparent dry matter digestibility (IVADMD). These analyses aided in the study of their nutritional characteristics and the possible substitution of one residue for another. There are different criteria for classifying inputs (^{Crampton & Harris, 1974}; ^{Ensminger, 1992}; ^{McDonald, Edwards, & Greenhalgh, 1981}). One criterion that is a common denominator is to consider their nutritional contribution and, in this way, classify them as energy, protein, or fibrous foods.

Agroindustrial energy residues were the byproducts from rice polishing, i.e., split rice grains (< 1/4 grain size), split rice grains (> 1/4 grain size), and rice dust. Moreover, coffee pulp and palm kernel cake were also found in this group. Net energy of lactation (NE_L) ranged from 1.60 Mcal/kg (coffee pulp) to 2.19 Mcal/kg (split rice in the two grain sizes assessed) on a dry basis (table 2), and the IVADMD is in a range of 41.9 % (palm kernel cake) to 99.6 % (split rice with < 1/4 grain size). The values of NE_L, EE, NFE, and in vitro digestibility of these residues were significantly higher than other residues (p< 0.05), confirming their energetic value. In vitro digestibility (IVADMD) of split rice (< 1/4 grain size) in this study was 99.4 ± 0.25 %; this value is due to the high content of nitrogen-free extract and the regular protein content (^{Reyes, 1991}).

In the current study, split rice grains showed an average of 9.2 % crude protein, a higher value than what was reported by ^{Shimada (2009)}, with 8.7 % of crude protein. According to ^{Fundación Española para el Desarrollo de la Nutrición Animal [FEDNA] (2016a)}, rice dust is a good energy source for all farm species, especially for ruminants, due to its high fat content (12.0-18.0 %). The results obtained in this work on the level of fat coincide with what is stated in the literature. Regarding palm kernel cake, the ndf of 55.0-65.0 % is compensated with the fat content of 7.0-10.0 % (^{FEDNA, 2015}). In contrast to the above, in the current study, a slightly higher content of 67.7 % was found with an ethereal extract level of 11.1 %, which must be related to differences in the agroindustrial process. Concerning crude protein, ^{FEDNA (2015)} reports a percentage of 16.7 %, i.e., higher to what was found in this study (14.2 %). The percentages of fat and protein in coffee pulp found in the current study (2.4 % and 12.9 %, respectively) were similar to those reported by ^{Pandey et al. (2000)}, where fat and protein values were 2.5 % and 12.0 %, respectively.

Regarding agroindustrial protein residues, coconut cake and cacao husk were considered within this group because they have significantly higher levels of protein compared to other residues (p< 0.05). The average cacao husk protein was 21.8 %, and the one of the coconut cake was 21.9 %, while the IVADMD was 77.4 % and 52.0 %, respectively. However, the coconut cake stood out for showing a value of 51.7 % for NDF, while the cacao husk was 28.5 %, which coincides with the IVADMD values. The cacao husk protein level of the current study was higher than what has been reported by other studies where the level ranged between 15.0 % and 19.0 % (^{Cardona, Sorza, Posada, & Carmona, 2002}; ^{Lecumberri, Mateos, Izquierdo, Rupélez & Goya, 2007}; ^{Soto, 2012}). With respect to the coconut cake, ^{FEDNA (2016b)} reported that the amount of crude protein in this input is 21.0 % with 90.9 % DM, percentages that agree with the results obtained in this work (21.8 % and 92.4 %, respectively). Besides, the in vitro digestibility of the coconut cake has been reported as limited by the high proportion of protein bound to the cell wall (70.0 %), as stated by ^{FEDNA (2016b)}.

The residues classified as fibrous were rice husk, palm fiber, and heart of palm husk, containing levels of neutral detergent fiber (NDF) higher than the other inputs (p< 0.05). The NDF content of the rice husk varied in a range of 68.0-77.1 %, being lower than what ^{Alemán (2012)} found. This author reported a value of 78.5 % for NDF. On the other hand, the NDF content of palm fiber ranged from 66.8 % to 72.7 %, values similar to those reported by ^{Cuesta, Conde and Moreno (2000)}, i.e., from 66.0 % to 77.0 %. The heart of palm husk showed a crude protein and fat values of 7.0 % and 1.3 %, respectively, lower values compared to those reported by ^{Mosquera, Martínez, Medina & Hinostroza (2013)} (i.e., 9.15 % protein and 5.15 % crude fat).

Table 2 shows the values for total digestible nutrients (TDN) and net energy of lactation (NE_L). The fibrous residues show a low NE_L value, e.g., rice husk with 0.8 Mcal/kg, compared to the energy residues that varied from 1.73 to 2.06 Mcal/kg of NE_L (Split rice with > 1/4 grain size).

The residues with the greatest use potential in cattle feed due to the higher energy contribution are those obtained from rice milling, such as split rice with < 1/4 grain size, split rice with > 1/4 grain size, and rice dust (2.1 ± 0.02, 2.1 ± 0.02 and 1.7 ± 0.02 Mcal/kg of NE_L on a dry basis, respectively); meanwhile, the protein contributions of the inputs with the greatest potential are coconut cake and cacao husk (21.9 % and 21.8 ± 1.34 % of dry-based protein, respectively). On the other hand, the residues with the lowest potential for use due to their low IVADMD value were palm fiber and rice husk (27.7 ± 2.47 % and 27.6 ± 5.02 %, respectively). In general, split rice with < 1/4 grain size was the agroindustrial residue with the best chemical and nutritional composition, as well as the best IVADMD in the current study.

The protein fractionation results are shown in table 3. The cacao husk showed a higher amount of fraction A (between 37.6 % and 43.8 %), indicating that this fraction of the input protein would be rapidly degraded in the rumen and converted into ammonia (^{Tham, Man, & Preston, 2008}). The input that showed the highest amount of fraction C was the rice husk (between 44.1 % and 68.0 %). This input would have a higher amount of lignin, silica, tannin-protein complexes, or other products resistant to hydrolysis by microbial enzymes (^{Tham et al., 2008}), which are directly linked to the amount of protein fraction C, being considered as a low-quality input.

Those inputs considered as protein, such as cacao husk and coconut cake, showed different percentages of protein fractions. As indicated above, the cacao husk contains a higher fraction A value than the other inputs, but it also had a considerable percentage of fraction B (between 36.5 % and 38.9 %), with the B2 fraction showing the highest concentration (between 16.8 % and 19.5 %). This fraction is normally fermented slowly in the rumen due to the bacteria that require ammonia as the sole source of nitrogen (^{Russell, O'Connor, Fox, Van Soest, & Sniffen, 1992}). Fraction C of this input ranged between 19.7 % and 23.5 %.

On the other hand, the coconut cake showed a higher amount of protein fraction B (75.5 %), where the slowly degradable true protein (B3) showed a higher value (52.8 %). The B3 fraction, since it only degrades in the rumen by 10.0-25.0 % and the rest passes into the intestinal tract, is the fraction with the highest use efficiency in ruminants (^{Sniffen et al., 1992}). Fraction C of the coconut cake had a value of 16.5 %. Of these protein inputs, the coconut cake would become more efficient due to the high content of fraction B3 compared to cacao husk, and as it has a lower fraction A (8.0 %) and fraction C (16.5 %), it is considered a good quality protein input. In this regard, ^{Guevara-Mesa et al. (2011)} reported that coconut flour presented fractional values A, B1, B2, B3, and C of 11.41 %, 3.15 %, 23.26 %, 48.02 %, and 14.17 %, respectively, i.e., similar values to those observed in the current study.

Several works indicate the fractioning of protein from different inputs and forages for ruminant feeding such as alfalfa, corn, soybean flour, wheat bran, cotton, sorghum, barley, oat, cropping residues, leguminous and grass forages (^{Gaviria, Rivera & Barahona, 2015}; ^{Guevara-Mesa et al., 2011}; ^{Sniffen et al., 1992}). However, it is still necessary to investigate in detail, information on the fractioning of proteins in non-traditional inputs and, with greater emphasis on agroindustrial residues from the tropics in Peru. Likewise, Agroindustrial residues considered as protein show similar usable crude protein uCP values (cacao husk between 21.6 % and 24.5 %, and coconut cake with 24.9 %) (table 3).

The relationship between neutral detergent fiber (NDF) and the in vitro dry matter digestibility (IVADMD) of the agroindustrial residues evaluated can be observed in figure 1.

Source: Elaborated by the authors

Figure 1 Relationship between the neutral detergent fiber (NDF) and the in vitro apparent dry matter digestibility (IVADMD) of the agroindustrial residues in the San Martín region, Peru. 

A significant negative correlation was found between both parameters (r = -0.98), i.e., when there was an increase in the NDF, the IVADMD value decreased. This relationship was most notable in those residues with high NDF content, such as rice husk, palm fiber, heart of palm husk and palm kernel cake, since the in vitro digestibility of the NDF would be reduced and, therefore, also the dry matter digestibility (^{Mertens, 2009}). ^{Mahyuddin (2008)} reported a negative correlation between NDF and IVADMD of tropical pastures; however, the coefficient of determination of this work (R²= 0.24) was lower compared to the results of the current study (R²= 0.96). The reason could be the variation between the chemical components among tropical forage species, while, in the current work, agroindustrial residues had less variation in these components. Another study also reported a negative relationship between the IVADMD and the NDF of Kikuyu grass, showing an R² value of 0.75, that is, a lower value compared to what was found in other local residues of the tropics (^{Rodríguez, 1999}). On the other hand, those residues with low levels of NDF and high values of IVADMD, such as coconut cake, coffee pulp, cacao husk, rice dust, split rice with < 1/4 grain size and split rice with > 1/4 grain size, would be residues with higher use.

Conclusions

Agroindustrial residues with increased energy potential and with good in vitro digestibility (IVADMD) were split rice with < 1/4 grain size, rice dust and split rice with > 1/4 grain size, with net energy of lactation (NE_L) values of 2.1 ± 0.02, 2.1 ± 0, 02 and 1.7 ± 0.02 Mcal/kg on a dry basis, respectively, and IVADMD values of 99.3 ± 0.2 5 %, 90.5 ± 0.4 2 % and 99.0 ± 0.68 %, respectively.

The agroindustrial residues with the best protein potential were cacao husk and coconut cake, with crude protein levels of 21.8 ± 1.34 % and 21.9 %, respectively. Agroindustrial residues with fibrous potential were rice husk, palm fiber and heart of palm husk with neutral detergent fiber (NDF) values of 69.8 ± 4.17 %, 72.6 ± 6.45 %, and 60.4 %, respectively. However, only the heart of palm husk reported a regular IVADMD value (57.2 %).

Agroindustrial residues from the region of San Martín, Peru, have varied energy and protein use potential and would be useful in cattle feeding in the tropics.