Introduction

In livestock terms, the department of Atlántico, Colombia, has a bovine inventory that represents 0.89 % (242,856 animals) of the national inventory (^{Instituto Colombiano Agropecuario [ICA], 2019}), which seems to be a poorly representative figure at the country scale. However, in the departmental economic context, livestock represents an important line in the Gross Domestic Product (GDP), with a local participation of 1.1 % —above important sectors such as mining (^{Departamento Administrativo Nacional de Estadística [DANE], 2019})— in land use that reaches 91.4 % of the total land destined for agricultural activity (278,211 ha). Furthermore, it has a production of 185,000 liters of fresh milk per day (^{Gobernación del Atlántico, 2016}).

For the rainy season 2010-2011, the department of Atlántico, particularly the southern cone, suffered one of the most serious floods in the recent history of Colombia. The water from Canal del Dique, i.e., approximately 2,200 million m³ of water, flooded the municipalities of Santa Lucía, Candelaria, Manatí, Campo de la Cruz, and Suan, and indirectly Repelón, Sabanalarga and Luruaco (^{Jabba, 2011}). This situation implied losses in agricultural production of around 60,984 tons, affecting 61,426 cattle heads and incalculable losses in native and introduced forest species (^{Toro et al., 2016}).

As a technical, economic and social assistance for the flood-generated crisis, Corporación Colombiana de Investigación Agropecuaria - AGROSAVIA (formerly Corpoica), with the support from Ministerio de Agricultura y Desarrollo Rural (MADR) and the Government of Atlántico, implemented the establishment of silvopastoral systems (^{Cajas, Carvajal, Barragán, & Portilla, 2014}), as part of the economic recovery strategy in the affected populations. However, when executing this strategy, due to the climatic and environmental characteristics of the southern cone of the Atlántico department, including climatic seasonality (dry period from December to March), low precipitation (1,089 mm/year) and the influence of the Trade Winds during the dry period (^{Martínez et al., 2005}), a complementary nutritional strategy is required. This will contribute to correct nutritional imbalances and improve nutrient availability, especially during the dry period.

The African palm (Elaeis guineensis Jacq.) (Arecaceae) industry in Colombia, in terms of hectares sown (in development and production), recorded a 3 % growth in the 2015-2016 period (^{Federación Nacional de Cultivadores de Palma de Aceite [Fedepalma], 2016}). This represents an offer of byproducts that have been used in animal feed, such as palm kernel cake (PKC) and oil palm decanter cake (OPDC) resulting from the oil production process (^{Alimon, Zahari, Kelantan, Pengkalan, & Kota Bharu, 2012}). Studies have estimated that approximately 350 kg of OPDC, and about 60 kg of PKC are generated per ton of crude palm oil production (^{Van Dam, 2017}). For 2017, the production of crude palm oil in Colombia reached 1,627,550 tons (^{Federación Nacional de Cultivadores de Palma de Aceite [Fedepalma], 2018}), which is equivalent to an estimated production of 569,964 tons of OPDC and 97,653 tons of PKC.

The attainment of the byproducts PKC and OPDC is generated at profit plants. During the processing flow (production of crude palm oil), the OPDC is obtained from the digestion and pressing of entire (whole) fruits, and later attainment and decantation of sludge and effluents; regarding fruit pressing, the nut is collected and then it undergoes a pressing process. In some cases, washing with a solvent is carried out to generate the PKC (^{Ali Hassan et al., 2001}; ^{Van Dam, 2017}). In nutritional terms, the PKC is characterized by being a high-value byproduct for ruminant feeding, with a nutritional range in protein content from 14 % to 19 %, neutral detergent fiber (NDF) between 66 % and 78 %, acid detergent fiber (ADF) between 41 % and 52 %, and a crude fat content that can vary from 1 % to 2 % employing a solvent extraction process. On the other hand, higher values between 4 % and 9 % are obtained using mechanical extraction (^{Alimon, 2005}; ^{Alimon et al., 2012}; ^{Bustamante, Campos, & Sánchez, 2017}; ^{Vargas & Zumbado, 2003}).

The PKC has been referenced for use in beef cattle and water buffalos (Bubalus bubalis (L.)), with levels of inclusion in the diet ranging between 30 % and 80 %, and for dairy cattle with percentages from 20 % to 50 % (^{Alimon et al., 2012}). In the case of the OPDC, ^{Zahari, Alimon and Wong (2012)} have reported the use of this byproduct to feed cattle and sheep, with inclusion levels between 40 % and 45 % in the diet, while ^{Gafar, Alimon, Sazili, Man and Abubakr (2013)} reported values of 30 % or more, in goat diet.

Accordingly, the aim of this study was to evaluate the effect of some byproducts of the African oil palm (E. guineensis) oil production process in the production and quality of milk in small dual-purpose producers in the south of the department of Atlántico, Colombia.

Materials and methods

Experiment location and producer selection

The study was carried out in the municipalities of Manatí, Campo de la Cruz and Repelón, in the department of Atlántico, Colombia. The users or participants were selected from the list of beneficiaries of the project "Phase III or Action 8: Assistance from Corpoica to the departmental program for the bovine repopulation and delivery of fruit materials in the southern municipalities of Atlántico" within the livestock component, linked to the establishment of silvopastoral systems (SPS). In total, 12 small producers were selected in the dual-purpose bovine system distributed as follows: six in Repelón, four in Campo de la Cruz and two in Manatí: geographical coordinates 10°26′53″ N and 74°57′33″ W, average temperature of 28.5 °C, altitude of 10 m a.s.l., and annual precipitation of 1,004 mm; Repelón: coordinates 10°29′40″ N and 75°07′27″ W, average temperature of 28.9 °C, altitude of 9 m a.s.l., and annual precipitation of 927 mm; and Campo de la Cruz: coordinates 10°22′40″ N and 74°52′53″ W, temperature of 28.4 °C, altitude of 7 m a.s.l., and annual precipitation 1,063 mm. The areas of the three municipalities are classified as tropical dry forest, according to ^{Holdrige (1971)} (figure 1).

Source: Elaborated by the authors using Google® images

Figure 1 Spatial location of the selected users or participants in experiments I, and II. Manatí 

In the period between July and October 2013 (rainy period), each farm established between 2.0 and 4.9 hectares in silvopastoral systems. The arrangement was comprised by 1-meter-wide strips (600 planting sites per strip, 0.5 m between streets × 0.5 m between plants) established with shrubs as Leucaena leucocephala (Lam.) de Wit (Fabaceae) and Crescentia cujete L. (Bignoniaceae), spaced every nine meters and associated with the Brachiaria brizantha grass (Hochst. ex A. Rich.) Stapf cv. Toledo (Poaceae). The establishments were carried out with mechanical intervention (renovation) and with Cynodon spp., in interventions with zero tillage (recovery). For the high stratum, timber species adapted to local conditions, such as Tabebuia rosea (Bertol.) DC., and Handroanthus chrysanthus (Jacq.) S.O. Grose (listed as Tabebuia chrysantha) (Bignoniaceae), were established (25 m between streets and 25 m between trees) (^{Cajas et al., 2014}).

The selected farms were characterized by having an average extension of 12.8 ± 6 hectares, with a bovine population ranging between 11 and 40 animals, which classifies them as small producers (^{Federación Nacional de Ganaderos [Fedegan], 2006}). Each farm was divided into six paddoks with sps and between 4 and 10 additional paddoks with grass monoculture (Bothriochloa pertusa (L.) A. Camus, Cynodon nlemfuensis Vanderyst, Dichanthium aristatum (Poir.) CE Hubb., and Cyperus rotundus L., among others). Rotational grazing was handled with 3 and 5 days of occupation and 25 to 35 days of rest. In the farms of these producers, an extractive production of the soil without fertilization or mechanization prevails. The average milk production was 6.7 liters for the rainy season and 3.6 liters for the dry period, with a herd characterized mostly by zebuine animals (Brahman and Gyr), with undetermined crosses (absence of genealogical records) with taurine races (Holstein and Swiss Brown). About 33 % of the selected users (participants) declared using traditional diet supplementation to withstand the dry season, mainly focused on the use of cut grass and sugarcane molasses. About 85 % indicated using mineralized salt (commercial or processed based on a mineral premix), of which only 70 % reported its use throughout the year.

Diets, animal selection and experimental design

Based on the byproducts palm kernel cake (PKC) and oil palm decanter cake (OPDC), two experimental diets were generated and were compared with a control diet (mineral supplementation with 8 % phosphorus). Each of the experimental diets included oil palm byproducts and the addition of an energy source (sugarcane molasses), as recommended by ^{Bustamante et al. (2017)} and ^{Zahari et al. (2012)}, as well as a source of non-protein nitrogen (urea) and sulfur (ammonium sulfate) (table 1). Each animal was given daily, 200 g of mineralized salt, and 1,782 g of the experimental diet (PKC or OPDC).

The chemical characterization of the experimental diets and the forage offered was carried out by establishing crude protein (Kjeldahl method), NDF and ADF (^{Van Soest, Robertson, & Lewis, 1991}), ashes (organic matter combustion), in situ digestibility of dry matter employing the nylon bag technique (^{Ørskov, Hovell, & Mold, 1980}) and protein fractionation (^{Licitra, Hernandez, & Van Soest, 1996}). These were carried out at the AGROSAVIA Animal Nutrition Laboratory, in the research center (CI) Turipaná. The mineral composition, including N, P, K, Ca, Mg, Na, S, Fe, Cu, Mn, Zn, and B, was quantified by atomic absorption spectrophotometry in the Soil and Water Laboratory of AGROSAVIA, CI Tibaitatá.

Table 1 Ingredients and chemical characterization of the experimental diets and the forage offered 

^aAverage result, two bromatological characterizations in each experiment.

^bAverage result, six bromatological characterizations in each experiment.

Source: Elaborated by the authors

The botanical characterization in experiment I recorded an average botanical composition of 57 % grasses, 7 % herbaceous legumes, 40 % undesirable species, a forage supply of 1,618.5 kg of available DM/ha, and 971 kg of usable DM/ha. In the farms linked to this experiment, the dominant grass species was Brachiaria brizantha cv. Toledo. For experiment II, the average botanical composition was 89 % grasses, 2 % herbaceous legumes, 9 % undesirable species, a forage production of 1,882.5 kg of available DM/ha, and 1,713.4 kg of usable DM/ha. The dominant grass in experiment II was Cynodon spp.

Due to the absence of zootechnical records in the selected farms, a reproductive diagnosis was made by rectal palpation, and the physiological and productive status of the cows was established. Based on this information, the cows in a lactating state were selected, identifying their third lactation period and number of births, information that was used as a covariate in the data analysis.

Although the farms had a relative closeness and were influenced by a similar climatic condition, there were particular conditions among them that marked differences in the response variables (i.e., rotation of grasslands, genetic groups, lactations, among others). Based on this, the use of a Latin square design with crossover and balanced for residual effects was proposed (^{Cochran, Autrey, & Cannon, 1941}; ^{Lehmacher, 1991}). In this design, two block factors were considered: one corresponding to the farms, and another to the evaluation periods (three periods, 15 days of accustoming and five days of evaluation); to do this, two complete replicas of the experiment were carried out, using six farms in Repelón (experiment I), two farms in Manatí, and four in Campo de la Cruz (experiment II).

In each of the experiments and per farm selected, one of the six possible supplementation sequences generated to guarantee degrees of freedom and evaluate the drag or carry effect, were randomized (table 2). The drag or carry effects were constructed following the guidelines described by ^{Cerón-Muñoz, Galeano-Vasco and Restrepo-Betancur (2013)}.

Table 2 Assignment of supplementation sequences for farms in experiments I and II 

MSalt: Mineral salt, PKC: Palm Kernel Cake, OPDC: Oil Palm Decanter Cake

Source: Elaborated by the authors

Results and discussion

Table 3 shows the results obtained for the supplementation with byproducts of the African palm industry on milk production and quality under small producer conditions in the south of the department of Atlántico.

Table 3 Production and compositional quality of milk in smallholder farms in the south of the department of Atlántico 

MSalt: Commercial mineralized salt 8 %; PKC: Palm kernel cake; OPDC: Oil palm decanter cake; Carry a: Carry effects in crossover influenced by MSalt; Carry b: Carry effects in crossover influenced by PKC; NB: Covariate number of births; DM: Covariate days in milk; RMSE: Root mean square error; CV: Coefficient of variation; TS: Total solids; MUN: Milk urea nitrogen. Different letters in the same column indicate statistical differences between averages according to Tukey’s test.

Source: Elaborated by the authors

The treatment significantly affected (p < 0.05) milk production (L/day) in favor of OPDC in experiments I and II. In neither of the two experiments, a significant difference (p> 0.05) was observed with PKC-based supplementation. However, the difference (p< 0.05) with the control treatment (MSalt) was consistent in experiments I and II, with 10.07 % and 14.88 % more milk production in OPDC, respectively.

The average proportion of fat in milk had a significant effect (p< 0.05) of the diet evaluated in experiments I and II. In the first case, the PKC diet showed the highest average fat value, with no significant difference (p> 0.05) with the OPDC treatment, but superior to the MSalt treatment (4.31 % vs. 3.85 %, respectively). Contrary to this, in experiment II, the PKC treatment registered the lowest proportion of fat in milk, with 0.33 percentage points less (p< 0.05) of fat compared to the MSalt and OPDC treatments.

Both the protein and the TS were only affected by the treatment in one of the two experiments. In the case of protein, the PKC recorded the lowest percentage in experiment I, with a significant difference (p< 0.05) of the percentage of protein in milk recorded for the OPDC and MSalt treatments. Concerning TS, in experiment II, the average value recorded in OPDC (12.76 %) and MSalt (12.78 %) treatments significantly exceeded (p< 0.05) the record reported in the PKC treatment (12.49 %).

In the concentration of urea nitrogen in milk, the OPDC-based supplement in experiment i recorded the lowest amount of MUN, with a reduction of 25 % and 32 % compared to MSalt and PKC supplements, respectively. In experiment II, the lowest concentration in MUN was observed for PKC supplementation, with a significant difference (p< 0.05) from the concentration recorded in MSalt and OPDC supplements.

The proposal to consider a crossover design by alternating the experimental subjects for each of the farms evaluated was a robust scheme to develop response research in milk production and quality based on byproducts of the African oil palm industry, under the conditions of small producers in the south of the department of Atlántico. ^{Engstrom, Sanchez, Stone and St-Pierre (2010)} report that research on farm demands robust experimentation models, among which they mention the crossover design, with the advantage of carrying out minimal changes in the daily activities of the farms. Likewise, ^{St-Pierre and Jones (1999)} and ^{Rumosa-Gwaze, Mwale and Chimonyo (2011)} indicated that research on livestock farms generates advantages in the subsequent process of technology adoption, given that the farm research process involves the producer closely, and obliges him/her to appropriate the activities inherent in research (i.e., animal selection, feeding process, response measurement, among others). Quite the opposite occurs with the experimental process in the research center, where the producer does not have tangible access.

Supplementation with OPDC increased milk production from about 10 % and 14 % in both experiments (I and II) compared to the control treatment, and without significant difference from the PKC supplementation. On average, in each supplement, an average production of 3.02 and 4.01 L/cow/day was observed for OPDC, and 2.9 and 3.8 L/cow/day for PKC in experiments I and II, respectively. ^{Holmann et al. (2004)} state that, for the Colombian Caribbean region, supplementation is not a widespread practice. On average, local livestock produce 4.2 L/cow/day with a supplement that does not exceed 0.68 kg/animal/day, mainly using wheat and rice bran, mineralized salt, sugarcane molasses, and commercial concentrates, in inclusion percentages of 16.4 %, 4.7 %, 20.7 %, 11.3 %, and 7.9 %, respectively. For the specific case of the south of the department of Atlántico, ^{Yepes-Vargas and Sarmiento-Moreno (2016)} point out that the average milk production in producers that implemented SPS reached 3.73 L/cow/day, and in those that do not implement this system, lower value of 3.07 L/cow/day are obtained in average.

Regarding OPDC, many references related to its use in milk production and quality for dual-purpose systems are not reported. In this regard, ^{Pallares-Cerchiaro and Medina (2014)}, in research evaluating the effect of glycerin on milk production and quality in the south of the department of Atlántico, indicated that the control treatment (without glycerin) registered 10.4 L/cow/day with the addition of 1 kg of OPDC in cows of high genetic potential for milk production in the tropics. In the case of the PKC, ^{Olafadehan, and Adewumi (2009)}, in an investigation under smallholder farm conditions by supplementing Creole cows in eastern Nigeria with 1.77 kg of supplement that included 40 % of PKC, managed to increase significantly milk production in 27.3 %, compared to the diet without supplementation (2.28 vs. 1.79 kg/cow/day, respectively). Likewise, ^{Bustamante et al. (2017)}, supplementing with 1 kg of PKC plus 0.35 kg of sugarcane molasses, significantly increased daily milk production in water buffaloes (B. bubalis) from 2.59 (without supplementation) to 3.82 L/animal/day.

The compositional quality of milk, in terms of fat, protein, and total solids, was affected by the supplementation scheme in both experiments. However, in all the treatments, the compositional records of milk indicated that, under the conditions evaluated, milk can be classified with high compositional quality (^{Calderón, García, & Martínez, 2006}). Furthermore, it complies with the framework regulation for the commercialization of raw milk (> 3.2 % protein, > 3.5 % fat, and > 12.2 % total solids), established by the Ministerio de Salud y Protección Social (Ministry of Health and Social Protection) (^{Decreto 616, 2006}; ^{Decreto 1880, 2011}). The milk quality results in this work are similar to those reported by ^{Botero, Vertel, Florez and Medina (2012)} for small dual-purpose producers in the north of the department of Sucre, who indicated that under these conditions, the quality of the milk has an average of 4.06 ± 0.13 %, 3.57 ± 0.1 % and 12.3 ± 0.3 % in fat, protein, and total solids, respectively. However, these results are higher than those recorded by ^{Romero, Calderón and Rodríguez (2018)} for three subregions of the department of Sucre, with protein in a range of 2.96 % to 3.29 %, fat from 3.37 % to 3.75 % and total solids from 11.71 % to 12.23 %.

The MUN is considered a metabolic indicator of the energy-protein relationship in the diet and the protein use efficiency in the rumen (^{Danes, Chagas, Pedroso, & Santos, 2013}; ^{Pardo, Carulla, & Hess, 2008}). The results of MUN in the current study were close to the upper limit in the optimal range (10 to 19 mg/dL MUN) cited by ^{Hess et al. (1999)}, except for the OPDC in experiment I. Further, ^{Ávila and Lascano (2001)} indicated that a 10 mg/dL value can be used as a reference point to define MUN and that this value depends on the degree of protein supplementation and the genetic group of the cow. In this sense, responses range from an increase in MUN and milk production as a result of the increase in the level of protein supplementation in the diet —mainly for mongrel cows—, to increases in MUN without response in milk production —in cows with a zebuine phenotype—.

Under the conditions of the savanna microregion of Cesar (department of Cesar, Colombia), and with a basic diet of Brachiaria bizantha cv. Marandú in cows with different crossbreeding between Bos indicus × Bos taurus, ^{Mojica-Rodríguez, Castro-Rincón, Silva-Zakzuk, Hortúa-Castro and García-Quintero (2013)} reported MUN values of 11.04, 13.49 and 17.78 mg/dL in the rainy season, and 9.81, 8.58 and 7.13 mg/dL in the dry period, for the thirds I, II and III of lactation, respectively.

The MUN results obtained in the current study indicate that the diet consumed could affect the nitrogen use efficiency in the rumen, possibly due to a deficiency of fermentable carbohydrates in the rumen. Given the conditions under which this experiment was carried out (on smallholder farms), and the degree of difficulty presented by the methodologies for the establishment of dry matter consumption in grazing for ruminants, it was not possible to quantify the total dry matter intake and nutrient intake. However, studies that evaluated MUN in cows that grazed in SPS with Leucaena spp., indicate that the voluntary intake of this legume influences the total balance (^{Bottini-Luzardo, Aguilar-Pérez, Centurión-Castro, Solorio-Sánchez, & Ku-Vera, 2016}; ^{Prieto-Manrique, Mahecha-Ledesma, Vargas-Sánchez, & Angulo-Arizala, 2018}) and the excretion (^{Piñero-Vázquez et al., 2017}) of the nitrogen ingested in the diet. This, together with an energy deficit, could be the factor that influenced the MUN results of this experiment, even in the control diet (without nitrogen).

Conclusions

The byproducts of the African palm industry, especially the oil palm decanter cake, were able to demonstrate a positive response in milk production increase under the conditions of small dual-purpose producers in the south of the department of Atlántico. However, it is necessary to consider longer-term experiments to assess the sustainability of production, milk quality, and the economic, social and environmental viability of these strategic supplementation schemes under the conditions of small double-purpose producers.