The experimental procedures were approved by the Instituto Valenciano de Investigaciones Agrarias (Moncada, Valencia) Animal Ethics Committee and fulfilled the European regulations for the Care and Use of Animals for Scientific Purposes. Forty Pietrain × (Landrace × Large White) finishing pigs (20 males and 20 females) were randomly selected at 76.8 ± 4.19 kg body weight (BW) and allocated individually in 3 × 3 m pens. Four experimental diets were formulated to contain different levels of ECP on a DM basis: 0 (basal diet), 50, 100, and 150 g/kg of diet. To obtain ECP, the original fresh citrus pulp obtained from a single juice citrus industry was ensiled in a trench silo for 20 days. This citrus pulp contained only orange wastes.

The basal diet consisted in a commercial pelleted diet for finishing pigs. The composition of the basal diet used in this study is shown in Table 1. The diets with 50, 100, and 150 g of ECP/kg (dry matter basis) were manufactured every 2 days by mixing the basal diet with the corresponding amount of ECP. Additionally, to reduce the differences in terms of crude protein (CP) and lysine (Lys) content among diets caused by the addition of ECP, different amounts of soybean meal and L-lysine were added to the diets containing ECP. Concretely, per kilogram of diet we added 28, 40 and 50 g of soybean meal and 0, 0.5 and 1 g of lysine in diets including 50, 100 and 150 g ECP/kg, respectively. All the ingredients (basal diet, citrus pulp, additional soybean meal and crystalline Lys) were mixed by using an automatic mixer of 150 L capacity in order to obtain a homogeneous mixture. Calculated and analysed nutrient concentration of the experimental diets and the ECP is shown in Table 2.

The trial lasted 5 weeks (experimental period) in which pigs were fed the experimental diets. At the beginning of the experimental period pigs weighed 91.5 ± 6.09 kg BW. Before the start of the trial (pre-experimental period), pigs were allowed one week to become accustomed to the pens (week –2) and one week to get adapted to the smell, texture and taste of ECP itself (week –1). During the week of adaptation to ECP (week –1), this was provided as free choice (separated from the feed) and ad libitum in the groups of pigs that were going to be fed the diets containing ECP. A fixed amount of 2.70 kg DM feed/d (2.5 times approximately maintenance requirement for net energy considering the energy content of the basal diet; NRC, 2012) was provided in two equal doses at 08:00 h and 12:00 h during all the experimental period. Water was provided ad libitum throughout the study. At the end of the study (125 ± 7.6 kg BW) all animals were slaughtered and carcass characteristics were recorded.

Pigs were weighed at allocation, at the beginning of the experimental period and the day before slaughter. The digestibility balance was performed on the last week of the experimental period using only males (n = 20) and males were also weighed 2 days prior the digestibility balance. The coefficient of total tract apparent digestibility (CTTAD) of nutrients was determined by recording average daily feed intake (ADFI) and total production of faeces for three consecutive days using the Velcro bag methodology (Van Kleef et al., 1994). A representative sample of each diet (1,000 g) was collected daily during the digestibility balance and stored at -20ºC until subsequent analyses. Faeces produced by each individual pig were weighed and homogenized every 24 h, and a sample (300 g) was stored at –20ºC until subsequent analyses of nutrient and volatile fatty acids (VFA).

In addition, faecal samples were aseptically removed directly from the rectum of 8 animals (4 males and 4 females) per treatment before the pre-experimental period (week –3) and on weeks +2 and +5 of the experimental period for bacterial (enterobacteria and lactobacilli) enumeration. These bacterial populations were used as indicators of gut health with a low number of Enterobacteriaceae and high number of lactobacilli being considered beneficial (Canibe & Jensen, 2003; Ewing, 2008; Ivarsson et al., 2012).

At the end of the study, pigs were fasted for approximately 20 h and slaughtered. On the day of slaughter, animals were transported from the pig house to the slaughterhouse and rested for a minimum of 1 h before slaughter. The pigs were stunned by CO₂, exsanguinated, scalded, cleaned, and eviscerated within ~30 min. After 2 h of sacrifice, the carcasses were placed at 4°C in a chilling room for 24 h. Carcass weight was recorded immediately after slaughter (hot carcass weight). At approximately 2 h post-mortem (at the chilling room), fat depth was measured with a metal ruler at the level of the gluteus medius in the left part of the carcass. Also at this time, a sample of subcutaneous fat was removed at the level of the second cervical vertebrae and the fatty acid (FA) profile of the two fat layers was determined by gas chromatography (O’Fallon et al., 2007). The experimental unit was the individual pig for all variables.

Before analysis, feed and faecal samples obtained during the digestibility trial were pooled across days for each pig. Feed samples were analysed for DM, ash and CP following AOAC (2000) procedures. Neutral detergent fibre (NDF), acid detergent fibre (ADF) and acid detergent lignin (ADL) content were analysed sequentially with a thermo-stable amylase pre-treatment (Van Soest et al., 1991). Gross energy (GE) was determined using adiabatic calorimetry (Gallenkamp Autobomb; Loughborough, UK). Composition of ECP was analysed before the start of the trial for DM, CP and ether extract (EE) using the method outlined by AOAC (2000), and for NDF and ADF as indicated for the diets. Additionally, the neutral detergent soluble fibre (NDSF) content of ECP was analysed using a modification of the methodology proposed by Hall et al. (1997) described by Martínez-Vallespín et al. (2011). Briefly, NDSF includes soluble and insoluble pectins, soluble hemicelluloses (arabinoxylans and β-glucans), fructans and oligosaccharides and was calculated as the difference between the organic matter (OM) fraction and the starch, protein, and NDF (corrected by CP content) fractions, all included in the ether-ethanol insoluble residue. Faeces were analysed for DM, ash, CP, GE, NDF and ADF as indicated for the diets. Additionally, the volatile fatty acid (VFA) concentration in the faeces collected in the digestibility trial was analysed by gas chromatography, with the addition of 4-metil valeric as internal standard (Jouany, 1982).

Immediately after collection, faecal samples for bacterial counts were diluted 1:10 (1 g faeces in 9 mL of peptone water) and decimal dilutions of each sample were prepared. The number of colony forming units (CFU) of enterobacteria and lactobacilli were counted by plating 0.1 mL of each dilution in selective media. Enterobacteria were isolated on McConkey agar (Liofichem, Roseto degli Abruzzi, Teramo, Italy), following aerobic incubation at 37ºC for 24 h. Lactobacilli were cultured on Man, Rogosa, Sharp agar (MRS, Liofilchem, Roseto degli Abruzzi, Teramo, Italy) following incubation at 30ºC for 72 h. All colonies on the MRS plates were counted as lactobacilli.

Data were analyzed using SAS System Software (SAS Inst. Inc., Cary, NC, USA). Differences in BW, ADFI, CTTAD, VFA content in faeces and carcass quality traits among experimental treatments were tested by analysis of variance using the GLM procedure of SAS, and applying polynomial orthogonal contrasts to test linear (L) and quadratic (Q) effects. Changes in bacterial counts with time were analysed according to a repeated measures model (MIXED procedure of SAS). Enterobacteria and lactobacilli counts were log₁₀ transformed before analysis. In all cases, the individual pig was considered the experimental unit and the treatment was considered the main factor in the models. For carcass quality traits analyses, sex was additionally included as a block factor in the model.

The CTTAD of OM, CP, GE, NDF and ADF for the different experimental diets was calculated. The digestible energy and digestible protein content of ECP was then estimated by extrapolation to a total substitution (100%) using lineal regression (REG procedure of SAS) between the digestible nutrient content of experimental diets and the substitution rates, following the methodology proposed by Villamide et al. (2003). This method was chosen because, as Villamide et al. (2003) suggested, this is a reliable method for imbalanced feedstuffs since it is based on the effect of the test ingredient in several diets and it does not only depend on only one diet that could have an unusual digestive behaviour. The digestible energy and protein of the extra soybean meal and Lys added to diets with ECP was estimated (14.1 MJ/kg of DM and 451 g/kg of DM for the soybean meal and 20.8 MJ/kg of DM and 964 g/kg of DM for the Lys, respectively; De Blas et al., 2010) and subtracted. The digestibility values of the diet containing 150 g of ECP/kg imbalanced the regression model in the case of energy. Thus, digestibility values of this experimental diet were not considered representative and they were eliminated from the model.

The DM content of the diets decreased from 895 g/kg to 480 g/kg and the gross energy (GE) content decreased from 19.7 to 18.6 MJ/kg of DM as the level of ECP increased from 0 to 150 g/kg of diet (Table 2). In terms of amino acids, the Lys level was similar among diets but the Met+Cys and Thr levels decreased with the inclusion of ECP. Neutral detergent fibre and ADF increased with the inclusion of ECP (NDF: from 197 to 220 g/kg and ADF: from 76 to 101 g/kg).

Male pigs weighed 122 ± 8.5 kg two days before the digestibility trial and feed intake expressed in DM was similar among groups of treatment (Table 3). Additionally, average daily gain (ADG) from males from the beginning of the experimental period to the end of the study was similar among treatments. The CTTAD of OM and CP was not different among treatments (Table 3). However, GE digestibility was reduced as the level of ECP in the diet increased following a L effect (p<0.05). The CTTAD of NDF and ADF increased with the inclusion of ECP in the diet, both following a significant (p<0.05) L effect. The DM content of faeces decreased significantly (p<0.001) as the inclusion of ECP in the diet increased being 292, 289, 277 and 265 g/kg for the animals fed the diets with 0, 50, 100 and 150 g ECP/kg, respectively (data not shown in tables). Following the regression equations obtained to calculate the amount of digestible energy and digestible protein in ECP (digestible energy [MJ/kg of DM] = 15.6 – 8.64x, R² = 0.56 and digestible protein [g/kg of DM] = 120.1 – 86.34x, R² = 0.66), the ECP contained on DM basis 7.0 MJ/kg of digestible energy and 37.7 g/kg of digestible protein. In the case of digestible energy, as previously indicated, ECP inclusion at levels higher than 100 g/kg in pig diets affected the digestibility of energy and nutrients of the basal diet and imbalanced the model in the present study. For this reason this treatment was eliminated from the prediction model.

The concentration and proportions of total VFA in faeces are presented in Table 4. As the Q effects were not significant these were not shown in this table. VFA are considered indicative of the total production and proportion in the gut (Williams et al., 2001). Total VFA content in faeces was not affected by the inclusion of ECP in the diet. Relative to total VFA concentration, acetic acid was the main VFA contributing to total faecal VFA in all diets, but especially in pigs fed with 150 g ECP/kg of diet. The contribution of acetic acid to total VFA in this group of pigs was significantly higher (p<0.05) compared to the rest of treatments, following a clear L effect among treatments (p<0.01). On the contrary, the contribution of the rest of VFA especially propionic and valeric acids to total VFA was higher in the group of pigs fed the basal diet, also following a significant L effect among treatments (p<0.05).

As regards to bacterial counts in faeces, the repeated measures analysis performed showed a significant effect of diet and week for enterobacteria and lactobacilli counts (p<0.01). Enterobacteria and lactobacilli counts before the pre-experimental period (week –3) were similar among treatments. On week +2 of the experimental period, pigs fed 150 g ECP/kg had the lowest enterobacteria counts (Table 5, p<0.05) compared to the other groups of pigs and at the end of the trial (week +5), all the groups fed with ECP showed significantly lower enterobacteria counts compared to the pigs fed the basal diet (p<0.05). Lactobacilli counts also decreased with the inclusion of ECP in diet on weeks +2 and +5 of the experimental period (p<0.05). Consequently, the ratio lactobacilli to enterobacteria was not different among treatments throughout the study.

Carcass weight was not affected by the diet (Table 6). However, carcass yield and fat depth at the gluteus medius level were reduced by ECP following a L (p<0.001) and a Q (p<0.05) relationship, respectively. For fat depth, the maximum levels were detected in the group of pigs fed with 50 and 100 g ECP/kg of feed and the minimum in the group of pigs fed the basal diet. The FA profile in the external layer of the subcutaneous fat was similar for all treatments, except for C18:3 (alpha-linolenic acid) which increased linearly with ECP inclusion. In all groups, the monounsaturated acids were the most important fraction of FA in this layer. In the internal subcutaneous fat layer, the monounsaturated was also the main fraction obtained. Among treatments, monounsaturated FA (especially C16:1, palmitoleic acid, and C18:1, oleic acid) increased and polyunsaturated FA (especially C18:2, alpha-linoleic acid) decreased quadratically (p<0.05) with ECP inclusion. The sex effect was also tested for these parameters, although the results are not shown in tables. Sex effect was significant (p<0.05) for fat depth at gluteus medius level and the monounsaturated and polyunsaturated FA. Females showed a greater fat depth level at gluteus medius and a greater percentage of monounsaturated FA and a lower percentage of polyunsaturated FA than males. However, the interaction diet × sex was not significant.

Ensiled citrus pulp is proposed as an alternative feedstuff for pig nutrition in this study. Its inclusion at levels up to 150 g/kg in finishing pig diets showed effects on nutrient digestibility reducing GE and increasing NDF- and ADF-CTTAD; however, the digestibility of GE was acceptable up to 100 g inclusion/kg of diet. These results are in agreement with those reported in the literature (Baird et al., 1974; Galassi et al., 2004; Watanabe et al., 2010a; Ruiz et al., 2012) in which the inclusion of dried citrus pulp or other pectin-rich feedstuff such as sugar beet pulp lowered nitrogen (N) and EE digestibility but significantly increased fibre digestibility. Also Rijnen et al. (2001) found increased fibre digestion in sows fed diets with sugar beet pulp silage. The fibre fraction of the ECP in this study was highly digestible and above that of other more lignified feed ingredients, which is due to its high pectin and soluble fibre content (Watanabe et al., 2010a). In fact, the NDSF fraction determined in this study for ECP was relatively high compared to that given for other feedstuffs such as sorghum DDGS (Cerisuelo et al., 2012) and corn DDGS (Stein & Shurson, 2009).

Typically, an excess of dietary fibre in pig diets, especially fermentable fibre, reduces feed intake (Kyriazakis & Emmans, 1995; Bindelle et al., 2008) due to the high water holding capacity of specific non-starch polysaccharides such as pectins. In the current study, ADFI was measured only in males during the digestibility trial and no significant differences were observed in terms of DM intake among treatments. Cerisuelo et al. (2010) also observed that after 3-4 weeks of feeding diets with ECP feed consumption stabilized. Accordingly, ADG did not differ among treatments. Other studies in the literature reported no differences in feed intake and weight gain when including up to 150 g/kg dried citrus pulp in pig diets (Amorim et al., 2014). However, Crosswhite et al. (2013) found a reduction in DM feed intake when levels of 150 g/kg (DM basis) of ECP were included in pig diets. According to Cerisuelo et al. (2010), the short (7 days) acclimation period followed in this study could partially explain this decrease in feed intake. On the other hand, it must be taken into account, that due to the low number of replicates (5 pigs per treatment) in the present study, data on ADFI in the present study should be carefully interpreted.

The digestible energy values for dried citrus pulp in pigs reported in the literature range from 9.6 to 12.1 MJ/kg of DM (De Mol, 1992). O’Sullivan et al. (2003) estimated the digestible energy value of dried citrus pulp for growing pigs to be 12.0 MJ/kg of DM. Watanabe et al. (2010a) reported digestible energy values for dried citrus pulp to be around 10.4 MJ/kg of DM). Both values are lower than those reported for most cereals used in animal nutrition. So far, no research has been published on the digestible energy content of ECP. In the present study, ECP had a low digestible energy content (7.0 MJ/kg DM), although higher than the estimated for other fibrous feedstuffs such as oat hulls, olive pulp or grape pulp, and comparable to those of alfalfa hay, soybean hulls and other tropical tree foliage (Leterme et al., 2005; De Blas et al., 2010). One of the reasons that could explain this low energetic value of ECP as compared with fresh or dried citrus pulp is a change in chemical composition during the ensiling process. Unpublished data from our group indicates that total soluble sugar content of fresh citrus pulp decreased from 354 to 122 g/kg and NDF content increased from 188 to 301 g/kg during the ensiling process (20 days). Additionally, Cervera et al. (1985) found that water soluble carbohydrates disappeared quickly during the first 10 d of the ensiling process of citrus pulp. In terms of antinutritional factors, Oluremi et al. (2010) reported that the concentration of phytonutrients such as oxalate, flavonoid, tannin, saponin and phytate in citrus fruit peels were below the levels reported to be toxic to livestock species and that fermentation caused a reduction in its concentration. Thus, it seems improbable that the presence of antinutritional factors in ECP can affect its nutritive value. Information on the digestible protein content of citrus by-products in pigs is also scarce in the literature. In the present study, digestible protein of ECP was low (33.8 g/kg of DM), but this was expected since the mean CP content of citrus by-products is generally low and often linked to fibre (Gasa et al., 1992). On the other hand, Watanabe et al. (2010a) reported digestible protein contents even lower for dried citrus pulp (8.9 g/kg of DM), thus indicating that the ensiling process could increase the availability of the inherent low digestible protein of fruits. Overall, in terms of nutritive value, ECP seems a low valuable feedstuff in pig nutrition. As suggested by Noblet & Le Goff (2001), although pigs are able to digest dietary fibre to a reasonable extent, the net supply of energy from fibre digestion can be low. Thus, the use of fibrous ingredients in pig diets as an alternative to cereals may not always be efficient in terms of nutrient contribution. However, as for other by-products, the economical asset of its use in pig production must be also considered.

VFA (mainly acetate, propionate and butyrate) from dietary fibre fermentation contribute to maintain gut integrity and also provide energy to the gut and the whole body of non-ruminants (Anguita et al., 2006). Its production and proportion may change according to diet composition. In the present study, VFA production was not measured but VFA content in faeces was similar among treatments. In terms of VFA proportion, the inclusion of 150 g of ECP/kg of DM increased the proportion of acetate (p<0.05), reaching levels of 0.64 in pigs fed diets with 150 g ECP/kg and decreased the proportion of other acids, especially propionic and valeric acids in faeces. Other studies (Zacharias et al., 2004; Anguita et al., 2006; Fung et al., 2010) reported that acetate is a direct product of soluble carbohydrate fermentation, especially pectins, which is in agreement with the results observed in the current study. Other acids such as valerate and the branched chain acids isobutyrate and isovalerate are usually formed as a result of proteolysis (Macfarlane et al., 1992; Williams et al., 2001). In the present study, valeric acid was significantly reduced with the inclusion of ECP indicating that ECP could decrease the proportion or activity of proteolytic bacteria in the hindgut.

Fermentation of dietary fibre in the gastrointestinal tract results in decreased digesta pH (Jensen & Jørgensen, 1994) and increased production of organic acids, which, in turn, are capable of inhibiting some intestinal pathogens such as E. coli and Salmonella spp. (Montagne et al., 2003; Ivarsson et al., 2012). In our study, the inclusion of 150 g of ECP/kg in the diet reduced enterobacteria counts after 2 weeks of administration of the experimental diets, and this reduction increased with time. A reduction of enterobacteria counts has been also reported when fermented liquid feed is provided to growing pigs (van Winsen et al., 2001; Canibe & Jensen, 2003). Dietary pH and VFA concentrations in the gut are considered the main factors affecting enterobacteria counts in these studies. Previous works with ECP from our group also showed a decrease in enterobacteria counts when feeding diets with 100 g of ECP/kg to finishing pigs (Cerisuelo et al., 2010). Both, the high content of fermentable fibre and the diet characteristics derived from its fermentation (low pH and high VFA content) might explain this result. This effect on enterobacteria is of importance since it may decrease the infection pressure of enteropathogens in animals fed ECP. Lactic acid bacteria can tolerate better these circumstances. However, in the present study, lactobacilli counts were also reduced in treatments with 100 and 150 g of ECP/kg on week +5 leading to no changes in lactobacilli:enterobacteria ratio. Therefore, it seems that the inclusion of high ECP levels into pig diets might induce changes in the gut environment capable of acting against lactic bacteria.

A crucial issue when evaluating the suitability of an unconventional feedstuff for animal nutrition is to know its effects on the final product. Diets that are rich in fermentable fibre are suggested to affect the quality of pork by means of reducing carcass yield or varying carcass fat content (Márquez & Ramos, 2007; Watanabe et al., 2010b; Crosswhite et al., 2013) due to its effects on gut fermentation. It has long been recognized that the weight and the volume of the whole gut, and of particular sections of it such as stomach, large intestine and caecum tend to increase in animals fed fibre-rich feeds (Kyriazakis & Emmans, 1995; Jørgensen et al., 1996; Watanabe et al., 2010b; Ivarsson et al., 2012) thus reducing carcass yield. In the present study, a reduction of two percentage units in carcass yield was observed in animals fed the diet including 150 g ECP/kg compared to animals fed the basal diet. Cerisuelo et al. (2010) also observed a reduction in carcass yield with the inclusion of 100 g ECP/kg in the diet. In the present study, a period of 20 h of fasting was practiced to the animals before slaughter. Thus, differences in gut weight could be related to the observed reduction in carcass yield, although this was not measured.

Fat depth at gluteus medius was quadratically affected by the ECP inclusion. Reductions of fat depth at gluteus medius (Cerisuelo et al., 2010) or at backfat level (O’Sullivan et al., 2003) have been observed with the inclusion of citrus by-products in diets. In the present study, pigs fed with 50 and 100 g of ECP/kg showed higher fat depth values compared to control and pigs fed the diet including 150 g of ECP/kg. Thus, it seems that the inclusion of ECP increases fat deposition at gluteus medius but when inclusion levels reach 150 g of ECP/kg, energy digestibility decreases and also fat deposition. The FA profile of the external subcutaneous fat layer was slighty affected by the addition of ECP in pig diets. However, the proportion of monounsaturated FA increased and that of polyunsaturated FA decreased in the internal layer in animals fed with 50 and 100 g of ECP/kg following a Q effect. Although an increase in the ratio of unsaturated to saturated FA in meat products is desirable due to health implications, the presence of a high percentage of polyunsaturated FA might impair meat quality because of reduced fat consistency (Wood & Enser, 1997) and, probably, also meat palatability. Thus, it seems that the inclusion of ECP at levels up to 100 g/kg diet might be beneficial in terms of carcass quality (high monounsaturated and low polyunsaturated FA), although the percentage of the different groups of FA observed in this experiment for all the treatments were maintained within the usual levels reported for pigs.

In conclusion, including up to 100 g of ECP/kg in finishing pig diets did not affect nutrient digestibility, carcass yield and subcutaneous fat FA profile. However, from the results obtained in the present study, higher inclusion levels (150 g of ECP/kg) are not recommended. The nutritional value of ECP in finishing pigs is low and its inclusion decreases faecal enterobacteria and lactobacilli counts.