Phosphorus (P) is one of the most expensive nutrients in poultry diets as well a source of environmental concerns regarding excessive P in ground water. In order to reduce the total dietary P, the use of exogenous phytase provides a more efficient use of dietary P and a reduced inclusion of feed phosphates. However, mineral sources of P in diets for broiler chickens are still a need to meet the requirement of poultry and prevent P deficiencies. The phosphate levels in the diets will depend on their P content; and P bio-availability is pivotal to formulate diets at a higher precision and avoid excessive P excretion.

Different P sources are mainly used in the poultry diets, such as di-calcium phosphate (DCP) and monocalcium phosphate (MCP). The MCP has a higher concentration of P (22.6%) than DCP (FEDNA, 2010), which content depends on the degree of hydration. The DCP dihydrate contains typically 17.7% P and 24% Ca and DCP anhydrous contains 20.1% P and 27% Ca (FEDNA, 2010). Commercial products labeled DCP are industrial products resulting from the acidulation of rock phosphate, frequently with sulfuric acid, yielding phosphoric acid, which is neutralized with Ca carbonate after purification (Lima et al., 1997). Alternatively, tri-calcium phosphate (TCP) provides a Ca:P ratio close to the recommended dietary ratios for growing chickens (17.4% P, and 34% Ca). However, a limited amount of information exists regarding their P availability or retention.

Phosphorus is also a nutrient that linearly reduces growth performance and bone characteristics when levels below P requirements are applied (Hamdi et al., 2015a). Based on this property, different methodologies exist to measure in vivo bio-availability of P or to compare P sources including bio-assays based on growth, bone weight and bone ash weight (Ravindran et al., 1995; Lima et al., 1997). These procedures are useful relative measurements for the comparison of different sources of Ca and P (Leske & Coon, 2002). Different sources have been used as a P reference in bio-assay studies, monosodium phosphates (Shastak et al., 2012), MCP (De Groote & Huyghebaert, 1997) or DCP (Ravindran et al., 1995; Fernandes et al., 1999; Coon et al., 2007). It is believed that P in MCP is more digestible than P in DCP (Grimbergen et al., 1985; Eekhout & De Paepe, 1997). Therefore, FEDNA feed evaluation tables describes higher availability for MCP than for DCP. However, Rostagno (2017) described similar P availability and digestibility in broilers for MCP, DCP and TCP. Because most feed phosphates designated as MCP or DCP are mixtures of MCP and DCP, differences within a source may also exist (Petersen et al., 2011).

Consequently, in this study we tested the hypothesis that different feed phosphates, varying in the amount of P and Ca, will affect the dietary P retention and overall productivity of broiler chickens. Two experiments were conducted to test this hypothesis and the objectives were to evaluate the effect of the P source provided at different levels in broiler diets on the availability of P and their effects on performance and bone mineralization for broilers. In the first experiment, P availability was compared between MCP and four different batches of DCP in broilers chickens up to d 21. In the second experiment, the inclusion of a source of TCP as a single source of Ca and P was compared to MCP in broiler diets with respect to their effects on performance, bone mineralization and P retention in broilers up to d 35.

All the animal experimentation procedures used in the two experiments were approved by the animal Ethics Committee of the Universitat Autònoma de Barcelona and were in compliance with the European Union guidelines for the care and use of animals in research (European Parliament, 2010).

Present results showed that feed consumption, bird performance and bone mineralization were decreased when low NPP levels (3.0 to 3.5 g/kg) were included in the diet as compared to 4 and 4.5 g P/kg. Similar results were observed by Akter et al. (2016) on the growth performance, and by Viveros et al. (2002) on bone mineralization. Through its involvement in metabolic and structural processes, P is essential for animals to attain their optimum genetic potential in growth and feed efficiency as well as skeletal integrity and development (Applegate & Richert, 2008). Thus, based on the effect a deficiency has on the body, P is classified as nutrient type II. Individuals with a Type II deficiency are stunted in growth and have no visual signs or differences from 'normal' individuals.

In our experiment, no differences in animal performance were observed between broilers fed the 4.0 and 4.5 g NPP/kg diets. However, greater NPP levels, from 4.0 to 4.5 g/kg, increased tibia weight and tibia ash (mg/tibia). Hamdi et al. (2015a) concluded that a level of 3.8 g NPP /kg improved the growth of chicks and increased bone mineralization on d 14, but no further increases were observed with 4.5 g NPP/kg in diets including the addition of phytase. The lower required levels observed for this experiment as compared to the actual results may be due to the effect of phytase inclusion in the diet and also the duration of the experiment.

Yan et al. (2001) reported that NPP requirements for BW gain and feed conversion were considerably less than required for tibia ash for broilers of 3 to 6 weeks of age in the absence of phytase. They suggested NPP levels of 0.330%; 0.186% and 0.163% to optimize tibia ash, BW gain and G:F ratio, when no phytase was included in the diet. However, with 800 FTU (phytase activity is expressed as phytase units or FTUs; one FTU is the activity of phytase required to liberate 1 µmol of inorganic phosphorus per minute at pH 5.5 from an excess of 15 M sodium phytate at 37°C), phytase diets, the suggested NPP were lowered to 0.240%, 0.151% and 0.109% respectively to optimize tibia ash, BW gain and BW G: F ratio. Ravindran et al. (1995) also observed that bone-mineralization criterion is a good, sensitive indicator of the P status of the birds. Despite phosphorus is being largely contained in whole-body tissues, bone is the main storage organ for P, containing 85% of the body's total P.

Létourneau-Montminy et al. (2008) showed that P deficiency may be aggrieved by high dietary Ca concentrations. In the present study the levels of Ca were formulated to remain constant, which increased the Ca:P ratios as lower were the NPP values. Hamdi et al. (2015a) also described lower performance with the use of diets with high levels of Ca (9 g/kg) containing limiting values of NPP (2.5 g/kg) for broiler at 14 d of age as compared to lower levels of Ca (5-7 g/kg). Al Masri (1995) showed that greater dietary Ca levels and its ratio to P may decrease P absorption. Calcium has the capacity to interact with organic P in the gut lumen limiting mineral and P absorption (Lonnerdal et al., 1989) and can have interference with other micromineral absorption (Simpson & Wise, 1990). Calcium may decrease the amount of P in a suitable form for absorption through the formation of precipitates of orthophosphate in the gut and of insoluble complexes with the phytate molecule (Wise, 1983; Maenz et al., 1999). Furthermore, soluble Ca may increase the intestinal pH and reduce mineral solubility and availability, as reported by Shafey & McDonald (1991). Ca-phytate complexes precipitate at pH values between 4 and 6, which coincide with the pH values of the intestine, where metal ion absorption, occurs (Tamim & Angel, 2003). Létourneau-Montminy et al. (2007) concluded in a literature review including 158 treatments from 14 references, that reduction in dietary P and Ca to (6.0 g Ca/kg and 3.1 g NPP/kg) allowed similar performance and bone mineralization to currently recommended levels by NRC (1994) (10.0 g Ca/kg and 4.5 g NPP/kg).

Considering the differences on P solubility observed in the literature among different phosphate sources, and the likely interactions among minerals on P precipitation in the poultry digesta, we hypothesized that phosphate sources would differ on P availability. Therefore, our study focused on the comparison of the broiler performance and bone mineralization from MCP, DCP or TCP sources, as well as the comparison among different batches of DCP. Diets were prepared on a total P basis in order to allow that likely differences on their P availability (between MCP and DCP in Exp.1, and between MCP and TCP in Exp. 2) could promote differences on performance or bone mineralization.

Results indicated that no differences were observed among P sources (MCP, DCP and TCP) with respect to their effects on the productive performance and bone mineralization of broiler chicken at ages of 21 and 35 days. Lima et al. (1997) also confirmed the lack of differences on broiler performance when they evaluated seven DCP sources from different origins. Similarly, Shastak et al. (2012) did not observe significant effects of the P source on feed intake and BW gain for broilers at 3 and 5 weeks of age, when comparing mono-sodium phosphate and DCP anhydrous in the diets. However, Gillis et al. (1962) reported higher P availability in purified grade MCP diets than in DCP diets. The authors suggested the differences on the hydration degree of P sources as explanation to differences on P availability. Specifically, P in the anhydrous DCP form is less available for poultry than the hydrated salt. It is noteworthy to describe that during the DCP manufacturing,conditionsincludingtemperatureare responsible for the formation of the dihydrate or anhydrate product. Grimbergen et al. (1985) showed that the growth response was lower when anhydrous-DCP was included in the diet as compared to MCP or hydrated-DCP. In addition, Grimbergen et al. (1985), did not detect any difference in the growth response between MCP and hydrated DCP. According to Rucker et al. (1968), the dihydrate form of DCP dissolved more rapidly in an acid environment than the anhydrous form.

Moreover, Lima et al. (1997) also suggested that the particle size of phosphate may have a role also on P availability. Specifically, the larger particle size phosphate sources are retained longer in the gizzard than smaller particles, likely allowing a higher solubility and availability (Burnell et al., 1990; Lima et al., 1997). This criterion of particle size was not registered or evaluated in this study.

No significant difference was either observed between high MCP and TCP in animal performance and bone mineralization in Exp. 2. However, Wilcox et al. (1954) reported that the P in TCP was poorly utilized by the turkey poults. The use of TCP for young broilers decreased the P retention. However, no difference was observed between high MCP and TCP in the grower phase. Hamdi et al. (2015b) showed that the supplementation of broilers diet with TCP, improved FI and WG at d 14 but ileal digestibility of Ca and P was lower for diets including TCP and limestone as compared to diets including Ca chloride, MCP and sodium phosphate. This difference could be associated with solubility of the P sources in birds of different ages.

The results of this experiment also showed that the dietary P retention (% values) is increased when the levels of NPP are lower in the diet. However, the results were not affected by the P source (MCP, TCP). Result is consistent with Leske & Coon (2002) who demonstrated that the retention of P from different P sources depends on the amount of the inorganic P included. They found that P retention from MCP declined from 98% to 59% when NPP was increased from 1.6 to 4.5 g/kg. Wasserman & Taylor (1973) suggested the existence of a saturable component in P absorption, which can be responsible for the decrease in P absorption in the intestine.

Our results allow us to conclude that despite the highly different physical structure and chemical properties among P sources (MCP, DCP and TCP), no evidences were observed of differences in their vivo P availability in growing broilers. Increasing the dietary P have a direct impact on performance and bone mineralization, but higher NPP levels are required in the broiler diets to optimize bone mineralization than to optimize growth performance.

We would like to acknowledge the farm staff of the Servei de Granges i Camps Experimentals of the UAB for their support and support and assistance during the above described experiments. This manuscript has been proof-read by Dr. Maurice Pitesky from the UC Davis School of Veterinary Medicine-Cooperative Extension, Department of Population Health and Reproduction. Manel Hamdi is also grateful to the Animal Nutrition and Welfare Service (SNiBA) for the provision of a research fellowship.