Effects of calcium level and source, acidication, and phytase on phytate degradation and the microbiota in the digestive tract of broiler chickens

Background Diet acidication, dietary calcium ( Ca ) level, and phytase supplementation are known inuences on the microbial community in the digestive tract and on phosphorus ( P ) utilization of broiler chickens. Effects of dietary factors and microbiota on P utilization may be linked because microorganisms produce enzymes that release P from phytate ( InsP 6 ), the main source of P in plant feedstuffs. This study aimed to detect linkages between microbiota and InsP 6 degradation by acidifying diets (i.e., replacing Ca carbonate (CaCO 3 ) by Ca formate or adding formic acid to CaCO 3 -containing diets), varying Ca levels, and supplementing phytase in a three-factorial design. We investigated i) the microbial community and pH in the digestive tract, ii) prececal ( pc ) P and Ca digestibility, and iii) InsP 6 degradation. Results All factors under investigation inuenced digesta pH and the microbiota composition. Predicted functionality and relative abundance of microorganisms indicated that diets inuenced the contribution of the microbiota on InsP degradation. Values of InsP 6 degradation and relative abundance of the strains Lactobacillus johnsonii and Lactobacillus reuteri were correlated. Phytase supplementation increased pc InsP 6 disappearance, with differences between Ca levels, and inuenced concentrations of lower inositol phosphate isomers in the digestive tract. Formic acid supplementation increased pc InsP 6 degradation to myo -inositol. Replacing CaCO 3 by Ca-formate and the high level of these Ca sources reduced pc InsP 6 disappearance, except when the combination of CaCO 3 +formic acid was used. 3 +formic acid led to the highest InsP 6 (52%) in the crop and increased myo -inositol concentration in the ileum digesta. Supplementing phytase leveled the effect of high Ca content pc InsP 6 to Ca and to in 5 Ca level j is the xed effect of the Ca level (5.6 or 8.2 g/kg dry matter); Phytase k is the xed effect of phytase supplementation (0 or 1,500 FTU phytase/kg); block l is the random block effect; and e ijkl is the residual error. The block effect included possible effects of location in the building and sampling time on day 21 or day 22 because four blocks were sampled each day. ANOVA was calculated using the MIXED procedure of the SAS software. Normal distribution and homogeneity of variance were tested prior to statistical analysis. P values described herein result from ANOVA, or t tests when two groups were compared.

functionality and relative abundance of microorganisms indicated that diets in uenced the contribution of the microbiota on InsP degradation. Values of InsP 6 degradation and relative abundance of the strains Lactobacillus johnsonii and Lactobacillus reuteri were correlated. Phytase supplementation increased pc InsP 6 disappearance, with differences between Ca levels, and in uenced concentrations of lower inositol phosphate isomers in the digestive tract. Formic acid supplementation increased pc InsP 6 degradation to myo -inositol. Replacing CaCO 3 by Ca-formate and the high level of these Ca sources reduced pc InsP 6 disappearance, except when the combination of CaCO 3 +formic acid was used.
Supplementing phytase to CaCO 3 +formic acid led to the highest InsP 6 disappearance (52%) in the crop and increased myo -inositol concentration in the ileum digesta. Supplementing phytase leveled the effect of high Ca content on pc InsP 6 disappearance.

Conclusions
The results point towards a contribution of changing microbial community on InsP 6 degradation in the crop and up to the terminal ileum. This is indicated by relationships between InsP 6 degradation and relative abundance of phosphatase-producing strains. Functional predictions supported in uences of microbiota on InsP 6 degradation. The extent of such effects still remains to be clari ed.
InsP 6 degradation can also be in uenced by pH effects caused by dietary concentration and solubility of the Ca sources.

Background
High utilization of phosphorus (P) provided by plant feedstuffs is advantageous because less or no mineral P is needed to ful ll the P requirement of animals. Plant-P is mainly bound in phytic acid [myoinositol 1,2,3,4,5,6-hexakis (dihydrogen phosphate); InsP 6 ] and present as phytate, which requires hydrolyzing enzymes to make P available for the animal. Hence, phytases are widely used as feed additives in non-ruminant nutrition. The main effect of phytases is the cleavage of P from InsP 6 to increase P utilization by animals. When diets are fed with the supplementation of mineral P and calcium (Ca), but without supplemented phytase, prececal (pc) InsP 6 degradation is reduced in broiler chickens [1].
The reduced degradation of InsP 6 upon Ca supplementation is usually explained, among other reasons, by the formation of Ca-InsP 6 complexes in the digestive tract at high Ca concentrations [2,3].
It was shown that bone mineralization of broiler chickens can differ as a result of varying Ca sources when supplied in equal Ca concentrations [4] and that using different Ca sources can have other consequences, such as impacts on the microbial community of the digestive tract of broiler chickens [5].
Studies investigating the impact of dietary Ca on P utilization usually used limestone as a Ca source, which mainly consists of calcium carbonate (CaCO 3 ), a compound with a high buffer and acid-binding capacity. Increasing pH in the digestive tract supports formation of Ca-InsP 6 complexes and, thus, reduces pc InsP 6 degradation [2,6,7]. Reducing intestinal pH might counteract effects of high dietary Ca concentrations. Ca salts of organic acids like Ca-formate are known to have a lower buffer capacity than CaCO 3 [8]. Hence, replacing CaCO 3 by Ca-formate might compensate the pH increase caused by CaCO 3 supplementation and thereby increase InsP 6 degradation and, as a consequence of changing the Ca source, affect the microbial community. Solubility of dietary Ca might also in uence P utilization through Ca-InsP 6 complexes or other mechanisms. Complexation of Ca and InsP 6 was shown to be increased at pH 5 and higher [9]. Using Ca sources with higher solubility than that of CaCO 3 might allow for a higher Ca absorption in the proximal section of the small intestine and thus reduce the amount of Ca cations available for Ca-InsP 6 complex formation.
A decrease in pH could also be achieved by adding formic acid to diets. As reviewed by Kim et al. [10], supplementation of formic acid was reported to affect the microbial community and lead to increased performance and nutrient digestibility in broiler chickens. Likely, the addition of formic acid to CaCO 3containing diets and the use of Ca-formate in uences the microbial community in the digestive tract, possibly through changes in digesta pH [11,12]. Therefore, different microbiota composition might be involved in causing the observed effects of Ca sources, organic acid supplementation, and Ca levels on P utilization.
The present study was conducted to investigate effects of dietary Ca and phytase supplementation on the microbiota of the crop and the terminal small intestine of broiler chickens based on target amplicon sequencing of DNA, and to study consequences on pc InsP 6 disappearance, and pc P and Ca digestibility.
To elucidate the underlying mechanisms, pH and inositol phosphate (InsP) isomers in the digesta of the crop, gizzard, and terminal small intestine were analyzed. It was hypothesized that replacing CaCO 3 by Ca-formate or adding formic acid to CaCO 3 -containing diets decreases the pH and in uences the microbial community, thereby increasing InsP 6 degradation and P digestibility.

Results
The initial group weight at the beginning of feeding the experimental diets, on day 16 of the experiment, was similar for all treatments, at 9,604 g/pen (standard deviation = 15 g/pen, P = 0.183). Mortality during the experimental period was low (0.6% of all animals) and not related to any treatment (six cases in six different treatments).

Growth performance
The average daily gain (ADG) and average daily feed intake (ADFI) were lower compared to other treatments when the Ca level was high and no phytase was supplemented (P < 0.001, Figure 1). The ADFI was signi cantly lower for Ca-formate compared to CaCO 3 and CaCO 3 +formic acid (P = 0.002). Phytase supplementation increased the gain to feed ratio (G:F) by 0.04 g/g (P < 0.001). Further effects on growth performance were not signi cant. pH in the digestive tract In the crop, the highest pH of 5.5 was observed for CaCO 3 . Crop pH was decreased by 0.6 and 0.3 units for CaCO 3 +formic acid and Ca-formate, respectively, compared to CaCO 3 (P < 0.001, Figure 2). With a 0.1unit increase, crop pH was marginally but signi cantly higher for the high compared to the low Ca level. In the gizzard, phytase supplementation increased pH by 0.2 units (P < 0.001) and decreased pH by 0.1 units (P = 0.005) at the low and high Ca level, respectively. In the ileum, phytase supplementation increased pH by 0.8 units at the high Ca level (P < 0.001) but had no signi cant effect at the low Ca level.

Microbial community
In both the crop and the ileum, mainly Lactobacillus species, including L. johnsonii, L. crispatus, L. reuteri, L. gallinarum, and L. vaginalis, were identi ed ( Figure 3 and Figure 4). Streptococcus alactolyticus was highly abundant in the crop and the ileum for diets with CaCO 3 at the low Ca level, irrespective of phytase supplementation. This was re ected by a high similarity between CaCO 3 treatments at the low Ca level and a separation of these treatments from the other diets in a cluster analysis ( Figure S1). Permutational multivariate analysis of variance (PERMANOVA) analyses (Table S3) revealed that the microbial communities in the crop and the ileum were signi cantly affected by all the main effects (P ≤ 0.034), with no interaction being signi cant. The microbial community in birds receiving CaCO 3 differed signi cantly in both sections of the digestive tract compared to CaCO 3 +formic acid or Ca-formate (P ≤ 0.005).
In the crop, between 34% and 53% of the detected relative abundance of microorganisms corresponded to operative taxonomic unit (OTU) 1 (L. johnsonii) and between 11% and 29% to OTU2 (L. gallinarum). We conducted analyses of variance (ANOVA) to get information on treatment effects on single OTUs being aware that explanatory power is partly impinged by physiological interrelationships between microorganisms. Abundance of OTU1 was higher for Ca-formate when phytase was supplemented (P ≤ 0.050). The abundance of OTU1 increased for Ca-formate at the high Ca level. OTU8 and OTU12 (both L. Page 5/32 vaginalis) were more abundant in the CaCO 3 treatment compared to the CaCO 3 +formic acid and Caformate treatments (P < 0.001). CaCO 3 increased the abundance of the OTUs closely related to S. alactolyticus (OTU5, P = 0.006 and OTU10, P = 0.008), L. reuteri (OTU9, P = 0.001; OTU11, P = 0.003; OTU15, P < 0.001 and OTU19, P < 0.001), and Gallibacterium sp. (OTU23, P < 0.001). The high Ca level led to an increased abundance of OTU2, assigned to L. gallinarum (P = 0.007), and OTU4 (P = 0.005), OTU9 (P = 0.036), and OTU11 (P < 0.001), which were assigned to L. reuteri. Phytase supplementation increased the abundance of OTU9, OTU11, and OTU13 (assigned to L. reuteri), and decreased the abundance of OTU2 (L. gallinarum). Crop pH was negatively correlated with the abundance of two out of three OTUs assigned to L. gallinarum (P ≤ 0.008) and positively correlated with seven out of nine OTUs assigned to L. reuteri (P ≤ 0.027, Table S5, Figure S2).
Similar to the crop, the most abundant OTUs in the ileum were OTU1 (L. johnsonii, 24-46%) and OTU2 (L. gallinarum, 15-41%). Phytase supplementation increased the abundance of OTU1 (P = 0.040) and OTU2 (P = 0.015). The abundance of OTU2 was increased in the ileum of chicken fed with the high Ca level diets (P = 0.009). The abundance of OTUs assigned to L. reuteri (OTU6, OTU15, and OTU18, P ≤ 0.049) and L. vaginalis (OTU8, P < 0.001) were decreased for Ca-formate and the abundance of OTU19 and OTU10 (L. reuteri and S. alactolyticus, respectively, P ≤ 0.049) were increased for CaCO 3 . The abundances of OTU7, OTU8, and OTU20 were signi cantly affected by the interaction between Ca level and phytase supplementation (P < 0.049). Unclassi ed Firmicutes (OTU7) and unclassi ed Bacterium (OTU20) were less abundant in the ileum of chicken at the high Ca level without phytase supplementation. Supplementation of phytase leveled this effect. Abundance of OTU8 (L. vaginalis) increased upon an increase in Ca level, but not when phytase was supplemented.

Functional prediction
The broad classi cation hierarchy of KEGG pathways of functions resulted in a similar distribution for crop and ileum. For both sections, the most abundant pathways were assigned to the metabolism category (74-75%), followed by environmental information processing (8-9%), and genetic information processing (8-9 %).
The same P-related pathways were signi cantly in uenced in the crop and the ileum ( Figure 5). No interaction related to genes connected to InsP metabolism was signi cant in the crop and the ileum. In the crop, genes encoding for InsP metabolism were higher in the CaCO 3 than in the CaCO 3 +formic acid and Ca-formate treatments (P ≤ 0.001) and more abundant at the low than at the high Ca level (P = 0.007). In the ileum, the abundance of genes related to InsP metabolism was lower in the Ca-formate treatment than in the CaCO 3 and CaCO 3 +formic acid treatments (P ≤ 0.011) and not signi cantly in uenced by Ca level. Phytase supplementation had no signi cant effect on InsP metabolism in the crop and the ileum. Other signi cantly in uenced pathways were mineral absorption, the phosphotransferase system, the phosphatidylinositol signaling system, and phosphonate and phosphinate metabolism in the crop and the ileum. Phytase supplementation had a signi cant in uence on the abundance of genes related to the phosphatidylinositol signaling system and phosphonate and phosphinate metabolism in the crop and in some treatments in the ileum, as indicated by a signi cant three-way interaction. The other pathways were signi cantly in uenced by acidi cation, Ca level, or the two-way interaction between acidi cation and Ca level.
Within the category of InsP metabolism, increasing the Ca level reduced abundance of genes coding for myo-inositol-1(or 4)-monophosphatase in the CaCO 3 and Ca-formate treatments (P ≤ 0.018), but not in the CaCO 3 +formic acid treatment (P = 0.520) in the crop ( Figure 6). In the ileum, myo-inositol-1(or 4)monophosphatase was increased by phytase supplementation (P = 0.040). Increasing dietary Ca had no effect on myo-inositol-1(or 4)-monophosphatase in the CaCO 3 and CaCO 3 +formic acid treatments (P ≥ 0.269) and decreased myo-inositol-1(or 4)-monophosphatase in the Ca-formate treatment (P = 0.008). Myo-inositol-1-phosphate synthase coding genes were lower in the CaCO 3 +formic acid and Ca-formate treatments than in the CaCO 3 treatment in the ileum (P ≤ 0.001). Other genes annotated to InsP and myoinositol degradation in the KEGG database were not signi cantly in uenced by the treatments used in this study.

InsP 6 disappearance and prececal digestibility of P and Ca
In the crop, phytase supplementation increased InsP 6 disappearance in the CaCO 3 treatments with and without formic acid by 20 and 37 percentage points (pp), respectively (P < 0.001), but not in the Caformate treatment (P = 0.090, Figure 7). Without phytase supplementation, Ca level did not affect InsP 6 disappearance in the crop (P = 0.536). At the high Ca level, the effects of phytase supplementation on InsP 6 disappearance in the crop increased by 6 pp to 37 % (P = 0.025).
Phytase supplementation increased pc InsP 6 disappearance (P < 0.001) to approximately 80 %, irrespective of the dietary Ca level. Without phytase supplementation, pc InsP 6 disappearance was 9 pp higher for the low compared to the high Ca level (P < 0.001). Ca level had no signi cant effect on pc InsP 6 disappearance for CaCO 3 +formic acid, but the high Ca level decreased pc InsP 6 disappearance by 4 pp for CaCO 3 (P = 0.047) and by 10 pp for Ca-formate (P = 0.016). Increasing dietary Ca decreased pc P digestibility. This effect was more pronounced for Ca-formate, with 12 pp (P < 0.001), than for CaCO 3 and CaCO 3 +formic acid, with 7 pp each (P < 0.001). The three-way interaction was signi cant for pc Ca digestibility (P = 0.012, Figure 8). The pc Ca digestibility in CaCO 3 and CaCO 3 -formate was increased by phytase supplementation at the low Ca level (P ≤ 0.012) but was not affected at the high Ca level (P ≥ 0.160). Phytase supplementation increased and decreased pc Ca digestibility for CaCO 3 +formic acid at the low and high Ca level, respectively (P ≤ 0.002).

Inositol phosphate isomer and myo-inositol concentrations
Signi cant interactions between acidi cation and phytase supplementation were detected for InsP 6 and two InsP 5 isomers in the crop (P < 0.001, Table S9). Decrease in concentrations of InsP 6 and InsP 5 isomers upon phytase supplementation was most pronounced in the CaCO 3 +formic acid treatment and less in the CaCO 3 and the Ca-formate treatments. Myo-inositol concentrations in the crop were not signi cantly affected.
In the gizzard, InsP 6 concentrations were lower in diets with supplemented phytase compared to those without (P < 0.001, Table S10). Highest InsP 6 concentrations were found in the Ca-formate treatment without phytase supplementation. Phytase supplementation decreased concentrations of InsP 5 isomers below level of detection, while making some InsP 4 and InsP 3 isomers detectable. Phytase supplementation increased myo-inositol concentrations to a greater extent at the low Ca level compared to the high Ca level (P = 0.022). Acidi cation had no signi cant effect on myo-inositol concentrations.

+formic acid and both
Ca levels for Ca-formate when phytase was supplemented. Concentrations of Ins(1,2,3,4)P 4 and InsP 3x were high when phytase was supplemented to CaCO 3 +formic acid and Ca-formate at the high Ca level.
An increase in myo-inositol concentrations upon phytase supplementation was more pronounced in the ileum at the low compared to the high Ca level. The highest myo-inositol concentration was determined for CaCO 3 +formic acid (P ≤ 0.003) with no difference between CaCO 3 and Ca-formate (P = 0.150).

Discussion
Effects of phytase on InsP 6 degradation and subsequent P utilization differ considerably between studies. Possible explanations include the use of acidifying ingredients in the diet and different Ca levels affecting the pH, phosphatase-producing bacteria, the probability that InsP 6 complexes are formed, and the e cacy of phytases in the digestive tract. Most studies only investigated one factor, making potential interactions impossible to detect. This study is the rst to examine the in uence of phytase and acidifying ingredients with different Ca levels on P utilization following InsP 6 degradation, the microbial community, and potential linkages between all traits. Our hypothesis was that replacing CaCO 3 by Ca-formate or adding formic acid to CaCO 3 -containing diets in uences the microbiome and increases InsP 6 degradation.

Crop
Crop microbiota effects on inositol phosphate degradation. In the crop, dietary treatments in uenced the microbial community and had potential in uences on InsP 6 degradation. Diets could have in uenced the microbiota composition and InsP degradation via independent mechanisms or could have in uenced the microbiota composition, which then altered InsP degradation as a consequence. Adding formic acid to CaCO 3 or replacing CaCO 3 with Ca-formate decreased pH in the crop content and concurrently shifted microbiota composition. A connection between pH and abundance of OTUs is indicated by correlations (Table S4) but it is not clear whether pH or other consequences of Ca level and acidi cation were causative. The microbial communities of the CaCO 3 +formic acid and Ca-formate treatments differed from that of the CaCO 3 treatment, while no difference of the microbial community was determined between the CaCO 3 +formic acid and Ca-formate treatments. Changes in pH were not completely concomitant with changes in the microbial community, suggesting that changes in the microbial community were caused by at least one other mechanism in addition to pH reduction. Likely, different mechanisms had an impact on the microbial community when the pH was reduced from 5.5 to 5.2 (CaCO 3 to Ca-formate) and from pH 5.2 to 4.9 (Ca-formate to CaCO 3 +formic acid). The microbial community may also have been affected by formic acid because formic acid is known to inhibit certain bacterial species [13].
Functional prediction and changes in the relative abundance of some bacteria point towards an in uence of the microbial community on P utilization. Phosphatase activity has been described for microorganisms, including L. johnsonii [14] and strains of L. reuteri [15]. Relative abundance of L. johnsonii ranged from 34-53% and depended on acidi cation and Ca level (Table S12). Hence, treatments probably in uenced the contribution of L. johnsonii to P utilization and affected other traits of L. johnsonii; probiotic characteristics are assigned to L. johnsonii, such as reduced adhesion to the epithelial cells and inhibited growth of pathogenic microorganisms in humans [16], and they are further known to positively in uence measures related to the immune system of broilers [17]. Relative abundance of OTUs assigned to strains of L. reuteri summed up to a range of 6-26%. At the low Ca level, the relative abundance of some OTUs assigned to L. reuteri (OTU4, OTU9, OTU11, and OTU13) were less abundant in the crop, which corroborates the nding that Ca can inhibit the growth of L. reuteri [15]. Hayek et al. [15] found that phosphatase activity (probably including phytase) produced by L. reuteri depended on strain and presence of chemical elements in a nutrient solution. Ca increased phosphatase activity of all L. reuteri strains under investigation. This effect was described to be caused by Ca-dependent activation of active sites of phosphatases, including phytase [18,19]. In the present study, it appears possible that lower abundance of most L. reuteri strains were compensated by a higher potential to produce phosphatases by L. reuteri and other microorganisms at the higher Ca level when phytase was supplemented. Irrespective of individual bacteria strains, the InsP metabolism and other P-related metabolic pathways of the microbial community in the crop were in uenced by acidi cation and Ca level. Within InsP metabolism, acidi cation and Ca level in uenced the abundance of genes connected to enzymes involved in degradation of lower InsP isomers and myo-inositol. This strongly indicates that the microbial community contributes to InsP degradation in the crop; however, the extent of this contribution cannot be su ciently quanti ed with the available data.
pH effects on InsP 6 disappearance in the crop. Acidi cation and Ca level effects were observed in diets with phytase supplementation only. With a value of 55%, the highest InsP 6 disappearance in the crop was observed when phytase was supplemented to CaCO 3 +formic acid. This effect seems to have been caused by acidi cation of the crop content by formic acid, which then shifted the pH closer to the optimum of the phytase used (pH optimum of 4.5 according to the product datasheet of the producer). This explanation is supported by a linearly negative relation between crop pH and InsP 6 disappearance in the phytase-supplemented diets (Figure 9). The regression indicates that InsP 6 disappearance in the crop was increased by 5.3 pp per each 0.1-unit reduction of pH in the crop.
Ca solubility effects. Different Ca solubility seems to have an in uence on InsP 6 disappearance in the crop, but clear inferences are di cult to make. Higher solubility can make free Ca cations more available for Ca-InsP 6 complex formation and, hence, reduce InsP 6 disappearance. Ca solubility of Ca-formate was reported to be substantially higher than that of CaCO 3 [20]. InsP 6 disappearance in the crop was higher for CaCO 3 compared to Ca-formate, consistent with the theory of higher Ca-InsP 6 complex formation as caused by higher solubility of Ca-formate. This is supported by the crop pH of 5.2 for Ca-formate and 5.5 for CaCO 3 , both being above the pH critical for complex formation [9]. As acidi cation can increase Ca solubility, CaCO 3 +formic acid probably had higher Ca solubility than CaCO 3 alone as a consequence of formic acid lowering the pH [21]. Formic acid supplementation would then make more Ca available for Ca-InsP 6 complex formation and thus reduce InsP 6 disappearance. However, higher InsP 6 disappearance in the crop was observed for CaCO 3 +formic acid than for CaCO 3 . When Ca-formate was used instead of CaCO 3 , InsP 6 disappearance in the crop was possibly determined more by the higher Ca-InsP 6 complex formation caused by higher Ca solubility than by the lower pH. With formic acid supplementation, acidi cation may have had a higher impact on InsP 6 degradation than the higher Ca solubility.

Ileum and gizzard
Ileum microbiota effects on inositol phosphate degradation. The reducing effect of the high Ca level on pc InsP 6 disappearance for CaCO 3 but not for CaCO 3 +formic acid may be caused by changes in endogenous phytase and other phosphatases produced by epithelial cells or by microbiota. The latter is supported by the effects of the dietary treatments on metabolic pathways of the microbial community connected to InsP 6 degradation. Changes in P-related metabolic pathways in the ileum were similar to that in the crop. The composition of the microbial community varied among treatments. ANOVA analyses of relative abundances of single OTUs revealed signi cant interactions for ve OTUs (1, 5, 7, 8, and 20; Table S13). However, an assessment of the contribution of these microorganisms to InsP degradation is di cult to derive because the phosphatase activity of the associated microorganisms is either not yet described in the literature or the OTUs de ne classes of strains. Therefore, it cannot be estimated whether these microorganisms contributed phosphatase to a relevant extent. Nonetheless, results indicated that microorganisms in the digestive tract have contributed to P utilization of the birds. There were ve positive correlations between pc P digestibility and OTUs assigned to L. reuteri and L. johnsonii; these microorganisms are known to produce phosphatase [14,15]. Another possible explanation for differences in the microbial community could be an alteration of availability of nutrients for the microbiota in the digestive tract as a result of the varied supplements.
Calcium solubility and pH effects on inositol phosphate degradation. High dietary Ca levels reduced pc InsP 6 degradation, but this effect was compensated by phytase supplementation. A decline in pc InsP 6 disappearance with graded levels of CaCO 3 in diets for broiler chickens was previously described [3]. In the present study, effects of acidi cation, Ca level, and phytase supplementation on InsP 6 degradation were interrelated. In contrast, Li et al. [7] and Dersjant-Li et al. [22] did not nd an interaction between CaCO 3 level and phytase supplementation on pc InsP 6 disappearance and pc P digestibility when up to 1,000 and 500 phytase units (FTU) phytase/kg were supplemented, respectively. However, dietary Ca levels in these studies were higher than in the present study. We are not aware of other studies on InsP 6 degradation that compared CaCO 3 with the more soluble Ca-formate. Hamdi et al. [4] determined no signi cant difference in pc P digestibility when 5.5 g Ca/kg was supplied by CaCO 3 or the more soluble Ca chloride. In another series of experiments, CaCO 3 was compared with a calci ed marine seaweed considered a highly soluble Ca source [23][24][25][26]. Results of this series were not consistent, likely owing to different Ca and phytase supplementation levels. Further, comparisons between calci ed marine seaweed and Ca sources investigated in these studies may be impeded by unknown substances contained in the seaweed products. Overall, it seems that one major factor of P utilization affected by Ca sources and levels is the solubility of Ca from differing sources that depends, inter alia, on pH and microbiota. This supports conclusions reached by Kim et al. [21] based on a comparison of CaCO 3 sources.
Differences in pc InsP 6 disappearance between CaCO 3 , CaCO 3 +formic acid, and Ca-formate were smaller compared to InsP 6 disappearance in the crop. A clear determination of pc InsP 6 disappearance as compared to InsP 6 disappearance in the crop has been found in the present study, irrespective of whether phytase was supplemented or not ( Figure 10). This indicates the cleavage of Ca-InsP 6 complexes due to decreasing pH in the proventriculus and the gizzard. Gizzard pH ranged from 2.8-3.3, which was probably su cient for cleavage of Ca-InsP 6 complexes because Ca-InsP 6 complexes occur at pH 4 and higher [27].
No difference in pc InsP 6 disappearance was found between CaCO 3 +formic acid at both Ca levels and the lower Ca level of CaCO 3 . A reason for this could be the higher e cacy of phytase caused by a decreased crop pH and less complete InsP 6 degradation for Ca-formate due to higher Ca solubility. We are not aware of other studies investigating a simultaneous supplementation of formic acid and phytase in poultry. However in pigs, total tract P digestibility was increased to a higher extent when phytase and formic acid were supplemented together than when their separate effects were summed [28].
Effects on inositol phosphate degradation pathways. Concentrations of lower InsP isomers give insights into effects on InsP degradation pathways, including limiting steps in the process of InsP degradation. InsP isomers that limit InsP degradation apparently differed among treatments. When no phytase was supplemented, InsP degradation by endogenous enzymes seems to have been limited by the high Ca level. High Ca supposedly inhibited the activity of endogenous enzymes at different stages of the degradation pathway of InsP 6 and lower InsP isomers depending on the Ca source. This interpretation is based on lower concentrations of myo-inositol and higher concentrations of InsP 6 and InsP 5 isomers in the ileum at the high compared to the low Ca level of the CaCO 3 and Ca-formate treatments. In the CaCO 3 +formic acid treatment, there was no difference in InsP 6 concentrations in the ileum between the Ca levels, but there was slightly higher InsP 5 isomer and lower myo-inositol concentrations at the high Ca level. Obviously, formic acid supplementation overcame a restricted release of the rst phosphate group from InsP 6 at the high Ca level that was observed in the CaCO 3 and Ca-formate treatments.
Degradation of lower InsP isomers in the ileum seems to be impaired by the high Ca level when phytase was supplemented. In the CaCO 3 treatment , higher InsP 4 and InsP 3 isomer concentrations and lower myoinositol concentrations were found for the high compared to the low Ca level. This suggests a diminishing effect of a higher Ca level on phosphatases degrading InsP 4 and lower InsP isomers. Similar InsP 3 to InsP 6 concentrations were determined between the low Ca level of the CaCO 3 and the CaCO 3 +formic acid treatments, but higher myo-inositol concentrations were determined for the CaCO 3 +formic acid treatment. This may indicate a higher or faster hydrolysis of InsP 2 and InsP 1 isomers for the CaCO 3 +formic acid treatment at the low Ca level. A fast degradation of InsP 2 and InsP 1 isomers by the use of formic acid could also be observed in the higher Ca level. At high Ca levels, higher InsP 5 to InsP 3 concentrations were found for the CaCO 3 +formic acid treatment compared to the CaCO 3 treatment, but there was no difference in myo-inositol concentration. Despite a slower degradation of InsP 5 to InsP 3 isomers, this points towards a high hydrolysis of InsP 2 and InsP 1 isomers for the high Ca level for CaCO 3 +formic acid. However, it should be of greater interest to degrade InsP with higher phosphorylation rst so as to diminish potential chelating effects. Comparing the low Ca levels in the CaCO 3 and Caformate treatments, higher concentrations of InsP 6 and InsP 5 isomers and similar myo-inositol concentrations were found, indicating that the initial steps of InsP 6 degradation were impeded for Caformate, and degradation of InsP isomers lower than InsP 5 was not the limiting factor. Degradation of InsP 2 and InsP 1 isomers seems to be particularly relevant for InsP degradation at both Ca levels for CaCO 3 +formic acid when phytase was supplemented. This makes the knowledge of InsP 2 and InsP 1 isomer concentrations necessary so as to determine InsP isomers that limit InsP degradation.
Effects on calcium digestibility. Responses in pc Ca digestibility differed from other traits under investigation, probably as a result of highly regulated Ca homeostasis and regulation of Ca absorption [29]. Phytase supplementation increased the pc digested amount of Ca irrespective of acidi cation at the low Ca level, whereas such an effect was only observed for Ca-formate at the high Ca level. This may be explained by a maximum pc digested Ca amount of about 0.42 g/d. This amount was met at the high Ca level without phytase supplementation for CaCO 3 and CaCO 3 +formic acid, but for Ca-formate only when phytase was supplemented. Particle size [25,30] and Ca solubility [21] of Ca sources were shown to be a determinant of Ca utilization. Possibly, the considerably higher mean particle size of Ca-formate compared to CaCO 3 was more limiting for Ca utilization than the higher solubility of Ca-formate.

Conclusions
Replacing CaCO 3 by Ca-formate and adding formic acid to CaCO 3 -containing diets decreased digesta pH, in uenced the microbial community, and had an effect on InsP 6 degradation. Results imply that InsP degradation in the crop and until the terminal ileum may partly be explained by the microbial community because relationships between InsP degradation and relative abundance of phosphatase-producing strains L. johnsonii and L. reuteri were observed. Functional prediction supports in uences of microbiota on InsP degradation. In addition to microbiota, Ca effects on InsP degradation and pc P digestibility were shown to depend on dietary concentration, solubility of the Ca sources, as well as on the consequences of Ca supply on the pH of the digestive tract. The results of this study are relevant to the industry because choice of concentration and source of Ca in the diet is an important factor of feed formulation that affects P utilization by the animals.

Animals and management
The experiment was conducted at the Agriculture Experiment Station of the University of Hohenheim. Unsexed Ross 308 broilers were obtained from a commercial hatchery (Brüterei Süd ZN der BWE-Brüterei Weser-Ems GmbH & Co. KG, Regenstauf, Germany). The hatchlings were placed in 72 oor pens (115 × 230 cm ground area, 260 cm height) in groups of 15. The temperature was gradually reduced from 34 °C at placement to 26 °C at the end of the experiment on day 22. The light regimen was 24:0 h of light:dark in the rst three days and 18:6 h of light:dark from day 4 until the end of the experiment. Birds were kept on wood shavings for the rst 15 days. On day 16, the litter was removed from the oor and birds were kept on perforated oors thereafter. Birds were reallocated among pens on day 16 to achieve a similar group weight in all pens. The pens were randomly assigned to the treatments in a completely randomized block design. Feed and water were available for ad libitum consumption throughout the experiment. Diets A commercial starter diet was provided for the rst 15 days and contained, per kg, 215 g crude protein, 11 g Ca, 7 g P, 12.5 MJ metabolizable energy, 110 mg monensin sodium, 10 IU endo-1,4-β-xylanase, and 750 FTU of a 6-phytase (Deutsche Tiernahrung Cremer GmbH & Co. KG, Düsseldorf, Germany). From day 16, the experimental diets were provided.
Twelve experimental diets were mixed to follow a 3×2×2 factorial design of the experiment with two Ca levels and two phytase supplementation levels. The third factor was the addition of formic acid to CaCO 3containing diets or replacing CaCO 3 by Ca-formate, which is termed "acidi cation" herein (Table S14).
Except for Ca and P, the diets were calculated to meet or exceed the supply recommendations of the Gesellschaft für Ernährungsphysiologie [31]. The diets were based on corn, soybean meal, rapeseed meal, and sun ower meal and were formulated without mineral P. High levels of oilseed meals were included to achieve high InsP 6 concentrations as a substrate for the added phytase. Titanium dioxide was included as an indigestible marker at a level of 5 g/kg. Diets contained CaCO 3 , CaCO 3 +formic acid (Amasil 85, BASF SE, Germany; >85% wt/wt formic acid), or Ca-formate. CaCO 3 and Ca-formate were added in two concentrations in order to achieve dietary Ca levels of 5.6 g Ca/kg dry matter ("low") or 8.2 g Ca/kg dry matter ("high"). One half of each diet was supplemented with 1,500 FTU phytase/kg ("+"; Natuphos E 5000 G, BASF SE, Germany). The other half remained without phytase supplement ("-"). Diamol (diatomaceous earth) was used to balance mass differences between diets. Overall, analyzed nutrient concentrations con rmed calculated values (Table S2). The diets were produced by Research Diet Services (Research Diet Services BV, Hoge Maat 10, 3961NC, Wijk bij Duurstede, Netherlands) and pelleted through a 3-mm die.

Measurements and sampling procedures
Animals and feed were weighed on day 16 and at the end of the experimental period on a pen basis. At the end of the experiment, half of the pens of each treatment were processed on day 21 and day 22, respectively, for capacity reasons. The animals were deprived of feed 2 hours before slaughter, and, feeders were moved back into the pens 1 hour before slaughter in order to standardize gut ll. The animals were stunned using a gas mixture (35% CO 2 , 35% N 2 , and 30% O 2 ) and euthanized by CO 2 exposure. Crop and gizzard content was removed with a spatula without scraping the mucosa and pooled on a pen basis. A subsample of the pooled crop digesta was used for microbiota analysis and pH measurement. The section between Meckel's diverticulum and 2 cm anterior to the ileo-ceco-colonic junction, herein de ned as the ileum, was removed. Digesta from the posterior half of the ileum was sampled because P absorption in the anterior third of the ileum may not have been completed [32]. Approximately 2 cm of this ileum section was used for pH and microbiota analysis. For this purpose, the digesta was carefully stripped out. Samples for microbiota analyses were immediately stored at -20°C.
The digesta of the remainder of this ileum section was ushed out using ice-cold deionized water and pooled on a pen basis. Digesta samples were immediately frozen at -20°C until they were freeze-dried.
Microbial community analyses DNA from crop and ileum digesta samples were extracted with the commercial DNA extraction kit FastDNA TM Spin Kit for soil (MP Biomedicals LLC, Solon, OH, USA). DNA was further quanti ed with a NanoDrop 2000 spectrophotometer (Thermo Fisher Scienti c, MA, USA) and stored at -20°C. Illumina library was prepared according to Kaewtapee et al. [33]. The V1-2 region of the 16S rRNA gene was ampli ed and 1 µl of the rst polymerase-chain reaction (PCR) product was used as a template in the second PCR with multiplexing and indexing primers as described previously [34]. Samples were sent for pair-end sequencing using the 250 bp paired-end sequencing chemistry on an Illumina MiSeq platform. Raw reads were checked for quality, assembled, and aligned using the mothur pipeline tool [35]. The data included 74,662 ± 3,399 sequences per sample. The UCHIME program included in the mothur pipeline was used to identify possible chimeras [36]. Reads were clustered at 97% identity into 681 OTUs. Only OTUs with an average abundance higher than 0.0001% and a sequence longer than 250 bp were considered for further analysis. The closest representative was manually identi ed using seqmatch from the Ribosomal Database Project [37]. Sequences were submitted to European Nucleotide Archive under the accession number PRJEB38378.
Prediction of functionality was carried out with the R package Tax4Fun2 [38], which relied on the SILVA database [39] and used the KEGG hierarchy, comprising of gene catalogs from sequenced genomes [40], for the assignations. To assign functionality, the BIOM table was obtained from the mothur pipeline [41].
Genomes from 16S rRNA gene sequences identi ed in this study were downloaded from the NCBI database (https://www.ncbi.nlm.nih.gov/home/genomes/) in order to produce a database specially made for the crop and ileum of chickens.

Chemical and physical analyses
Samples of all diets were pulverized using a vibrating cup mill (Pulverisette 9, Fritsch GmbH, Idar-Oberstein, Germany) for chemical analyses of gross energy, P, Ca, Ti, InsP isomers, and myo-inositol, or ground to pass through a 0.5 mm sieve (Ultra Centrifugal Mill ZM 200, Retsch GmbH, Haan, Germany) for all other analyses. Digesta samples were pulverized using the same vibrating cup mill. Ground samples were analyzed for proximate nutrients and ber fractions according to the methods of Verband Deutscher Landwirtschaftlicher Untersuchungs-und Forschungsanstalten [42]. The concentrations of Ti, P, and Ca in pulverized feed and digesta samples were analyzed using inductively coupled plasma-optical emission spectrometry following wet digestion [43]. InsP 6 and InsP 3-5 isomers were analyzed in pulverized feed and digesta samples according to methods described by Zeller et al. [43] with modi cations noted by Sommerfeld et al. [1]. Separation of enantiomers is not possible using this methodology; therefore, we were unable to distinguish between the d-and l-forms. Some InsP 3 isomers could not be identi ed because standards were unavailable. Clear discrimination of the isomers Ins(1,2,6)P 3 , Ins(1,4,5)P 3 , and Ins(2,4,5)P 3 was not possible because they co-eluted, and therefore the term InsP 3x was used for these InsP 3 isomers of unknown proportions. Myo-inositol in feed and digesta samples was analyzed according to Sommerfeld et al. [44], using gas chromatography/mass spectrometry following derivatization. Phytase activity of the diets was analyzed according to an ISO standard method [45].
Measurements of pH in the content of the digestive tract were done using a CG 840 digital pH meter (Schott-Geräte GmbH, Mainz, Germany) equipped with a temperature probe and InLab pH combination puncture solid-state electrode (Mettler Toledo Inc., Columbus, USA).

Calculations and statistics
The ADG, ADFI, and G:F were calculated on a pen basis from day 16 to the end of the experiment and corrected for mortality. Prececal InsP 6 disappearance and pc digestibility of P and Ca (y) were calculated on a pen basis using the following equation: Statistical evaluation of all traits was performed according to the following model: y ijkl = µ + Acidi cation i + Ca level j + Phytase k + (Acidi cation i × Ca level j ) + (Acidi cation i × Phytase k ) + (Ca level j × Phytase k ) + (Acidi cation i × Ca level j × Phytase k ) + block l + e ijkl [2] where y ijkl is the mean value of each treatment; µ is the mean of all treatments; Acidi cation i is the xed effect of diets containing CaCO 3 , CaCO 3 +formic acid, or Ca-formate; Ca level j is the xed effect of the Ca level (5.6 or 8.2 g/kg dry matter); Phytase k is the xed effect of phytase supplementation (0 or 1,500 FTU phytase/kg); block l is the random block effect; and e ijkl is the residual error. The block effect included possible effects of location in the building and sampling time on day 21 or day 22 because four blocks were sampled each day. ANOVA was calculated using the MIXED procedure of the SAS software. Normal distribution and homogeneity of variance were tested prior to statistical analysis. P values described herein result from ANOVA, or t tests when two groups were compared.
In order to describe the particle size distribution of the diets, the equation described by Siegert et al. [47] was tted to the results of the sieve analysis using the NLMIXED procedure of SAS for Windows (version 9.4, SAS Institute, Cary, NC, USA): where y is the cumulative percentage of particles smaller than the sieve size x (mm), a is the slope of the regression, and b is the in ection point, which can be interpreted as the mean particle size (mm).
The sequencing dataset was statistically analyzed as described by   Figure S1. Cluster analysis similarity in crop content and ileum digesta. Figure S2. Relationship crop pH and relative abundance of OTU2 in the crop content. Figure S3. Relation between the relative abundances of OTU1 and OTU2 with other measured traits.

Ethics approval
The experiment was conducted in accordance with German Animal Welfare Legislation following approval of the Regierungspräsidium Tübingen, Germany (approval no. HOH53-18TE).

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analyzed during this study are included in this published article and its supplementary information les.

Competing interests
DF is an employee of BASF SE. The authors declare that they have no competing interests.

Funding
This study was nancially supported by BASF SE, Ludwigshafen. All authors declare that the funding source had no in uence on this study, including on its study design, analysis of results, and  Table S1.  Table S2.  Metabolism pathways. Relative proportion of genes assigned to P-related KEGG pathways in the crop content and the ileum digesta of broiler chickens fed with differently acidi ed diets with low and high Ca levels without (-) and with (+) supplementation of 1,500 FTU phytase/kg. Only signi cantly in uenced pathways are presented (P < 0.050). Columns within a statistical comparison not sharing the same letter are signi cantly different (P ≤ 0.050). Details of the statistical evaluations are shown in Table S6. Enzymes. Relative proportion of genes assigned to enzymes related to inositol phosphate and myoinositol degradation listed in the KEGG database in the crop content and the ileum digesta of broiler chickens fed with differently acidi ed diets with low and high Ca levels without (-) and with (+) supplementation of 1,500 FTU phytase/kg. Only signi cantly in uenced enzymes are presented (P < 0.050). Columns within a statistical comparison not sharing the same letter are signi cantly different (P ≤ 0.050). Details of the statistical evaluations are shown in Table S7. P and InsP. InsP6 disappearance in the crop, prececal (pc) InsP6 disappearance, pc P digestibility, and amount of pc digested P of broiler chickens fed with differently acidi ed diets with low and high Ca levels without (-) and with (+) supplementation of 1,500 FTU phytase/kg. Only signi cant (P < 0.050) effects are shown. Columns within a statistical comparison not sharing the same letter are signi cantly different (P ≤ 0.050). Details of the statistical evaluations are shown in Table S8.  Table S8.