Molecular Scale Studies of Phosphorus Speciation and Transformation in Manure Amended and Microdose Fertilized Indigenous Vegetable Production Systems of Nigeria and Republic of Benin

This study investigated the speciation, transformation, and availability of P during indigenous vegetable production by employing a combination of chemical and spectroscopic techniques. The study focused on sites in two ecozones of SSA, the dry savanna (lna, Republic of Benin) and rainforest (Ilesha, Nigeria). Both sites were cultivated with two indigenous vegetable species: local amaranth (Amaranthus cruentus (AC)) and African eggplant (Solanum macrocarpon (SM)). The soils were treated with 5 t/ha poultry manure and urea fertilizer at the rates of 0, 20, 40, 60, and 80 kg N/ha. Soil samples were collected before planting and after harvest. Phosphorus Kedge X-ray absorption near-edge structure (XANES) spectroscopy was used to determine P speciation in these soils. Quantitative analysis showed that adsorbed and organic P were the two dominant P species in the manure amended dry savanna (DS) soils before planting and after harvest in soils cultivated with both AC and SM, with the addition of urea (40 kg N/ha) causing an increase in the organic P form in dry savanna soils cultivated with AC. Soils of the rainforest (RF) cultivated with AC initially had large amounts of apatite P in the manure amended soils prior to planting, which was transformed to adsorbed and organic P after harvest. Urea addition to the rainforest soils shifted the dominant P species from organic P to adsorbed and apatite P, which was likely to limit P availability. Soils cultivated with SM had similar proportions of both organic and adsorbed P forms, with 40 kg N/ha addition slightly increasing the proportion of adsorbed P.


Introduction
Soils of Sub-Saharan Africa (SSA) are generally highly weathered, sandy textured, and characterized by poor fertility [1] and low phosphorus (P) and nitrogen (N) levels [2]. These soil characteristics pose limits on agricultural productivity, which in turn leads to food, household, and economic insecurity for West Africa [3][4][5]. With a rapid population growth rate of 3 to 5% [3], there is a need to intensify agricultural production sustainably to meet the food needs of this region. Accordingly, Integrated Soil Fertility Management (ISFM) is a package of agronomic recommendations designed to improve both soil fertility and agricultural yields. The ISFM incorporates the use of inorganic fertilizers and other organic amendments with improved crop varieties and crop rotations [6,7]. One of the ISFM fertilization techniques is fertilizer microdosing; the precise application of relatively small quantities of fertilizer to plants near their planting holes, which reduces the volume of fertilizers farmers need to apply due to improved nutrient uptake efficiency. This technique has proven to be economical and efficient in a wide variety of agricultural systems [2,8]. Indigenous leaf vegetables such as Amaranthus cruentus and Solanum macrocarpon are an integral part of agriculture in SSA; they contribute to human nutrition and traditional medicine and play a key role in the rural economy and family subsistence [9][10][11][12]. This MicroVeg project is incorporating the fertilizer microdosing technique to increase indigenous vegetable production in Nigeria and the Republic of Benin. Fertilizer microdosing has proven to increase yields in cereals across SSA, but has been much less frequently utilized for vegetable crops.
The practice of microdosing is not without negatives. Since farmers are applying reduced rates of inorganic fertilizer and experiencing large yield increases, there is a concern that repeated use of fertilizer microdosing may mine nutrients from the soil, reduce soil pH, and accelerate land degradation [13]. Specifically, in systems where microdosing uses urea instead of NPK fertilizer, nutrient mining of P may limit yields for microdosing. Phosphorus is essential for plant growth and development [14]. However, P is often strongly retained in the soil by chemical mechanisms such as adsorption to mineral surfaces, especially Al and Fe oxides [15,16] and precipitation as phosphate minerals with Al and Fe in acidic soils or with Ca and Mg in circumneutral soils [17]. As a result of these fixation reactions, plant available P in the inorganic orthophosphate form (HPO4 2− or H2PO4 − ) is sparingly found in soil solution, and instead, dissolved P is largely in the organic form [18], which can limit availability.
The application of manure to agricultural soils is beneficial in many ways: it provides essential macronutrients such as N and P [19][20][21][22], improves soil organic matter content [23], and helps to maintain soil quality [24]. However, manures can also alter the physical and chemical properties of the soil [25], with P minerals in the soil transforming into new phases even after short term manure application [26], and pose an environmental threat to surface waters if not properly managed. The bioavailability of P in manure and manure amended soils varies with the type of manure and soil, the solubility of the P compounds in them, and the reaction products formed upon their application [20,27,28]. Therefore, understanding how the applied manure and nitrogen fertilizer influences the forms and plant availability of P is important in minimizing its negative effects on the environment [29] in any agricultural system. X-ray absorption near-edge structure (XANES) spectroscopy has been used by many researchers in studying the speciation of P in soils [26,[30][31][32][33][34][35][36][37]. This direct speciation technique is preferred because it is element specific and non-destructive [38] and it avoids bias in distinguishing different P species that can arise from chemical fractionation methods [39,40]. Typically, P species in soil samples can be identified and quantified using linear combination fitting (LCF), which requires a robust spectral library of P reference compounds [41,42]. The spectral features of different phosphate standard materials have been reported by several researchers [30,31,34,41,[43][44][45] and can be used to determine overall P speciation in soils amended with organic or inorganic P fertilizer.
The objective of this study is to identify P species present in the soil before and after the cultivation of two indigenous vegetables (Amaranthus cruentus and Solanum macrocarpon) with the MicroVeg agronomic practices in two ecozones (dry savanna and rainforest) of West Africa. This study utilizes a combination of both the direct solid state speciation via synchrotron X-ray absorption spectroscopy (XAS) and indirect speciation via chemical analysis to determine the linkages between availability and mobility of P and its speciation in soils.

Materials and Methods
The study sites were located in (a) Ina, Republic of Benin (9° 95 N, 2°72 E), which has an elevation of 380 m, an annual precipitation of 1073 mm, and a mean temperature of 26.5 °C, and (b) Ilesha, Nigeria (07° 29′ 21.9"N 004° 34′ 17.3"E), which has an altitude of 372 m, an annual precipitation of 1386 mm, and a mean temperature of 25.5 °C. Ina is in the dry savanna ecoregion, whereas Ilesha falls within the rainforest ecoregion. Plots from both sites were cultivated with two indigenous vegetable species shown in Figure 1: local amaranth (Amaranthus cruentus) and African eggplant (Solanum macrocarpon). The experimental plots consisted of 1 control and 4 treatments: manure only (T0), as the control, manure + 20 kg/ha urea (T20), manure + 40 kg/ha urea (T40), manure + 60 kg/ha urea (T60), and no manure + 80 kg/ha urea (T80) using a randomized complete block design and replicated four times. Manure was applied at 5 t/ha, and it contained 20% C, 2.5% N, and 1.5% P by weight. Soil samples were randomly collected from 9 to 10 points on each plot before planting (BP) and after final harvest (AH, i.e., after 3 cycles of harvesting the vegetable leaves) at a 0-20 cm depth using a soil auger. A sub-sample of a mixed composite sample was air dried, ground to pass through a 2 mm sieve, packaged in airtight and sealed bags, and shipped to the University of Saskatchewan in April 2017. The soils were stored in airtight vials at room temperature and analyzed for pH, total P, available P, estimated cation exchange capacity (eCEC), and P XANES analysis. Due to the limited volume of soil samples shipped, bulk density could not be measured. Soil pH was measured in triplicate in a 2:1 water:soil suspension (Carter and Gregorich, 2008), and particle size analysis was done using the hydrometer method following dispersion with sodium hexametaphosphate [46]. Organic C was determined by combustion using a Leco C/N analyzer (Leco Corp., St. Joseph, MI, USA), and total P was measured colorimetrically with an auto-analyzer (Folio AA3) in triplicate following digestion using the H2SO4-H2O2 acid block digestion method of Thomas et al. (1967). Available P and eCEC were determined using the Mehlich-III extraction procedure, as described by Carter and Gregorich (2008). This method can extract multiple elements of interest to this study and is known to be appropriate for tropical soils [46]. Available P was determined from the Mehlich-III extracted soil solution on an auto-analyzer, and CEC was calculated by measuring the concentrations of exchangeable Ca 2+ , Mg 2+ , Na + , and K + cations in vacuum filtered Mehlich-III solutions on a Microwave Plasma Atomic Absorption Spectrometer (MP-AES 4100, Agilent Technologies). Estimated CEC was then calculated as the sum of exchangeable base cations. All chemical analyses were carried out in triplicate.

Phosphorus K-Edge XAS
Prior to XANES analysis, all triplicate soil samples of each treatment were mixed into one composite sample, finely ground, homogenized, sub-sampled, and spread as a thin film (area of 2 cm 2 ± 0.5 cm) on graphite tape and then carefully mounted on a sample holder. Phosphorus K-edge XANES data were collected at the Soft X-ray Micro-characterization Beamline (SXRMB) of the Canadian Light Source (CLS), Saskatoon, Canada. Measurements were performed with an InSb (III) double crystal monochromator with an energy resolution of 0.5 eV in partial fluorescence mode (PFY) using a 4 element solid-state detector (Bruker Optics) optimized for P emission. Two to three scans were taken for each sample and merged into one spectrum. The beamline energy was calibrated after every beam fill using a ZnPO4 reference compound. The acquisition time for the spectrum varied between 2 s and 5 s depending on the total P concentration in the sample. All spectra were baseline corrected with a linear pre-edge (energy region: 2115.5-2140.5 eV) and edge normalized (energy The LCF approach assumes that the spectrum from a heterogeneous soil sample is an average of all the P species in the soil. Our previous studies suggested that a viable P speciation strategy includes one organic P phase (phytic acid was used as a representative compound), one adsorbed phosphate reference, a poorly crystalline calcium phosphate (brushite), a highly crystalline calcium phosphate (apatite), and an Al and Fe phosphate mineral. These phases were all included as potential components in the LCF of each sample, but a maximum of four standards per sample was allowed in the fits. Fits were accepted when the sum of the percentages of included standards were in the range of 90% to 110%, but then, the relative contribution was normalized to 100% for comparison. Goodness of fit values (as reduced chi 2 ) in Athena were typically 0.002-0.005, suggesting a statistically significant fit. Two samples that had very different bulk XANES speciation were also imaged using the tender X-ray microprobe of the SXRMB beamline using a 4 element Vortex detector, a 10 × 10 µ m incident beam spot size with an overall 0.5 × 0.5 mm dimension, and a constant incident energy of 7200 eV to capture the regions of interest between Si and Fe. Soil samples were loaded as a dried, ground, thin layer on carbon tape. X-ray fluorescence map data analysis was processed using the SMAK software package (Sam Webb, Stanford Synchrotron Radiation Lightsource) to subtract the inline beamline I0 and generate tricolour images to visualize co-locations of elements of interest in the samples.

Statistical Analysis
The physico-chemical analysis data were analyzed using the Statistical Package of Social Sciences (SPSS Inc.) at p ≤ 0.05 level. The significance of treatments and time effects, as well as the interaction of treatments and time effects was analyzed using repeated measure analysis of variance and determined significant at p ≤ 0.05 probability levels.

Yields
The RF had higher vegetable yields than the DS for both vegetable species (Figure 2a,b). For the AC, the T40 had the highest yield in the RF, while the T60 gave the highest yield in the DS ( Figure  2a). For the SM, the T60 gave the highest yield in both ecoregions ( Figure 2b). However, for economic profitability, the T40 was considered the recommended rate.

Soil Physico-Chemical Analysis
Particle size analysis results showed that soils in both ecoregions were coarse textured, having sand content as high as 81% in Ina and 85% in Ilesha. However, the rainforest (RF) soils had more fine texture particles than those in dry savanna (DS), with clay contents as high as 33% in some samples (Tables 1 and 2). The soils in the DS were slightly basic to neutral in pH (6.1 to 7.0) ( Table 1), while the RF soils ranged from acidic, 3.5, to slightly basic, 6.7 (Table 2). Generally, for soils cultivated with Amaranthus cruentus (AC), the pH reduced over time in both the DS and RF ecoregions (Tables  1 and 2). For the soils cultivated with Solanum macrocarpon (SM), the pH significantly increased after harvest in all treatments at the DS site (Tables 1, Figure 3a), while in the RF, the pH significantly decreased over time except for T80, i.e., the samples with heavy urea addition (Table 2, Figure 4b).
The DS soils had a significantly lower total P than the RF soils. The maximum total P measured in the DS was 347.25 mg kg −1 (Table 1), versus 1775.50 mg kg −1 in the RF (Table 2). At both sites, the available P was a small fraction of the total P, with only 3.3% and 11% of the total P being available in the DS and RF sites, respectively (Tables 1 and 2). In the soils cultivated with AC, total and available P reduced after harvest compared to before planting at both sites (Figures 5b and 4a), while in the soils cultivated with SM, total P reduced after harvest in all treatments as available P increased at the DS site (Figure 3b, Table 1). However, in the RF, for soils cultivated with SM, total P increased in all treatments except T40 (Table 2, Figure 4a), albeit within the margin of error, while available P had a different trend, increasing in all treatments except T20 at the RF site for soils cultivated with SM (Table 2, Figure 4a).  In the DS, eCEC reduced with time in all treatments except for T80 in the soils cultivated with AC, while the eCEC increased with time for all treatments in the soils cultivated with SM (Table 1). In the RF, the eCEC reduced with time in all treatments in soils cultivated with AC (Table 2), and in the soils cultivated with SM, eCEC increased in T0 and T40 and reduced in T20 and T80 ( Table 2). The trend observed in the eCEC of soils of both ecoregions cultivated with AC was almost similar to the trend observed in the pH, where both pH and eCEC reduced over time in treatments of both the DS (except for T80) and RF (except for T0) ecoregions for soils cultivated with AC (Tables 1 and 2).

P XANES Analysis
For this study, we chose to compare the control soils, T0, with the T40 treatments as this represented the MicroVeg fertilizer recommendation in both ecoregions. The XANES spectra for all the DS soils were almost identical in shape and energy position. Visual inspection of the sample spectra presented in Figure 6a shows features that are characteristic of organic and adsorbed phosphates such as a post-white line peak at 2171.3 eV (Zuo et al., 2015). The LCF results of the DS in the AC control, AT0, showed that adsorbed P (64.9%) and organic P (35.1%) were the dominant P species (Table 3). Upon harvest however, both P forms slightly reduced by 1% and 2%, respectively, apatite accounting for the remaining 2.9% of P present in the soil. In the soils amended with 40 kg N/ha of urea (AT40), the P forms present prior to planting were adsorbed P (67.9%) and organic P (32.1%); after harvest, however, the adsorbed P was reduced by 27.2%, while the organic P increased by 26.1% (Table 3, Figure 6b). Table 3. Proportion of P species in the amended soils from linear combination fitting (standard error of the fit in parentheses). In the DS soils cultivated with SM, the LCF showed an almost equal distribution of organic P (51.6%) and adsorbed P (48.4%) (Figure 6a,b) as the P forms present prior to planting in ST0. After harvest, the adsorbed P and organic P reduced slightly to 47.8% and 52.7%, respectively, and there was the occurrence of some apatite P (3.1%) (Figure 6a,b). In the ST40 treatment, adsorbed and organic P had an even distribution as the dominant P form, accounting for 49% and 51%, respectively, of P forms in the soil before planting and having no significant change after harvest.

Ina (Dry Savanna)
Organic P Adsorbed P Apatite P In the RF, visual inspection of the sample spectra presented in Figure 7a,b showed features that were characteristic of Ca phosphates such as apatite, a post-edge shoulder near 2157.5 eV, and a post-white line peak at 2164.7 [47], as prominent in the soils cultivated with AC. LCF results showed apatite and adsorbed P to be the P forms present in AT0 before planting, with both P forms accounting for 81% and 18.9% of P forms, respectively (Table 3, Figure 6a,b). Upon harvest, there was a major change in the P forms found in the soil: organic and adsorbed P were the dominant P forms found after harvest, accounting for 62.9% and 21.5% of P forms, respectively, while apatite accounted for 14.7% of the P forms present (Table 3, Figure 7a,b). In AT40, prior to planting, organic P accounted for 47.7% of the P present, while adsorbed and apatite P accounted for 42.1% and 8.1% of the P forms present, respectively (Table 3, Figure 7a,b). After harvest, there were notable changes in the P forms present, as apatite P more than tripled and accounted for 32.9% of P forms, adsorbed P decreased to 27.5%, while organic P saw a reduction to 37.5% of the P forms present in the soil (Figure 7a,b).
Furthermore, in the rainforest soils cultivated with SM, the LCF showed adsorbed P and organic P as the dominant P forms in ST0 before planting, at 51.4% and 47.6%, respectively (Table 3, Figure  7a,b). After harvest (ST0AH), there were changes in the P speciation as organic P increased to 60.1%, while adsorbed P decreased to 38.5% (Table 3, Figure 7a,b). In ST40 treated soils, organic P accounted for 57.6% of the P present in the soil prior to planting while adsorbed P accounted for 40.5%. After harvest, organic P reduced slightly to 50.9% while adsorbed P increased, accounting for 49.1% of the P species present (Figure 7a

Ilesha (Rainforest)
Organic P Adsorbed P Apatite P

X-ray Fluorescence Imaging
Further investigation into the RF samples AT0Bp and AT40Ah on a microprobe showed that the AT0Bp samples had concentrated discrete particles that contained high levels of both Ca and P (turquoise "hotspots" in Figure 8a), which was consistent with the high apatite content observed by LCF analysis of the bulk XANES for this sample. In contrast, the AT40Ah image showed P that was both correlated with Fe (yellow regions in Figure 8b) and spread diffusely throughout the entire sample. This was what would be expected based on the LCF model that described the sample with P adsorbed on iron oxides and P of organic origin.

Discussion
The RF soils were generally more acidic and had higher P content than the DS soils (Tables 1 and  2). It is possible that the presence of Fe (as evident from the soil acidity) or higher clay content in the RF soils allowed the binding and retention of more P. The concentrations of extracted soil P forms were affected by the interaction of treatment and time, but the interaction between these two factors showed inconsistent effects for all P forms (Tables 1 and 2). In the DS, total P significantly reduced with increasing rates of urea addition (Figures 5b and 3b); T0 had significantly higher total and available P than T40 in soils cultivated with both AC and SM (Table 1), but the decline in total and available P was more in T0 than in T40 ( Table 1), suggesting that urea addition slowed down P mining in these soils. In the RF, soil pH declined with increasing rates of urea addition (Figure 4b); the significant reduction in the soil pH of the T0 and T40 treatments in soils cultivated with AC correlated with the increment in total P ( Table 2). This suggested that urea addition was contributing to soil acidification in these treatments and simultaneously increasing the soils' total P content. There was also a significant decline in available P in both samples at harvest (Table 2). For soils cultivated with SM, the significant increment in the total and available P content of the soils over time in the T0 treatment suggested that the manure added to the soil raised the soils P content over time, whereas the decline in the total P in the urea addition (i.e., T40) and the significant increment in available P in that sample suggested that urea addition caused P mining in this soil.
The LCF results suggested the adsorption of P to mineral phases of Fe, organic P, and Ca phases (in some of the RF samples). These results were in line with other studies [26,31,33] that observed a dominance of adsorbed P in manure amended soils and highlighted the role of Fe in the chemistry of P in manure amended soils. In the DS, the marginal shift in the dominant P species in AT0, from 64.9% adsorbed P and 35.1% organic P (Table 3, Figure 6b) before planting to 63.9% adsorbed P and 33.1% organic P, and the occurrence of apatite P (2.9%) after harvest (Table 3, Figure 6a) were also accompanied by a reduction in the pH and total and available P after harvest (Table 3, Figure 5a,b). This reduction in both the adsorbed and organic P species and the formation of apatite suggested a (a) (b) slow but gradual conversion of P into mineral (apatite) P, which may become soluble as pH decreases by rhizosphere acidification [48]. Apatite can dissolve under acidic conditions that are achievable by rhizosphere acidification and be broken down to adsorbed P [49], which is plant accessible in alkaline and neutral soils because acidification increases the solubility of Ca-P minerals, making Ca-P and sorbed P phases preferentially extracted [49].
In the DS soils amended with manure and 40 kg N/ha, the dominant P form shifted from 67.9% adsorbed P and 32.1% organic P to 40.7% adsorbed P and 58.3% organic P, indicating a significant transformation of P over time from the adsorbed form to the organic form. This was also accompanied by a reduction in pH and total and available P (Table 1, Figure 5a,b). The shift in the P phase in AT40 from adsorbed P to organic P may be due to the microbial immobilization of the available P in the soil into the organic P phase (i.e., from adsorbed P-solution P-organic P). The accompanying reduction in the soil pH may be caused by the production of organic acids by the plant roots and/or soil microorganisms or a product of acidification from the process of nitrification of ammonium [34] in soils treated with urea.
The same speciation of adsorbed and organic P in the AC cultivated soils was also observed in the P K-edge XANES results of the DS soils cultivated with SM. In ST0, both organic and adsorbed P were the dominant P species, accounting for 51.6% and 48.4% of P species, respectively (Table 3, Figure 6b) prior to planting; however, upon harvest, there was a slight reduction in both P species, possibly because of its conversion into apatite P, hence its occurrence with an abundance of 3.1% after harvest (Figure 6a,b). The gradual shift in the P pool from adsorbed and organic P towards apatite P (Figure 7a,b) was consistent with the findings of Kar et al. (2017), who reported the gradual conversion of adsorbed P into apatite P, and Dendougui and Schwedt (2004), who noted the strong affinity of organic P to Ca minerals. This observed shift in P speciation may be due to precipitation of P with Ca 2+ in the high CEC ST0Ah soil of the DS (Table 1), leading to a slow transformation of P into the mineral phase in this soil as the soil pH increased (Table 1, Figure 3a), increasing available P while reducing total P (Figure 3b). The ST0Ah LCF spectra however did not have a well pronounced shoulder feature around 2154 to 2158 eV typical of Ca phosphates [44], possibly due to the low P concentration in the soil.
In the ST40 treatment, there was only a marginal change in the P speciation prior to planting and at harvest, with adsorbed and organic P accounting for 40.5% and 57.6% of total P respectively before planting and 49.1% and 50.9% after harvest (Figure 6a,b), although the pH and available P increased as the total P reduced (Table 1, Figure 3a,b). This decrease in total P as the soil pH and P availability increased suggested P mining from the soil by the vegetables, hence explaining the optimum yield from the ST40 treatments (Figures 2). It is possible that as the soil microbes gain access to the N from the urea, they mineralize the organic P in the soil, making them more plant available, hence the reduction in the proportion of the organic P and total P while available P increased (Table 1). This could cause a potential increase in P susceptibility to loss by leaching or runoff in this soil.
In the RF, the major shift in the dominant P species in the AT0 treatment from 81% apatite P (possibly present due to poultry manure amendment, which is known to be high in Ca-P phases [43,50,51]) and 18.9% adsorbed P before planting to organic P (62.9%), adsorbed P (21.5%), and 14.7% apatite P (Table 3, Figure 7b) after harvest strongly suggested that the dominant P cycling process here was microbial mediated immobilization. The apatite P could be expected to weather slowly due to organic acid exudates from soil microbes and/or plant roots, and then, P becomes more plant available and thus taken up out of the soil. It is also possible that some of the P is lost via leaching/runoff as the P is transformed from the mineral phase into the adsorbed/organic phase, hence accounting for the reduced total and available P over time (Table 3).
In the RF AT40 treatment, prior to planting, organic P (47.7%), adsorbed P (42.1%), and apatite P (8.1%) were present. Upon harvest, the dominant P forms shifted to organic P (37.5%) and apatite P (32.9%), while adsorbed P decreased to 27.5% (Table 3, Figure 7b). This suggested a conversion of organic P to the mineral P phase; this is possibly a microbial mediated process of mineralization that makes P become more plant available. However, the XANES LCF showed that mineralized P rapidly formed precipitates with Ca, thereby reducing the availability of P in the soil (Table 3). Furthermore, the huge reduction in the total P after harvest (Table 2) suggested that some of the P may be lost to the environment through erosion and runoff. Furthermore, in the RF, the P K-edge XANES for ST0 soils showed adsorbed and organic P as the only two P species present in the soils, both accounting for 47.6% and 51.4% of the P species respectively before planting (Table 3, Figure 6b), with a small change in the P phase occurring after harvest as organic P increased by 8.7% while adsorbed P reduced by 9.1%. Explanations for this include (a) either some of the adsorbed P is desorbed and then becomes immobilized by the soil microbes or (b) as the manure remains in contact with the soil, it begins to get slowly immobilized by the soil microbes. The latter sounds more feasible, given that the soil pH slightly decreased between planting and harvest, which would favor P immobilization over P desorption. However, in the ST40 treatment, organic P (57.6%) and adsorbed P (40.5%) were the initial P forms (Table 3, Figure 7b), whereas organic P reduced to 50.9% after harvest while the adsorbed P increased to 49.1%. This shift in P phase suggested that as N becomes more available to plants and microbes in the soil, some of the organic P in the soil becomes mineralized and increases in availability ( Table 2). As availability increases, some of the P may adsorb to Fe minerals given the low soil pH ( Table 2) [52].

Conclusions
This study investigated the solid phase speciation of P in manure amended and urea fertilized soil of the rainforest ecozone of Nigeria and dry savanna ecozone of the Republic of Benin under cultivation with two indigenous vegetable species: local amaranth (Amaranthus cruentus) and African eggplant (Solanum macrocarpon), using P K-edge XANES spectroscopy. The quantitative analysis of the XANES spectra of the dry savanna soils of Ina, Republic of Benin, under cultivation by the local amaranth (AC) and the African eggplant (SM) showed that adsorbed P and organic P were the two dominant P species in the manure amended soils before and after planting. In the manure amended soils of the dry savanna, the formation of apatite after harvest as pH and total P reduced showed the gradual conversion of the organic and adsorbed P into mineral P (Ca-P), which may become soluble and dissolve under acidic conditions. It is however difficult to conclude on how N addition affected P speciation and transformation, as the addition of 40 kg N/ha caused an increase in the organic P form in soils cultivated with AC and an increase in adsorbed P in soils cultivated with SM.
In the rainforest soils of Ilesha, Nigeria, XANES analysis showed the presence of apatite P in the manure amended soils cultivated with AC, which was then transformed via microbial immobilization into organic P. When N was added to this soil (i.e., AT40), XANES analysis showed a shift in the dominant P species from organic P before planting to adsorbed P and apatite P, most likely through a combination of mineralization and precipitation processes. In the soils cultivated with SM, organic and adsorbed P were the two dominant P species in both the manure amended (ST0), and the manure amended + urea fertilized soils (ST40). However, at harvest, there was a slow immobilization of P from the adsorbed phase into the organic phase in the manure amended soils (ST0), with N addition reversing this effect and mineralizing the P pool from the organic P to adsorbed P. We concluded that the N addition to the soils of both ecoregions caused P to cycle differently, thereby suggesting that we may be mining P from these soils. analyses. Special thanks also go to members of the Environmental Soil Chemistry research group at the University of Saskatchewan for their support.