A rapid ammonium fluoride method to determine the oxygen isotope ratio of available phosphorus in tropical soils

Rationale The isotopic composition of oxygen bound to phosphorus (δ18OP value) offers an opportunity to gain insight into P cycling mechanisms. However, there is little information for tropical forest soils, which presents a challenge for δ18OP measurements due to low available P concentrations. Here we report the use of a rapid ammonium fluoride extraction method (Bray‐1) as an alternative to the widely used anion‐exchange membrane (AEM) method for quantification of δ18OP values of available P in tropical forest soils. Methods We compared P concentrations and δ18OP values of available and microbial P determined by AEM and Bray‐1 extraction for a series of tropical forest soils from Panama spanning a steep P gradient. This involved an assessment of the influence of extraction conditions, including temperature, extraction time, fumigation time and solution‐to‐soil ratio, on P concentrations and isotope ratios. Results Depending on the extraction conditions, Bray‐1 P concentrations ranged from 0.2 to 66.3 mg P kg−1 across the soils. Extraction time and temperature had only minor effects on Bray‐1 P, but concentrations increased markedly as the solution‐to‐soil ratio increased. In contrast, extraction conditions did not affect Bray‐1 δ18OP values, indicating that Bray‐1 provides a robust measure of the isotopic composition of available soil P. For a relatively high P soil, available and fumigation‐released (microbial) δ18OP values determined by Bray‐1 extraction (20‰ and 16‰, respectively) were higher than those determined by the AEM method (18‰ and 12‰, respectively), which we attribute to slightly different P pools extracted by the two methods and/or differences resulting from the longer extraction time needed for the AEM method. Conclusions The short extraction time, insensitivity to extraction conditions and smaller mass of soil required to extract sufficient P for isotopic analysis make Bray‐1extraction a suitable alternative to the AEM method for the determination of δ18OP values of available P in tropical soils.


| INTRODUCTION
Tropical forest soils sustain a large net primary production despite low phosphorus (P) availability. 1 Given the importance of understanding how tropical forests will react to future environmental change, and the role of soil P in regulating these responses, there is an urgent need to better understand P cycling in tropical forest soils. 2 This requires the development of novel procedures that can provide information on the dynamics of P in the soil-plant-microbe continuum.
A promising technique for the investigation of soil P cycling involves the determination of the 18 O: 16 O ratio of oxygen (O) bound to P (δ 18 O P value). [3][4][5][6] The δ 18 O P technique has been used to investigate the importance of microorganisms for P cycling 7 and can provide information about hydrolysis by phosphatase enzymes 8,9 and the origin of P inputs into aquatic systems. 10,11 However, information on δ 18 O P values in tropical forest soils remains scarce, despite the importance of P in the ecology of this hyper diverse biome. 1 Indeed, the only study so far involved the quantification of δ 18 O P values in soils from litter and fertilization experiments in Panama, which suggested the importance of microorganisms for P cycling in lowland tropical soils. 12 The main method for quantifying the δ 18 O P values of available P is extraction via an anion-exchange membrane (AEM). 13 However, a number of potential issues limit the use of the AEM method for tropical soils, including the low available P concentrations (often <1 mg P kg −1 ) 14  and 50 L of water would be necessary for tropical soils. 12 In addition, the relatively long extraction time for the AEM method might influence results for δ 18 O P values, particularly for the determination of δ 18 O P values in microbial biomass, if enzymatic activity leads to hydrolysis of organic P during the extraction. It is therefore recommended that AEM extractions for δ 18 O P measurement be performed at 4 C, 15 which presents an additional limitation on the procedure.
Several alternative extraction procedures exist for soil available P that might be suitable for the determination of δ 18 O P values, including extraction in water and sodium bicarbonate. 16,17 Water extracts, however, can contain considerable concentrations of fine clays, which are difficult to remove by filtration and interfere with analysis, and water-extractable P concentrations in tropical soils are usually even lower than in AEM extracts. 18 In contrast, P concentrations in sodium bicarbonate extracts are usually greater than in water extracts, but the high solution pH, carbonate and salt concentration could lead to problems during the purification of P for δ 18 O P determination.
Degassing prior to the purification and precipitation of brucite is recommended to further clean the extracts. 19,20 In addition, sodium bicarbonate extracts are slightly alkaline, and can therefore extract a considerable amount of organic P. The purification protocol for the δ 18 O P determination only targets inorganic P, but extracted organic P could be hydrolysed under the acidic conditions of the colorimetric assay of orthophosphate. 21 As the orthophosphate concentrations are used to calculate the δ 18 O P values of microbial P, hydrolysis of organic P might lead to erroneous results.
An alternative procedure involves the extraction of available P in acidic ammonium fluoride (Bray-1 extraction; 30 mM NH 4 F + 25 mM HCl). 22 The method is appropriate for tropical soils because it is designed to extract P from acidic soils and extracts little organic P (the extraction is conducted at pH 2.5). 23 The NH 4 F prevents readsorption of P onto metal oxides, which are abundant in strongly weathered tropical soils. Importantly, the extraction time for the Bray-1 method is considerably shorter than for the AEM method (minutes compared with hours), which favours the accurate determination of   (Table 1).
Soil samples were taken from the upper 10 cm of the soil, sieved (<2 mm) fresh, stored at 4 C and extracted within 2 weeks of sampling.

| Extractions
All extractions involved fresh soils, and solution-to-soil ratios were based on fresh weights and not dry weights. However, data is reported on the basis of oven-dry soil. Phosphorus concentrations in the extracts are referred to as P unf (P in unfumigated extracts) and P fum (P in liquid (hexanol) or gaseous (chloroform) fumigated extracts).
Based on pre-tests, we decided not to replicate the extractions for the determination of P concentrations, because the error associated with replicate extractions was <5%.
For AEM extractions we followed the protocol of Turner and Romero. 28 In brief, 10 g fresh soil, 80 mL ultrapure (18.2 MΩ) water and five resin strips (1.5 × 4 cm) were used (unfumigated extracts).
Fumigated extracts received an additional 1 mL hexanol. To test for a temperature effect on P concentrations, the samples were shaken overnight at 22 C or 4 C. On the following day, the resin strips were removed, cleaned with ultrapure water and eluted for 1 h in 50 mL 0.25 M sulfuric acid (H 2 SO 4 ). Table 2 summarizes the different extraction characteristics tested for the Bray-1method (fumigation with CHCl 3 vapor). 29 We tested the effect of fumigation time by using three different times. Two After extraction, samples were centrifuged (3000 g, 15 min) and filtered through Whatman 42 filter papers. The P concentrations in all extracts were determined by molybdate colorimetry. 30 Phosphorus released by fumigation (fumigationreleased P) was calculated as the difference between the concentrations of the fumigated and unfumigated extracts. We did not determine P recovery to correct for P adsorption during the extractions, as the recovery of P spikes is not comparable with the recovery of microbial P released during chloroform fumigation in acidic soils. 31 For the δ 18 O P values of AEM P unf and P fum , we used the same solution-to-soil ratio as for the determination of the P concentrations but, depending on the P concentrations, we used 200-600g fresh soil for AEM P unf and 100-200 g fresh soil for AEM P fum (instead of the normal 10 g) to obtain sufficient P for analysis.  (Table 1). In addition, the soil from Madden Dam was used to investigate the effect of the solution-to-soil ratio and extraction temperature on the δ 18 O P value of Bray-1 P unf . The solution-to-soil ratios were: 5, 10 and 50. A ratio of 10 is the standard solution-to-soil ratio used for Bray-1 extractions. 32 The other two ratios were a compromise between amount of P extracted and volume of Bray-1 solution needed.

Soils from Plantation Road and Madden
Extractions of soil from Madden Dam were carried out with 18  Dam, we assume that this would also be the case for soils with lower organic/condensed P concentrations.

| Measurement of oxygen isotope ratio
The AEM and Bray-1 extracts were purified following Tamburini et al, 34 but with the addition of 1 mL concentrated H 2 SO 4 during the ammonium phosphomolybdate (APM) step to facilitate the precipitation of the crystals. 35 Measurement of the δ 18 O P values was undertaken by weighing approx. 300 μg of Ag 3 PO 4 into a silver capsule to which a small amount of fine glassy carbon powder was added to aid combustion. 34 The sample was converted into carbon The oxygen isotope ratios are reported in the conventional delta notation: where R = 18 O/ 16 O and R standard is the VSMOW.

| CALCULATIONS
The effect of the solution-to-soil ratio and extraction temperature on and not as a separate set of samples. One-way ANOVAs followed by

| Phosphorus concentrations
With increasing solution-to-soil ratio the P unf concentration in the Bray-1 extracts increased between 6-and 112-fold, depending on the soil (Figure 1). The largest proportional increase in P unf concentrations was for Plantation Road, which has high total P concentrations, and the lowest for Plot 5, which contained a relatively low total P concentration ( Table 1). The largest absolute increase in P unf concentration was observed for Madden Dam  Figures 2B and 2C).
Extraction at 4 C compared with 22 C increased the Bray-1 P unf concentrations slightly, but significantly (p < 0.05), for Madden Dam, but not for Plantation Road (p > 0.1) ( Table 3).
The extraction temperature did not affect the AEM P unf concentrations for Plantation Road, but the AEM P unf and P fum concentrations increased by a factor of 1.6 for Madden Dam when extracted at 22 C compared with 4 C (Table 3).

| The δ 18 O P values of Bray-1 extracts and the influence of extraction conditions
For Madden Dam, increasing the fumigation time reduced the δ 18 O P values of Bray-1 P fum slightly, but not significantly. The differences between the δ 18 O P values of P unf and P fum by Bray-1 were small, which makes it difficult to accurately calculate the δ 18 O P values of microbial P. It is most likely that at this site the δ 18 O P values of microbial P and P unf are similar. For Plantation Road, the concentrations of Bray-1 P unf were around the lower limit of the purification method and we could not obtain a sufficient amount of silver phosphate for the δ 18 O P determination. We would have needed at least 100 g of fresh soil to yield a sufficient amount of P. This is still an order of magnitude less than the 1 kg of fresh soil needed in the case of the AEM method, but would require the volume of the Bray-1 extract to be reduced, for example by using the MAGIC method. 37 For Plantation Road, the δ 18 O P values of Bray-1 P fum did not change with fumigation time, but nor did the Bray-1 P fum concentrations.
Consequently, the contribution of microbial P to Bray-1 P fum might be too small to detect in the δ 18 O P values of Bray-1 P fum or, as for Madden Dam, the δ 18 O P values of microbial P and P unf were similar.
T A B L E 3 Concentrations of phosphorus (P) (in mg kg −1 soil) extracted with anion exchange membrane (AEM) and Bray-1 solution without (P unf ) and with addition (P fum ) of hexanol (AEM) or chloroform (Bray- with increasing solution-to-soil ratio, but only to a certain threshold ( Figure 1). Given the advantages of the procedure, it seems likely to also have application for acidic soils in a variety of ecosystems worldwide.