Sensitive determination of total particulate phosphorus and particulate inorganic phosphorus in seawater using liquid waveguide spectrophotometry.

Determining the total particulate phosphorus (TPP) and particulate inorganic phosphorus (PIP) in oligotrophic oceanic water generally requires the filtration of a large amount of water sample. This paper describes methods that require small filtration volumes for determining the TPP and PIP concentrations. The methods were devised by validating or improving conventional sample processing and by applying highly sensitive liquid waveguide spectrophotometry to the measurements of oxidized or acid-extracted phosphate from TPP and PIP, respectively. The oxidation of TPP was performed by a chemical wet oxidation method using 3% potassium persulfate. The acid extraction of PIP was initially carried out based on the conventional extraction methodology, which requires 1M HCl, followed by the procedure for decreasing acidity. While the conventional procedure for acid removal requires a ten-fold dilution of the 1M HCl extract with purified water, the improved procedure proposed in this study uses 8M NaOH solution for neutralizing 1M HCl extract in order to reduce the dilution effect. An experiment for comparing the absorbances of the phosphate standard dissolved in 0.1M HCl and of that dissolved in a neutralized solution [1M HCl: 8M NaOH=8:1 (v:v)] exhibited a higher absorbance in the neutralized solution. This indicated that the improved procedure completely removed the acid effect, which reduces the sensitivity of the phosphate measurement. Application to an ultraoligotrophic water sample showed that the TPP concentration in a 1075mL-filtered sample was 8.4nM with a coefficient of variation (CV) of 4.3% and the PIP concentration in a 2300mL-filtered sample was 1.3nM with a CV of 6.1%. Based on the detection limit (3nM) of the sensitive phosphate measurement and the ambient TPP and PIP concentrations of the ultraoligotrophic water, the minimum filtration volumes required for the detection of TPP and PIP were estimated to be 15 and 52mL, respectively.

World Precision Instruments, Sarasota, FL, USA), and a miniature fiber optic spectrometer 119 (USB4000, Ocean Optics, Dunedin, FL, USA). The spectrometer was connected to a 120 computer, and an absorbance at 708 was operated using Spectra Suite software (Ocean Optics, 121 Dunedin, FL, USA). The analytical reagents (molybdate and ascorbic acid solutions) were 122 prepared by using the methodology of Hansen  The absorbances of procedural blank (GF/F filter + 3% K 2 S 2 O 8 + purified water) and 141 reagent blank (3% K 2 S 2 O 8 + purified water) were compared to check P contamination of GF/F 142 filter. In this case, the absorbance of purified water (+colorimetric reagent) was set to zero. 143 The procedural blank was prepared by filtering 1L of purified water and it was processed 144 following the outlined digestion procedure. 145 The absorbance of standard solutions (20, 50, 100, 200, 500 and 1000 nM) was measured 146 in order to draw a calibration curve. Each standard that was dissolved in 1.5% K 2 S 2 O 8 was 147 prepared by mixing phosphate standards dissolved in purified water (40, 100, 200, 400, 1000 148 The reproducibility of TPP determination was obtained by analyzing field samples.  In this case, the absorbance of purified water (+colorimetric reagent) was set to zero. The 171 procedural blank was prepared by filtering 1L of purified water and it was processed through the 172 outlined extraction procedure. filter was extracted with 1 M HCl and the extract was dispensed into duplicate tubes, one for the 186 conventional protocol (ten-fold dilution with purified water) and another for the improved 187 protocol (neutralization with 8 M NaOH). After the ten-fold dilution and neutralization, the 188 two types of solutions were analyzed by the LWCC method. 189 The reproducibility of PIP determination through the improved protocol was obtained by 190 analyzing field samples, which were collected at the same station as the TPP samples. Sample The mean ± standard deviation (SD) of the absorbances of the procedural and reagent 199 blanks were 0.009 ± 0.001 and 0.009 ± 0.003, respectively (n = 3) (  (Fig. 1). The regression equation obtained is y = 0.0010x -0.0089, with r 2 = 213 0.9997 (n = 14), where y is the absorbance and x is the concentration of phosphate. The wide 214 linear dynamic range could be applicable to various oceanic samples. For example, if a 100 215 mL filtration volume is used, then 3-1000 nM phosphate corresponds to 1.2-400 nM of ambient 216 TPP, according to the following equation: 217 where C a is the ambient TPP concentration (1.2-400 nM), C p is the phosphate concentration (3-219 14 1000 nM), V r is the reagent volume (20 mL), DR is the dilution ratio (2) and V f is the filtration 220 volume (100 mL). 221 222

Concentration and reproducibility of the field sample 223
TPP concentrations of the field samples were 8.4 ± 0.36 nM (mean ± SD, n = 5) ( Table 2).
The filtration volume estimated was 67-800 times lower than that of previous studies (1-12 L) 232

Filter blank 236
Mean ± SD of the absorbances of procedural and reagent blanks were -0.016 ± 0.002 and -237 0.018 ± 0.002, respectively (n = 3) ( Table 1). The mean absorbance between the two blanks 238 was not significantly different (t test, p > 0.05), as was the case for the filter blank test for TPP. 239 This indicates that P contamination of the GF/F filter was also negligible in the case of PIP 240 determination. The absorbances of both procedural and reagent blanks were lower than that of 241 purified water. This is probably due to the difference in refractive index between ionic 242 solutions (1 M HCl + 8 M NaOH) and purified water [28]. Therefore, it is necessary to use the 243 neutralized solution as an analytical blank. 244 245

Calibration curve 246
A calibration curve was obtained from the absorbances of each duplicate standard dissolved 247 in the neutralized solution (Fig. 2)

Absorbance comparison with the conventional protocol 256
A calibration curve for the conventional protocol was also obtained from the absorbances 257 of each pair of phosphate standards that were dissolved in 0.1 M HCl (Fig. 2). The curve 258 showed a strong linear correlation up to 1000 nM (r 2 = 0.9998), which was the same as that by 259 the improved protocol. However, the absorbances of the standards in the conventional 260 protocol were significantly lower than those of the improved protocol (paired t test, p < 0.05, n 261 = 7). Aspila et al.
[6] used the ten-fold dilution of 1M HCl with purified water to remove the 262 effect of acidity on phosphate analysis. However, the incomplete removal of acid could be the 263 reason behind the lower absorbances in the conventional protocol [26,27]. In addition to 8.9 264 times higher sensitivity in the improved protocol than the conventional protocol by decreasing 265 dilution ratio from 10 to 9/8, sensitivity of the improved protocol further increased by 2.3% 266 compared to that of the conventional protocol when taking into account a slope ratio of two 267 regression lines (0.001069/0.001045). 268 PIP concentrations of the field samples were 1.3 ± 0.08 nM (mean ± SD, n = 4) ( Table 2). 277 Because of the low CV (6.1%), this method provides high-precision measurements even for 278 ultraoligotrophic water. Yoshimura et al. [5] reported that typical proportions of PIP to TPP in 279 subtropical and subarctic regions range between 10 and 20%. In this study, the proportion of 280 PIP to TPP was 15%, which is within the typical range, and the concentration of POP (which is 281 obtained by subtracting PIP from TPP) was estimated to be 7.

Conclusions 288
The present study established sensitive methods for the determination of TPP and PIP in 289 the oligotrophic oceans. The proposed methods possess two distinct advantages over the 290 conventional methods. Firstly, significant decreases in filtration volumes for TPP and PIP 291 were performed through the application of the LWCC method. Secondly, the improved PIP 292 protocol was more sensitive than the conventional protocol in terms of the decrease in the 293 dilution ratio of 1 M HCl extract and the increase in the absorbance of the colorimetric 294 determination of phosphate. This also contributes to the decrease in the filtration volume. 295 The small filtration volumes enable rapid sample accumulation in the field. Field observations 296 revealed that the methods could detect very low concentrations of TPP and PIP with high 297 precisions even in ultraoligotrophic water. The methods are considered to be valuable in 298 understanding the role of particulate P in the oceanic P cycle. 299 300 Acknowledgments 301 We thank the officers, crew members, and members of the scientific party of the RT/V 302