How to Design a Study Including the Analysis of δ¹⁸O_P

Abstract

To plan a research study, one needs to (1) establish a research question, (2) make a set of observations, (3) form a hypothesis in an attempt to explain the observations and (4) test the hypothesis based on the data collected. The following questions should be addressed when designing a study including the analysis of δ¹⁸OP: (i) what is the research hypothesis? (ii) what is the main objective of the study? (iii) what are the aims to address these objectives? and (iv) which techniques are appropriate to address such research question. In addition, one needs to consider (1) which kind of samples needs to be collected, e.g. soil, vegetation or water? (2) in case of soil and sediment samples, which sampling depths and increments need to be sampled? (3) which P pools need to be extracted and analysed for the corresponding δ¹⁸OP values? (4) when and how often should samples be taken and (5) how many samples can be processed per week?

5.1 Planning or Designing a Study

To plan a research study, one needs to (1) establish a research question, (2) make a set of observations, (3) form a hypothesis in an attempt to explain the observations and (4) test the hypothesis based on the data collection.

The following questions should be addressed when designing a study including the analysis of δ¹⁸O_P:

• 1. What is the research hypothesis? (i) What is the main objective of the study? (ii) What are the aims to address these objectives? (iii) Which techniques are appropriate to address such research question?

• 2. Which kind of samples needs to be collected, e.g., soil, vegetation or water?

• 3. In the case of soil and sediment samples, which sampling depths and increments need to be sampled?

• 4. Which P pools need to be extracted and analysed for the corresponding δ¹⁸O_P values?

• 5. When and how often should samples be taken?

• 6. How many samples can be processed per week?

  In addition to the above questions, (1) monitoring of progress towards results, resources consumed and budget, (2) a reflection and a shared experience and lessons drawn from success and failure, should be considered a set of observations, (3) form a hypothesis in an attempt to explain the observations and (4) test the hypothesis based on the data collection.

5.2 Comments to the Questions

• Question 2: For example, when investigating P cycling in a lake, collecting samples from all potential P inputs into the lake is necessary.

• Question 3: For soils, the top 20 cm are often the most biologically active and hence most studies focus on this layer. When investigating the fate of fertilizer P, taking samples at different depths might be necessary.

• Question 4: This depends a lot on the research hypothesis. When investigating inorganic P for example, purifying the organic P in NaOH-EDTA extract is probably not necessary.

• Question 5: As mentioned by Pistocchi et al. (2017), timing is crucial when working with river (water and sediment) samples. Fast flow in streams, for example after storm events, can re-suspend a large amount of river sediments, which is making it more challenging to determine other sources of particulate P. Low flow in streams, could on the other hand lead to an overprinting of the original δ¹⁸O_P values due to a longer residence time, and hence more time for the equilibrium to be reached.

• Question 6: Especially question 6 should not be underestimated because ideally samples should not be stored for an extended period due to the potential alteration of the δ¹⁸O_P value during storage. In most laboratories, places on the stirring plate and/or in the water bath are limiting the number of extracts that can be purified at once. Realistically, in most cases, only 15 samples can be processed for the purification within one week.

Once questions 1 to 6 are answered, one could think about analyses that complement the δ¹⁸O_P analysis. Essential analyses which should always be included are:

• A. P concentrations (inorganic and total) in all extracts and water samples from aquatic systems.

• B. The δ¹⁸O_w of soil, plant or river water.

• C. The ambient temperature of the air, water or soil.

Depending on the hypothesis, it might also be useful to include further analyses. The δ¹⁸O_P has been successfully combined, in incubation and glasshouse studies, with ³³P in soils (Helfenstein et al. 2018; Siegenthaler et al. 2020) and plants (Pfahler et al. 2017). Also, the combination with metagenomics and other microbial analyses becomes more popular (Bi et al. 2018; Shen et al. 2020). Table 5.1 shows examples of analyses and data typically used along the δ¹⁸O_P for investigating P cycling in the environment.

Table 5.1 Examples of studies using the isotopic composition of oxygen bound to phosphorus (δ¹⁸O_P) to investigate P cycling in different environments

5.3 External Quality Assurance/Control

Quality assurance of isotope measurements by TC/EA-IRMS is based on the trueness and precision of the values from external standards that are analysed along with the samples in daily sequences (Watzinger et al. 2021). Most commonly, researchers use two benzoic acid standards provided by the IAEA (IAEA-601 with a δ¹⁸O value of 23.14‰ and IAEA-602 with a δ¹⁸O value of 71.28‰), a commercially bought silver phosphate, and sometimes also silver phosphate produced in-house. Before a new batch of silver phosphate, bought or produced in-house, is used as a standard, it should be sent to at least one additional laboratory for cross-validation of the δ¹⁸O_P value.

An inter-laboratory study for silver phosphate standards was conducted by Watzinger et al. (2021). A silver phosphate reference material was produced by the University of Natural Resources and Life Science (Austria) and sent to four other laboratories: The University of Western Australia (Australia), the ETH Zurich (Switzerland), the University of Helsinki (Finland) and the Helmholtz Centre for Environmental Research (Germany). This new reference material has an δ¹⁸O_P value of 13.8 ± 0.3‰ and is available for research laboratories.

In addition, the purification protocol should be tested in an inter-laboratory study using different materials: pure potassium dihydrogen phosphate and silver phosphate as a control and a set of different dried soil and sediment samples. A water sample would also be nice; however as water samples should be processed right away after sampling, storing and sending water samples to different laboratories might alter the δ¹⁸O_P signature due to an extended storage period.

5.4 Interpretation of Isotopic Data from the TC/EA-IRMS and Its Applications

There is no one-fits-all approach when it comes to the interpretation of δ¹⁸O_P data especially since not all variables influencing δ¹⁸O_P values are known. How to interpret δ¹⁸O_P data strongly depends on the research question, the research subject (soil, lake water, marine sediments etc.), and the data itself. However, there are some general steps/rules which can help to interpret the δ¹⁸O_P data:

• The first step in the interpretation of the obtained δ¹⁸O_P values is the calculation of the temperature-dependent equilibrium δ¹⁸O_P value. This value is an indication for intracellular cycling of P via the enzyme inorganic pyrophosphatase (PPase) and is thus assumed to be a good approximation for the expected δ¹⁸O_P of microbial P. As a general rule of thumb, a difference of 1‰ between a measured δ¹⁸O_P value and a calculated equilibrium value is often not very relevant due to the uncertainties associated with the calculation of the equilibrium value.

• The next step is the calculation of theoretical δ¹⁸O_P values of inorganic P released by the hydrolysis of organic P via enzymes. If organic P δ¹⁸O_P values are unknown, already published organic P δ¹⁸O_P values could be used as an approximation.

A good starting point for the interpretation of δ¹⁸O_P data is the comparison with the theoretical equilibrium value. Values lower than the equilibrium value are often an indication for hydrolysis of organic P but could also be due to other P inputs with a lower δ¹⁸O_P value like P from igneous rocks. Values higher than the equilibrium value could for example be caused by inorganic P leached from plants as plant inorganic P δ¹⁸O_P values tend to be enriched in ¹⁸O compared to other P pools (Table 5.1).

The following examples are very simple case studies and only show the general workflow when designing a δ¹⁸O_P study.

5.5 Example Research Study—Small Lake

A small lake, surrounded by five agricultural fields, is suffering from eutrophication during the summer. Each of the agricultural fields has a drainage didge which flows into the lake. Three of those agricultural fields are fertilized with mineral P fertilizers, the other two with farm-yard manure. One river, with a wastewater treatment plant (WWTP) upstream, is fed into the lake.

• 1. Hypothesis/objective: Identifying the main P sources into a lake

• 2. Samples: Water samples from the lake, the river, the drainage didges and the WWTP; fertilizers applied to the fields; and soil samples from the agricultural fields

  • a. Eight water samples

  • b. Five fertilizer samples

  • c. One bulked soil sample from each field

• 3. Sampling depths/increments: only relevant for soil, topsoil (0–20 cm); if possible, samples should be taken also throughout the soil profile to determine background δ¹⁸O_P values

• 4. P pools: soluble reactive P in water samples; water extractable and HCl P of fertilizers; resin and HCl P from soils

• 5. Sampling time points: winter and summer

• 6. Processing samples: 15 per week (based on laboratory equipment).

To calculate the total number of samples for δ¹⁸O_P analysis, one needs to now consider the following:

• A. Factor 2, because two sampling time points

• B. At each time point:

  • a. Eight water samples (SRP only)

  • b. Two P pools for each fertilizer sample, including a factor 2 for the HCl P due to using ¹⁸O-labelled and unlabelled HCl: 15 samples

  • c. Two P pools for each soil sample, including a factor 2 for the HCl P due to using ¹⁸O-labelled and unlabelled HCl: 15 samples.

In total, 76 samples need to be analysed and afterwards purified, with 38 samples at each sampling time point, and 38 samples cannot be analysed within one week. It is recommended to first deal with the samples/extracts which are more susceptible to changes due to biological activities. One would therefore start with the water samples and extract the more labile P pool (water-extractable or resin P) from the soils and farm-yard manure.

5.6 Example Research Study—Agricultural Fields

To investigate P cycling at agricultural fields where rapeseed is cultivated, five agricultural fields are selected as study sites. Those five fields are in the same climatic zone, but differ in their soil properties, like soil pH and P saturation index. All five agricultural fields are fertilized with the same mineral P fertilizers.

• 1. Hypothesis/objective: Does P cycling change along a soil profile (100 cm) where rapeseed is grown?

• 2. Samples: mineral fertilizer applied to the fields; soil samples from the agricultural fields; plant samples

  • a. One fertilizer sample

  • b. Soil samples from each field at different depths

  • c. Only above-ground plant sample

• 3. Sampling depths/increments: 100 cm, divided into 10 cm increments; if possible, samples should also be taken throughout the soil profile to determine background δ¹⁸O_P values

• 4. P pools: water-extractable and HCl P of fertilizers; resin, microbial, NaOH-EDTA Pi and P_org, and HCl P from soils; TCA P and NaOH-EDTA P_org for plant samples.

• 5. Sampling time points: peak of plant P demand

• 6. Processing samples: 15 per week (based on laboratory equipment).

To calculate the total number of samples for δ¹⁸O_P analysis, one needs to now consider the following:

• A. Factor 5, because five fields

• B. Two P pools for the fertilizer sample, including a factor 2 for the HCl P due to using ¹⁸O-labelled and unlabelled HCl: Three samples

• C. From each field:

  • a. Five P pools for each soil sample, including a factor 2 for the HCl P due to using ¹⁸O-labelled and unlabelled HCl: 60 samples

  • b. One bulked plant sample; two P pools; including a factor 2 for the TCA P due to using ¹⁸O-labelled and unlabelled TCA: Three samples.

In total, 318 samples need to be analysed and afterwards purified, with 63 samples at each field and three samples from the fertilizer. Those samples cannot be analysed within one week. It is recommended to first deal with the samples/extracts which are more susceptible to changes due to biological activities. One would therefore start with extracting the most labile P pool (resin and microbial P) from the soils. It might also be useful to analyse other parameters relevant for P cycling like enzyme activities.

Glossary

δ¹⁸O

The oxygen isotope ratio is conventionally given in the delta notation: δ¹⁸O = (R_sample/R_standard) − 1, where R_sample is the ¹⁸O/¹⁶O ratio of a sample and R_standard is the ¹⁸O/¹⁶O ratio of the Vienna Standard Mean Ocean Water (V-SMOW). δ¹⁸O_P is the δ¹⁸O value of a P compound or pool.

Soluble reactive P

SRP; considered the most bio-available P pool in water samples.

¹⁸O-labelled/unlabelled solution

¹⁸O-labelled and unlabelled solutions are used in case of acidic extractions, e.g., with 1 M HCl, to account for any oxygen exchange between phosphate and the solution during the extraction.

TC/EA-IRMS

A thermal conversion elemental analyser (TC/EA) coupled to an isotope ratio mass spectrometer (IRMS) is commonly used to determine the δ¹⁸O_P. The oxygen in silver phosphate is converted, via pyrolysis, into carbon monoxide, whose isotopic composition is then measured in the IRMS.

³³P

Radioisotope of phosphorus (P); half-life 25.4 days, beta emitter.