In situ observation of localized, sub-mm scale changes of phosphorus biogeochemistry in the rhizosphere

Aims We imaged the sub-mm distribution of labile P and pH in the rhizosphere of three plant species to localize zones and hot spots of P depletion and accumulation along individual root axes and to relate our findings to nutrient acquisition / root exudation strategies in P-limited conditions at different soil pH, and to mobilization pattern of other elements (Al, Fe, Ca, Mg, Mn) in the rhizosphere. Methods Sub-mm distributions of labile elemental patterns were sampled using diffusive gradients in thin films and analysed using laser ablation inductively coupled plasma mass spectrometry. pH images were taken using planar optodes. Results We found distinct patterns of highly localized labile P depletion and accumulation reflecting the complex interaction of plant P acquisition strategies with soil pH, fertilizer treatment, root age, and elements (Al, Fe, Ca) that are involved in P biogeochemistry in soil. We show that the plants respond to P deficiency either by acidification or alkalization, depending on initial bulk soil pH and other factors of P solubility. Conclusions P solubilization activities of roots are highly localized, typically around root apices, but may also extend towards the extension / root hair zone. Electronic supplementary material The online version of this article (10.1007/s11104-017-3542-0) contains supplementary material, which is available to authorized users.

Plant P availability and soil pH. P availability to plants is generally considered to be controlled by Fe and Al in acidic soils, and by Ca at higher soil pH, therefore, plant P availability is commonly considered to be highest between pH 6 to 7 (Price, 2006). However, (Barrow, 2017) challenges this concept by presenting evidence for P availability to continuously increase as pH decreases from ~7 to ~4, as both P desorption from the solid phase and plant P uptake increase concomitantly with decreasing pH. Below pH 4, plant P uptake decreases again, which might be connected to increasingly soluble Al. In an experimental and modeling study on soil P desorption, Weng et al. (2011) observed, that P solubility remains low between pH 3 and 7 in low-P soils, whereas there was a solubility maximum around pH 4, low P solubility between pH 6-8, and an increase of P solubility between pH 8 and 10 in high-P soils. Eriksson et al. (2016) also reported experimental data on pH-dependent solubilization of P in non-calcareous and calcareous soils of Swedish longterm experiments. In the Swedish soils, a minimum of P solubility between pH 5 and 7 in unfertilized control soils was found, whereas no consistent pattern was observed in the fertilized treatments. This evidence suggests, that P solubility is generally low between pH 5 and 7, but the location of the minimum depends on P load and soil properties such as carbonate and organic matter content, the presence of polyvalent cations, and distinct P minerals.
Association of P with the soil solid phase. Recent advances in solid-state P speciation techniques such as XANES revealed Ca phosphate minerals (apatites) derived from the parent material mainly to be present during the earliest stages of soil formation. Apatites dissolve during weathering, and the contained P redistributes to organic forms and is sorbed to Fe and/or Al oxides and hydroxides (Liu et al., 2013;Prietzel et al., 2013). Therefore, native, low P soils typically contain no or only small amounts of Ca phosphates, especially in the presence of larger amounts of Fe (Eriksson et S-3 al., 2016;Hashimoto and Watanabe, 2014;Zhang et al., 2014), while fertilized soils almost consistently show neo-formation of Ca phosphates in response to high P and Ca inputs (Eriksson et al., 2016;Luo et al., 2017;McLaren et al., 2015;Zhang et al., 2014). Ca phosphates can occur even in acidic, but continuously fertilized soils (Beauchemin et al., 2003). Moreover, defined Fe or Al phosphate minerals have been detected in soils, but these observations are scarce and typically show only minor amounts of these minerals (Beauchemin et al., 2003;McLaren et al., 2015). These detailed data on mineral P species in soils support the accepted concept of inorganic soil P being mainly associated with Fe and Al via sorption, and only small amounts being present as defined Ca, Fe and Al minerals.
Phosphorus content classes of the experimental soils. The P content classes used for P fertilization in Austria, based on calcium acetate lactate (CAL) extraction, are given in Table S1 (Baumgarten et al., 2017).  and Zr-Hydroxide/Hydromed D4 solution (for P) was used. This lead to a homogeneous loading of target analytes located at the top of the standard gels. Four standard gel sheets with varying loading were produced using this method and four gel discs were cut from each of these sheets.
Three gel discs were digested using microwave assisted digestion while the fourth gel was used as laser ablation standard. The fourth gel disc replicate was dried and used as calibration standard during LA-ICP-MS (Kreuzeder et al., 2013). The isotopes 31 P, 27 Al, 44 Ca, 57 Fe, 55 Mn were analyzed along with 13 C, which was used as internal normalization standard.

S-7
Organic carbon, malate and citrate exudation of wheat and buckwheat. Figure S2. Release of organic carbon by (a) wheat and (b) buckwheat roots. (c) exudation of malate and of (d) citrate by wheat and buckwheat. NF: not fertilized. Error bars show the standard error (n = 3). Letters indicate significant differences between fertilizer×soil and fertilizer×plant×soil treatments (Student-Newman-Keuls, P < 0.05).

S-8
Development of P depletion zones around roots. Figure S3. Development of a P depletion zone around a root of wheat. The shorter root (right root) had a delay of 3 days in comparison with the longer root. The imaging areas for the DGT are indicated with dotted lines. The relative P-flux is indicated in the calibration bar. The wheat plant was grown on calcareous soil with a NH4 fertilizer treatment. The length of the scale bar is 1 cm.

S-9
Representativeness of this study and sampling related errors. DGT has been used as a sampling method for the investigation of sediments and soils previously and is a well-established technique (Lehto et al., 2012;Stockdale et al., 2008;Williams et al., 2014). In soils, however, sampling based on DGT requires a diffusive flux and therefore a relatively high water saturation is required which may lead to anoxia and changed redox-conditions. In this study, the used water saturation was kept as low as possible (50-80% of the maximum water holding capacity) to avoid such effects and the sampling was carried out swiftly. In previous work large Mn-patches were observed when high water saturation was used, which were related to redox-artefacts (Hoefer et al., 2015). In this study, no such effects were observed.
The occurrence of air bubbles or the behavior of plant roots sometimes impede the sampling with planar optodes and DGT gels. All experimental treatments were carried out in three replicates to ensure a complete dataset. The experimental data provided in Table 2 of the main document gives an overview on the observed plant variability and reproducibility of the results. Furthermore, the manipulation of the rhizotron covers always bears the possibility of root injury. This was avoided by using a protective membrane throughout the experiment and by gentle handling of the rhizotron covers. No obvious artifacts could be observed due to root injury in this study.