Manipulations to increase soil arsenic availability, fern biomass, and fern arsenic uptake through nutrient application and mycorrhizal fungi inoculation have been investigated to increase phytoextraction rates. In particular, complex relationships between arsenic and phosphorus availability in soil and uptake in P. vittata vis-à-vis fern nutrient deserve fur ther elucidation, including the effects of soil phosphorus relative to supple mental phosphorus . According to the phosphorus starvation theory , arsenic uptake in P. vittata could be a byproduct of nutrient acquisition, especially iron and phosphorus. Phosphate and arsenate are chemical analogues found in soils associated with iron oxides and can be released in the rhizosphere through similar bio-geochemical processes . Furthermore, P. vittata associates with mycorrhizal fungi including Funneliformis spp. . Like hyper accumulators, mycorrhizal fungi evolved under phosphorus starvation conditions, can support plants under nutrient deficiency, drought, and metal stress, and have stress tolerance mechanisms that differ from and are possibly complementary to those of hyper accumulators . Inoculation with mycorrhizal fungi could increase P. vittata biomass , nutrient, water, and possibly arsenic access , with species adapted to local soils important in addition to generalists . Moreover, soil clay content affects nutrient and arsenic availability . Clay-sized particles, the smallest of soil particles, include mineral phases such as iron oxides, aluminum oxides,nft hydroponic and silicate clay minerals that provide a large surface area for adsorption of arsenic, phosphorus, and other nutrients .

Arsenic has been shown to be more plant available and more leachable in soil with lower clay content. Few arsenic phytoextraction studies have investigated arsenic leaching during P. vittata growth or computed arsenic budgets. After P. vittata growth, similar or lower levels of arsenic in leachate were found compared to non-phytoextracted soils. In long-term field studies, greater arsenic depletion has been observed for soil below the fern root zone, compared to surface depths , which could indicate leaching of arse nic below the root zone. It is important to quantify arsenic cycling in the P. vittata-soil-water system during growth, because P. vittata growth processes might lead to spatiotemporally heterogeneous arsenic leaching. Unsaturated flow-through soil column studies with plants are a powerful setup to quantify arsenic cycling in controlled whole plant-soil-water systems that approximate natural systems. With such systems, water balances including transpiration can be calculated, and arsenic transport in soil pore water can be quantified . Combining such soil columns studies with X-ray absorption near edge structure spectroscopy and X-ray fluorescence imaging at the micron scale allows us to investigate rhizosphere processes and relate rhizosphere processes to arsenic cycling at the system scale. We conducted a 22-week soil column study with moderately contaminated soil planted with P. vittata to determine the effects of soil texture , phosphorus application, and fungi inoculation on arsenic cycling in the plant-soil-water system. We used synchrotron-based spectromicroscopy to determine arsenic speciation in bulk and rhizosphere soils and propose mechanisms of arsenic mobilization for plant uptake and/or leaching. We observed greater arsenic uptake and leaching with smaller fern biomass in soil with lower clay content across all treatments, suggesting significant plant growth trade-offs associated with arsenic hyper accumulation and phytoextraction.

Two soils moderately contaminated with arsenic and with contrasting textures were excavated from the 0–30 cm depth in a former railroad right-of-way . Soils were sieved and stored field-moist at 5 °C in sealed containers under oxic conditions. Soil pH was measured on triplicate samples and cation exchange capacity and percent organic matter were measured on duplicate samples as described previously . Bulk density was measured in triplicate on intact cores of a known volume from which the volume of the fraction >2 mm was subtracted, following a modified protocol for rocky soils . Porosity was calculated from bulk density . Soil texture was measured in triplicate with the hydrometer method . In both soils, clay minerals identified with X-ray diffraction following USGS protocols included nontronite, trioctahedral montmorillonite, and/or vermiculite.Experiments with novel unsaturated flow-through soil columns with plants were performed in triplicate to estimate arsenic uptake by P. vittata, arsenic desorption and transport in soil, and arsenic leaching. We compared control columns without amendments with columns fertilized with phosphorus, and columns inoculated with the fungus F. mosseae . Phosphorus was mixed into soil before packing columns. Fungal inoculant was applied to each hole at transplanting. Column experiments without plants were also performed, with one replicate per soil per treatment for control and phosphorus treatments. However, because fungal inoculant was not expected to survive in the absence of ferns, unplanted column experiments with fungal inoculation were not performed. For the remainder of the manuscript, any mention of columns in text refers to planted columns, unless specifically noted. Soil columns were made of acrylic tubes . At the bottom of each column a 0.2 μm filter and glass fiber filter separated soil from the effluent port, which was filled with quartz sand with glass wool on either end. To ensure reproducibility between columns, they were all packed similarly at bulk density representative of field conditions .

Soil was well-mixed be fore weighing into 20 increments of equal mass appropriate for packing into a 2 cm depth. Soil was packed to a total depth of 40 cm using a custom-made 3-pronged device to apply downward force over a small surface area to avoid formation of layers. P. vittata ferns with bare roots and 3–7 fronds 10–20 cm in length were planted with roots 0–10 cm deep in columns. Pore water samplers and tensiometers were filled with degassed synthetic rain and inserted horizontally at depth with the tip 7.5 cm inside planted columns.A synthetic rain solution that served as column influent was made following rain composition of the field site adjusted to current rainfall pH . Globally, rainfall pH ranges from 3.5 to 8.0 so pH 5.33 was considered widely applicable. The synthetic rain was kept in the dark during experiments. The influent solution was supplied by a single port at the rhizome, and was eluted with a peristaltic pump through planted columns for 22 weeks at a constant flow rate equivalent to maximum daily rainfall . The flow rate was chosen to approximate conditions encountered in the field, made constant to simplify experiment design, and is representative of high rainfall climates and irrigated environments. A conservative bromide tracer was eluted through unplanted control columns for half a pore volume and the synthetic rain was eluted for 2 pore volumes at the same flow rate as used in the planted columns. The column experiments were conducted in a greenhouse with a 16-h photoperiod.Pore water samples were extracted from planted columns approximately every 2 weeks using acid-washed syringes maintained at 4 °C during collection. Pore water samples were immediately filtered under anoxic conditions, sub-sampled for analyses,blueberry pot size and stored at −18 °C in the dark until analysis. Planted column effluent was collected weekly into polyethylene bottles pre-acidified with concentrated HNO3 . After collection, effluent was adjusted to 2% HNO3, filtered , and stored for analysis. Unplanted column effluent was col lected every 228 min using a fraction collector. After collection, samples for bromide analysis were reserved and samples for arsenic analysis were combined on a weekly basis and treated the same as the planted column effluent. Pinnae samples were collected at 11 and 21 weeks. At the end of the experiments, all above ground biomass was removed 2 cm above the rhizome and separated based on development of sori and tissue senescence .

Each soil column was sliced at designated depths . The entire rhizome was removed, though not the entire root mass. All root and soil samples were immediately stored at 4 °C.Above ground biomass was washed 3 times in deionized water to remove soil particles and dust . Rhizomes and root samples for elemental analysis were washed in deionized water until clean of soil. Fronds, rhizomes, and root samples for elemental analysis were dried for 24 h or until constant mass was achieved. Whole root samples for microprobe and DNA sequencing were flash-frozen in liquid nitrogen within 6 h of harvesting and stored at −80 °C until analyses. Additional root samples were dried, processed, and maintained under anoxic conditions for X-ray absorption spectroscopy . Soil samples were either air-dried for elemental analysis or dried, processed, and maintained under anoxic conditions for XAS.Dry plant biomass was measured on pinnae, frond, and rhizome samples. Pinnae sample biomass served as a proxy for whole plant biomass during growth. Plant tissue and soil samples were digested following a modified EPA 3050B protocol described previously . Total arsenic, phos phorus, and iron concentrations of soil and fern digests were determined using inductively coupled plasma optical emission spectroscopy . The detection limit was 20 μg/L for each analyte. Extractable soil nutrient concentrations were measured as described previously . Total arsenic concentrations in effluent and pore water samples were analyzed using either ICP-OES or, if concentrations were less than 50 μg/L, hydride generation-ICP-OES after addition of 0.8 mL of 40% KI/8% ascorbic acid to 4 mL sample and 2.7 mL 1.1 M HCl . In pore water, arsenic concentrations were determined using HG-ICP-OES in samples buffered at pH 5.0 through addition of 2.5 mL 0.5 M disodium citrate to 2.5 mL sample, dissolved organic carbon concentrations were analyzed using an O-I-Analytical analyzer, and pH was determined using a Denver Instruments meter with pH/ATC Sartorius ATC combination electrode. Bromide concentrations were measured using a Dionex Ion Chromatograph with a detection limit of 2.4 mg/L. Samples of roots with rhizospheric soil, of soil aggregates, and of bulk soil were prepared for XAS, and measurements were collected as described pre viously . Briefly, whole roots and root or aggregate thin sections were mounted onto a Peltier cooling stage for X-ray micro probe analysis, X-ray flfluorescence mapping and X-ray absorption near-edge structure , at the Advanced Light Source XFM beamline 10.3.2 , Lawrence Berkeley National Laboratory . Bulk X-ray XANES spectra were collected at Stanford Syn chrotron Radiation Laboratory beamline 7.3 on bulk powdered samples mounted on filters. Least-square linear combination fitting was performed with custom LabVIEW software and XAS databases of iron and arsenic compounds available at beamline 10.3.2. Details of the spectromicroscopy measurements are available in the Supplemental Information.Transport and desorption of total arsenic within the soil columns was modeled using a solute transfer model through unsaturated soil amended with a retardation factor to account for the desorption of arsenic from soil. In unsaturated soils, the transport of solutes occurs in two phases: a mobile phase where water flows through the porous medium and an immobile phase in which water is stagnant within dead-end pores or between soil particles. Briefly, bromide tracer breakthrough curves collected from unplanted columns were used to constrain transport parameters and to check for non-ideal flow conditions in the columns. Adsorption constants were fit that best described the total arsenic desorption from each of the four unplanted soil columns. Changes in arsenic concentrations in the solid phase , where Cs_m and Cs_im are the arsenic concentration within the mobile and immobile solid phase , respectively, were derived from isotherm equations, . Since the binding sites in the soil are non-uniform, the Freundlich isotherm was used in the model as it represents multiple types of binding sites. Full details are available in the Supplemental Information.Statistical analysis was performed in R . Analyses of covariance were performed on linear models to analyze main effects of explanatory variables including time , depth , soil, and treatment on response variables including fern arsenic concentrations, biomass, arsenic accumulation, arsenic uptake rate, and changes in soil arsenic concentrations during phytoextraction. Models were selected based on AIC criteria using the MuMIn package . Two-way interactions were included when those models were most highly ranked. Regression summaries were used to compare effects of explanatory variables, and differences in means were determined with Tukey’s Honestly Significant Difference test on means normalized to initial soil arsenic concentration where appropriate. A paired t-test was used to determine the mass balance across all columns, com paring mass arsenic leached per column to mass arsenic accumulated per fern. Effects of soil and treatment on the AMF assemblage were determined using a permutational multivariate analysis of variance with pairwise comparisons.