Elemental analysis of the packaged sand yielded levels of total carbon and nitrogen below the detection limit, indicating that there was essentially no organic matter or nitrogen to account for in the sand media at the start of the ITNI experiments. The 15N labelled tracer used in the ITNI experiments was made by dissolving ACS grade KNO3 and 98 Atom Percent 15N-KNO3 in 18 megaohm deionized water in a 1-liter volumetric flask to achieve an isotopic composition of 1.01 AP 15N, and a nitrate concentration of 360 millimoles per liter. The concentration of nitrate was adapted from similar hydroponic experiments conducted in Fiest and Parker 2001. The 1.01 AP 15N tracer abundance produced a strong isotopic enrichment in the ITNI system, but was still within the range of 15N concentrations that could be accurately measured on our existing isotope ratio mass spectrometer. Other macro and micro nutrients were added to the KNO3 solution to essentially produce a 15N-enriched Hoagland solution . During ITNI deployment, seedlings were gently planted into the sand reservoirs of the ITNI modules, with any green, aerial leaves or stems resting above the sand surface. Then, each module’s liquid reservoir was dosed with 50 milliliters of the 15Nenriched Hoagland solution resulting in an initial isotopic enrichment of 1.01 15N AP, an initial nitrate concentration of 2 millimoles per liter and an initial concentration of 0.475 g/L of standard Hoagland solution. Following dosing,growing berries in containers the air lift systems were immediately engaged to circulate the nutrient solution within each ITNI module. The ITNI modules were stored for two weeks in a climate controlled greenhouse to ensure thorough distribution of 15N tracer into the different PLS components before deployment at the two field sites.
Beginning with the second ITNI deployment, the modules’ liquid reservoirs were wrapped in aluminum foil to discourage algal growth. During all deployments, the modules were spaced equidistant apart from one another surrounding a storage bin, in a square pattern . This storage bin housed the air compressor, the timer for the air compressor and air-line manifolds used to distribute compressed air to the air lift systems in the ITNI modules. At Motte reserve, since there were no electrical hookups for the pump and timer, a solar panel and accompanying deep cycle marine battery and solar controller were utilized as an energy source for the air lift system . Modules were watered 4 to 5 times throughout the day for approximately 1-2 minutes duration, depending on the season and transpiration needs of the plants. The ITNI modules were “shuffled” every two weeks while deployed to reduce confounding influences from a singular deployment location. Deionized water was added to the ITNI modules every 2-3 days to maintain the liquid reservoir volume at 9 liters, but no additional spikes of isotopically-labelled Hoagland’s solution were made. The modules were exposed in the field during the following time periods: Deployment 1: March 8, 2013 to May 20, 2013, Deployment 2: May 28, 2013 to August 12, 2013 and Deployment 3: November 15, 2013 to March 24, 2014. At the end of a deployment the modules were disconnected from the air lift system and immediately returned to the greenhouse for sample processing. Above-ground plant material from each module was individually clipped above the sand surface and placed into paper bags. Plant material bags were dried at 60°C for 2 days to drive off moisture and prevent microbial decay of the plant matter. We collected solution from the liquid reservoirs at the end of the deployments by passing the water through a 0.45 membrane filter using a filter holder and syringe. These 125 milliliter aliquots were put into clean HDPE bottles and stored in a freezer at -20 °C until analyzed for nitrate and ammonium concentration and the stable isotope composition of nitrate.
A second 125 aliquot of unfiltered water was collected from each module and stored frozen for later determination of total nitrogen. Total nitrogen was analyzed using a persulfate digestion and EPA method 353.2 . Total nitrogen measured by this technique includes several forms of N: i) particulate nitrogen , ii) dissolved inorganic N and iii) dissolved organic N . Because there is no currently available method for measuring the AP 15N of TN, we could not include its mass and isotopic composition in the deposition computations. At the end of the deployments, the sand in each module was passed through a ¼ inch metal screen to separate out the plant roots which were dried and weighed as previously described. The sifted sand from each module was individually weighed, thoroughly-mixed and then duplicate sub-samples were collected from each module for measurement of KCl extractable nitrogen . Note: We decided against measuring nitrogen content and stable isotope composition of the sand directly using the elemental analyzer inlet to the mass spectrometer. First, the N-content of the sand was very low, necessitating that relatively large samples be analyzed in the EA, and which proved problematic to combust properly. Second, the spatial heterogeneity of N-content in the sand was fairly large so that many, 25 mg replicates would be required to get a representative sample of the sand’s N-content and δ 15N for the ITNI computation. KCL extracts allowed us to base our estimates of Ncontent and isotope composition on much larger sub-samples of the sand . Water samples from the modules and KCl extracts from the sand were analyzed for nitrogen concentration and 15N abundance. Inorganic nitrogen concentrations were measured on a discrete analyzer using the following methods: nitrate+ nitrite = EPA 353.2 and ammonium = EPA 350.1. Ammonium levels in the water and KCl extracts were below the detection limit so we confined isotopic measurements to nitrate only.
The δ15N and δ18O of nitrate+nitrite were measured using the microbial denitrifier method at the Facility for Isotope Ratio Mass Spectrometry at UC Riverside. δ15N and δ18O values were measured using a Thermo Delta V isotope ratio mass spectrometer . Three reference standards, USGS-32, USGS-34 and USGS-35, were used for calibration. Isotopic abundances are expressed in standard delta notation relative to VSMOW for oxygen and relative to atmospheric N2 for nitrogen. The nitrogen content and stable isotope composition of above- and below-ground plant material were analyzed using a Costech elemental analyzer connected to the Thermo Delta V IRMS. Prior to analysis, the plant materials were ground and homogenized using a Pyrex-glass mortar and pestle. The δ15N values for the plant materials are expressed in standard delta notation relative to atmospheric N2. As our study was a proof-of-concept investigation for the use of the ITNI method in arid and semi-arid regions,blueberry containers we employed a variety of treatments in the three deployments . During Deployment 1 we created 3 treatments at both Motte and Riverside: i) modules spiked with 15N-enriched Hoagland solution that contained water, sand and Red Brome plants, ii) control modules spiked with 15Nenriched Hoagland solution that contained water and sand only and iii) control modules containing water and sand that were spiked with 15N-enriched nitrate only . The design of the Deployment 1 allowed us to: i) compare deposition rates to Red Brome at Motte and Riverside, ii) observe a complete ITNI module in relation to a module without a plant and iii) to compare the behavior of the 15N spike in plant-free modules containing complete Hoagland solution versus ones containing only potassium nitrate. In Deployment 2 we created 2 main treatments at both Motte and Riverside: i) modules spiked with 15N-enriched Hoagland solution that contained water, sand and summer mustard , and ii) modules spiked with 15N-enriched Hoagland solution that contained water, sand and California buckwheat . A single Hoagland control was also run at each site, but the data were not used to compute N deposition rates. The design of the Deployment 2 allowed us to compare deposition rates for both an invasive and native plant at a single site and between sites. Based on the results from Deployment 1, all of the modules in Deployment 2 were wrapped with aluminum foil to reduce light-levels and algal growth in the liquid reservoirs. Prior to Deployment 3 we had trouble growing seedlings of California buckwheat so we focused this experiment on summer mustard. Two treatments were created and deployed at Riverside only: i) modules spiked with 15N-enriched Hoagland solution that contained water, sand and summer mustard , and ii) control modules containing water and sand that were spiked with 15N-enriched Hoagland solution . A single California buckwheat module was created and operated, but the data from this module was not included in the figures and data analyses for Deployment 3.
The design of Deployment 3 allowed us to measure N-deposition at Riverside and to further assess the fate of the 15N tracer in modules not containing plants. Detailed ITNI module specifics for each Deployment are addressed in Tables 1.1, 1.2, and 1.3. Mean values of treatments will be addressed in this section and in the accompanying tables. During Deployment 1, all module treatments recovered less than 100% of the tracer spike for both Riverside and Motte Reserve . At Riverside, the Hoagland Control treatment recovered the most tracer spike among all treatments at 91%, and the N-Only Control treatment recovered the most at Motte Reserve with 76% . Motte Hoagland Control treatment had a tracer spike recovery of 53% . At both sites, the lowest tracer spike recovery occurred in the Invasive Plant treatments at 66% and 16% , for Riverside and Motte respectively. Motte Invasive Plant treatments yielded the overall least tracer recovery due to herbivory and interference by rabbits and birds. Hoagland Control and N-Only Control modules showed incomplete reduction of the initial tracer spike concentration of 28 mg N-NO3/L within the liquid portion of the ITNI module while levels in the Invasive Plant modules decreased to the detection limit . This depletion had a very clear response at Mott with nitrogen concentrations of 11.6 and 19.6 mg N-NO3/L, for Hoagland Control and N-Only Control treatments . This response was less clear at Riverside, where Hoagland Control and N-Only Control treatments resulted in 17.2 and 22.0 mg N-NO3/L . Modest depletion of the initial tracer spike was noted in the liquid reservoirs of the in the Hoagland Control treatments at Riverside and Motte . The tracer in N-Only Control treatments was also depleted from the initial tracer spike, however the treatments were also significantly lower than the Hoagland Control modules for both Riverside and Motte . In the Invasive Plant modules, AP 15N levels approached the natural abundance level of 0.366 indicating little of the original tracer remained in the liquid reservoirs of modules containing plants. Unmeasured N in the Invasive Plant treatments averaged 1.3% of the mass of nitrogen in the modules at harvest . The control at Riverside, which contained only water, had a similar unmeasured N: 1.4%. However, the Motte Control had 11% unmeasured. Riverside Nitrogen Only Controls had an average of 41.6% unmeasured N, while Motte Nitrogen Only Controls had 14.3%. Hoagland Controls at Riverside and Motte averaged 21.8% and 17.4% unmeasured N, respectively. The Module Average method of calculating nitrogen deposition for Riverside and Motte yielded deposition rates of 5.9 kg/ha and 3.5 kg/ha for the deployment period, respectively . Nitrogen deposition calculated via only plant material, resulted in a lower deposition rate , and lower still with the above ground plant matter only . However, owing to relatively large variability among replicates, there were no statistical differences among the three methods at Riverside . At Motte, N deposition was significantly different among all three computation methods : Module Average >Plant >Above ground . In contrast to Deployment 1, nitrogen recovery during Deployment 2 was generally higher and in some modules exceeded 100% . At Riverside, Invasive Plant treatments recovered between 74.6% and 190.1% of the added nitrogen. At Motte higher N recoveries relative to Deployment 1 were likely a result of the installation of a rabbit fence to reduce herbivory, resulting in recoveries between 106.8% and 253.5%. Native Plant modules had significantly lower N recovery than Invasive Plant modules at both Motte and Riverside. Recoveries in Native Plant modules ranged from 89.7% to 101.7% at Riverside and 61.3%-94.5% at Motte. Invasive Plant modules had average recoveries of 125% and 167% at Riverside and Motte respectively.