Riverside Native Plant treatments recovered on average 96% of the nitrogen tracer spike, and Motte Native Plant modules recovered 78%. In Invasive Plant modules at both sites, the above ground plant material held the majority of the total nitrogen within the plant . The Native Plant treatments at both sites exhibited less variable, smaller ratios of above ground to below ground N-biomass than the Invasive Plant treatments . Motte Native Plant modules showed a strong positive relationship with above ground biomass of nitrogen being greater than below ground biomass . Riverside Native Plants exhibited a weaker relationship between above ground and below ground plant material ; this weaker relationship might be a result of rabbit grazing of above ground plant biomass at Riverside. Though there were no obvious signs of herbivory of these plants; it is possible since the modules were exposed in an area were rabbits have been present in the past. Invasive Plant modules at both Riverside and Motte exhibited greater above ground nitrogen biomass than nitrogen mass in below ground plant biomass . Below ground nitrogen biomass ranged from <5 to 25 mg, while the above ground biomass ranged from about 150 to 450 mg nitrogen. The invasive plants at both sites had roots with much smaller overall biomass than that of the above ground plant biomass. Riverside and Motte Native Plant treatments showed relatively strong positive relations between above and below ground AP 15N . At Riverside, best indoor plant pots the California buckwheat plants had less tracer in the above ground portion of the plant than in the root system . At Motte, the California buckwheat plants had similar amounts of tracer in the above ground portion of the plant relative to the root system .

In the case of Invasive Plant modules, we observed no correlation between the tracer content of above and below ground plant parts at Riverside , while at Motte we observed an inverse relationship between above and below ground tracer concentrations At Riverside there was a clear pattern of higher N deposition measured by the Module Average method, however, it was difficult to detect statistically significant differences among the three computational methods of determining deposition rates . In the case of Riverside Native Plant modules, the Module Average approach yielded significantly higher N deposition than the Above ground only approach. The Module Average method yielded deployment nitrogen deposition rates of 8.6 kg N ha-1 and 9.8 kg N ha-1 for the Native Plant and Invasive Plant treatments, respectively. For the Module Average and Plant computation methods we noted higher apparent N deposition for Invasive Plant modules relative to the Native Plant modules, however these differences were not statistically significant. We observed significantly higher N deposition for the Invasive Plant treatment computed using the above ground biomass method relative to the Native Plant Treatment .Deployment 1 incurred two experimental errors with the treatments at Riverside, the first being the Nitrogen Only treatment. This treatment was shown to be statistically different from other treatments; however, this was due to the accidental and unknown supplement of additional 14N-KNO3 to all Nitrogen Only modules, after the deployment had begun. This error highly diluted the 15N tracer, which resulted in overestimated nitrogen deposition to the modules by mimicking an unrealistically high nitrogen deposition rate. Accidental additions were also made to some Complete Nutrient and Invasive Plant modules, but the effected modules were removed from analysis . This error was also evident in TN analyses where Nitrogen Only Controls at Riverside exhibited unmeasured N on the order of 41.6%, while Motte measured an average of 14.3% unmeasured N.

The other error during Deployment 1 was the placement of the watering ring, approximately 2 inches above the sand surface, which resulted in minor splashing of nutrient solution along the inside surfaces of the ITNI module . This resulted in some visual salt accumulation on the sand reservoir surfaces that would represent lost 15N and N mass leading to errors in the nitrogen deposition rate. Riverside and Motte experienced different biological interferences during Deployment 1. These interferences later prompted changes to the ITNI design with subsequent deployments to prevent as much interference as possible without compromising ITNI function. An internal biological interference was that of microbes in the ITNI experimental setup. This would be the case as Complete Nutrient treatments recovered less than 100% , although no plant components were available to be eaten, destroyed, or lost, in this treatment.This is because the addition of both nitrogen in a consumable form of KNO3 and a nitrogenfree Hoagland solution to the ITNI modules, would supply all other macro and micro nutrients necessary for microbial growth. We attempted to avoid this through dilute bleach rinses of modules before preparation and through baked sand additions to the modules. However, over time, and during prep, microbes could enter the ITNI systems through many pathways including dust deposition, and introduction of microbes from other species “visiting” the ITNI modules such as rabbits, birds, and insects. Upon harvest, biofilms were noted on the in-side surfaces of the modules, but could not be accurately quantified. ; leading to the belief that nitrogen was most likely immobilized by algae in the liquid reservoir. The presence of algae and microorganisms in the nutrient liquid were supported by TN analyses . Module treatments, such as Nitrogen Only Control and Hoagland Control, contained significant amounts of unmeasured N in their liquid reservoirs at harvest. However, Invasive treatments had very low levels of unmeasured N in the liquid reservoir. This suggests that by having the plant in the module, the nitrogen was taken up too quickly for microbes to fully utilize the nutrient media. Lastly, looking at the water-only Control at both Riverside and Motte, it is apparent that biological interference from more than just microorganisms was present at Motte, with the Riverside Control containing only 1.4% unmeasured N, while Motte had 11%. The source of this additional biological interference from rabbits and birds will be discussed in the following paragraphs.

Herbivory was present at both Riverside and Motte. Riverside modules experienced herbivory from what appeared to be aphids . The aphid like insects fixed to the Bromus rubens along the stems of the invasive grass. The insects were noted, in high densities during the 6 th week of exposure. This herbivory would result in the removal of both 14N and 15N through the xylem of the plant tissue. Motte, on the other hand, experienced plant tissue consumption and subsequent removal from the ITNI system, by rabbits. Modules containing invasive plants at Motte had bite marks after two weeks of exposure and were eventually grazed down to the base of the plant above the sand surface. Though they were not observed directly,blueberry container size rabbits were suspected to be the largest contributor to this loss because of identifying teeth marks on the plant surfaces and their scat and footprints discovered on the module sand surface.Rabbit interference had a significant effect on ITNI function, since all plant material above the plant surface was consumed. This could lead to two, somewhat compensating errors. First, since the plant material was more enriched with 15N than other module components, this would increase the magnitude of the quantity inside the parentheses of Equation 1.3 leading to an overestimation of N-deposition for Invasive modules at Motte . However, loss of plant biomass also reduced the value of Ns in Equation 1.3 results in an underestimate of N deposition. Based the relatively low rate of N deposition for these modules, we suspect that the N biomass effect overwhelmed the isotope effect leading to unrealistically low estimates for N deposition in the rabbit-grazed modules . The removal of above ground plant material might have contributed as much as 40% error to Motte Invasive modules, based on the proportion of above ground biomass available at Riverside. The last observed biological interference to the ITNI systems was bird visitation. Since Deployment 1 was exposed during the months of March and May, they were most likely prime targets for bird interference due to spring migration. When ITNI modules were serviced, it was noted that birds were congregating around the modules and resting on the watering ring, above the sand surface. Bird scat was noted in modules, and promptly removed to prevent it entering the ITNI system. N inputs to the modules from animal waste will increase the Ns term and decrease the a’s term in Equation 1.3 leading to overestimated N deposition.

Lastly, birds were also observed drinking from the watering ring apparatus that was installed approximately 2 inches from the surface of the sand, resulting in removal of nitrogen from the system. Since the watering ring functioned similar to a drip system, it is estimated that birds could have removed approximately 5 ml of water during each watering session. Based off a 9 L reservoir, this would contribute an approximate 11% error in the recovery of N over the course of the deployment. Deployment 2 incorporated the lessons learned from Deployment 1. This included the addition of aluminum foil on the outside of the module, the addition of a fence around the exposure area, and modified heights of the module watering rings. These improvements reduced the unintended losses and gains of nitrogen experienced in Deployment 1. Foil was wrapped around the liquid reservoir portion of the ITNI to reduce light levels in the liquid reservoirs in order to prevent algal growth. At harvest, the modules showed little or no evidence of biofilms in the liquid reservoir. This improvement may help explain the higher nitrogen recoveries in Deployment 2 relative to Deployment 1 . Lower algal or microbial N uptake also contributed to more accurate estimates of average AP 15N levels in the modules. Modules at Motte were fenced-in to prevent consumption of plants by rabbits. However, our first attempt at fencing , was not adequate to fully deter rabbits from borrowing or climbing into the modules. Wildlife cameras installed at the site during week 2 confirmed this. Changes to the fence design led to the complete exclusion of rabbits by week 4. However, during the fencing trial, 2 modules with native plants and 3 modules of invasive plants had evidence of grazing by rabbits and were therefore removed from the experiment. During Deployment 2 the watering rings were repositioned to lie completely flush with the surface of the sand in order to deter birds. Wildlife cameras confirmed that the lack of access to free-flowing water from the ITNI system eliminated the visits from birds and reduced the overall population of birds in the adjacent area. We are confident that the ITNI N deposition values for Deployment 2 are not significantly affected by rabbit grazing or bird visitation.During Deployment 2, we noted that the plant pool of nitrogen was substantially more enriched with 15N than the sand and water despite the fact that most of the N in the modules resided in the plant tissues . This is most likely attributed to exponential plant growth during the first few weeks of the deployment when there was abundant 15N in the liquid reservoir to fuel plant growth. This physiological process would promote the rapid transfer of nitrogen from the liquid reservoir and into the plant, leaving the nutrient solution and sand depleted in 15N tracer and N mass. After this initial plant growth-spurt fueled by Hoagland solution, we speculate that plant growth slowed because it was dependent on atmospheric N inputs. Interestingly, atmospheric N inputs had a larger impact on the AP 15N of the sand and liquid since these pools were small , while the plant pool of N was strongly buffered against changes in AP 15N owing to its large size. The growth and reproductive habits of the plants selected in this study might help explain why we observed lower N deposition rates when using only the plant in the ITNI computations or when we used only the above ground plant biomass. The majority of nitrogen in the Invasive Plant modules was in the above ground plant material . This was most likely due to the annual life history of the invasive plants, leading them to maximize above ground, seed producing organs, as opposed to putting growth into below ground roots.