«A Thesis Presented to the Faculty of the Graduate School of Cornell University In Partial Fulfillment of the Requirements for the Degree of Master of ...»
Hotspot: The hotspot occurred at the low topographic position where soil water content was approximately 60% VWC at subplot BLT (fig. 3C). Mean daily maximum temperature of the soil at subplot BLT was the highest observed (fig. 4) and the chamber where the spike occurred contained the least above-ground biomass of all 24 chambers (fig. 4, 5). These conditions are similar to those reported in the literature for soils that produce spring-thaw pulses of N2O. Corre, vanKessel (12) found that fluxes were higher at foot-slope positions in the landscape compared to shoulders and that differences in flux between cultivated and uncultivated land were more pronounced at the foot-slope where anoxic conditions are more prevalent. This is consistent with the position of the hotspot at BLT-2 observed during this study. It is has been observed that soil temperature is a driving factor in N2O emissions (26). In this case, the higher temperatures observed at the hotspot confirm this relationship. In addition, the low above-ground biomass at this chamber echoes the findings of Dorsch, Palojarvi (15), Dietzel, Wolfe (23), and
Hot-moment: The single greatest flux, 77.6 µg N2O-N m-2 hr-1 was observed at chamber BLT-2 on April 4th. The magnitude of this flux can be compared to the results of Singurindy, Molodovskaya (17) who observed a maximum flux of 200 µg N2O-N m-2 hr-1 during a thaw event from manure amended corn (Zea mays L.), and Molodovskaya, Oberg (29) who observed a maximum of 114 µg N2O-N m-2 hr-1 from manure amended alfalfa (Medicago sativa L.). In our study, the maximum flux observed meets the criteria of a hot-moment, exceeding the outlier threshold which we defined as a flux observation greater than the median plus 3 times the interquartile range (12). The hot-moment threshold in this case was calculated to be 21.8 µg N2O-N m-2 hr-1. An elevated flux (21.3 µg N2O-N m-2 hr-1) nearly qualifying as a hot-moment itself, was observed at the same chamber one day earlier, suggesting sustained biological activity.
The abrupt rise in N2O emission at the hotspot occurred at the tail end of a multi-day cooling period characterized by air temperatures that only briefly broke 0°C during the day, and settled as low as -7.2°C at night (fig. 3A, 3B). This sequence of events is similar to the findings of Wagner-Riddle, Furon (16) and Dietzel, Wolfe (23), and can be explained by reports that the magnitude of thaw induced N2O fluxes are related to freezing intensity and duration (20) and triggered by a rapid increase in temperature (11). It is known that intensity and duration of soil freezing is related to the amount of insulating plant cover (21) and we found a similar relationship between above-ground biomass and soil temperature dynamics (fig. 5). Minimal above-ground biomass in chamber BLT-2 suggests that the soil was most susceptible to freezing and hence increased nutrient availability. We suggest that the sudden presence of newly-liberated
temperatures above 5°C at BLT-2 (fig. 3B) created the conditions for increased activity of denitrifying microbes and the genesis of significant amounts of N2O on April 4th.
It is notable in our study that this observed pulse was not at the initial thaw, and that heavily insulated soils that thawed for the first time did not exhibit a pulse of N2O. This suggests that the abruptly released N2O constituting the hot-moment was newly generated, and that trapping of N2O in the subsoil over a long period does not appear to be a relevant mechanism at our site. It is possible that the hot-moment was the result of either de novo denitrification at the time of phase change from solid to liquid water in the soil surface (0-10cm) where most N2O production is thought to take place (21, 26), or due to generation of N2O in anoxic water films during the freezing period that was subsequently released upon thaw, or a combination of these two mechanisms.
Statistical results: The Kruskal-Wallace rank-sum test provides evidence that there are differences among treatments (p = 0.0002). Applying Wilcoxon’s rank-sum test to each topographic position – crop type combination against all others shows that Low - RCG is the most different (p = 0.0001). Wilcoxon’s rank-sum test, when applied as Low - RCG against High
- RCG, High - FGL, and Low - FGL individually, yields p-values of 0.0002, 0.0367, 0.0001 respectively, which were also the most significant of all (6) possible topographic position - crop type comparisons. The same pattern was observed with the application of Wilcoxon’s signedrank test using paired daily subplot flux averages to compare Low - RCG against High - RCG, High - FGL and Low - FGL with p-values of 0.0004, 0.0295, and 0.0001 respectively.
Removing the outlier from the dataset tended to slightly increase the p-values of the nonparametric statistical tests, but not significantly and did not alter the outcome of any of the non
normally distributed (Shapiro-Wilks test p-value = 0.07) with only a slight positive skew (skewness = 0.28) and are suitable for parametric operations. Temporal autocorrelation was observed to be not significantly different from zero and, with no need for repeated measures analysis, a least-squares linear regression model was constructed from the data. We found that the interaction model (p = 0.00004) represents the data better than topographic position or crop type alone, but explains only 6.5% of the variation in flux, though this low value might be mainly due to the relatively low magnitude of emissions and a low signal to noise ratio (the greatest modeled flux was about twice the detection limit, there were 18 detects disregarding the single removed outlier). The Residual Standard Error of the model was 1.34 * 10-7 g N2O m-2 min-1 and this value closely matches the standard deviation of simulated fluxes (1.4 * 10-7 g N2O m-2 min-1) from the Monte Carlo detection limit estimation. A Least Squared Difference (LSD) test of the fluxes grouped by topographic position – crop type shows that the Low - RCG group has the highest flux and is significantly different from the other 3 groups. The average flux from the Low - RCG subplots (outlier removed) is 4.4 µg N2O-N m-2 hr-1, almost six times higher than for the Low - FGL. 88% of the total emissions were not classified as outliers and were included in the model, this shows that the difference between treatments is important even at low flux levels.
Both parametric and non-parametric statistical analyses confirm that the Low - RCG subplots exhibit significantly elevated N2O fluxes compared to the other groups.
Overall, the statistical and modeling results and the hot-moment analysis indicate that elevated flux levels occurred in the Low topographic positions that were converted to RCG. As discussed in the hot-moment and hotspot analysis sections, this is likely due to the combination
other treatments, and these factors created conditions that are conducive to denitrification and subsequent N2O production: Presence of organic carbon and nitrate, daytime temperatures above 5°C, and anaerobic soil conditions (see introduction). It is unlikely that a difference in nutrient availability was directly caused by residue from the previous year’s fertilizer application because significant time had elapsed allowing for uptake by vegetation during the growing season, as well as by the microbial community. Additionally the species of vegetation in the FGL was largely reed canarygrass (which pre-existed at the field before conversion) and it is therefore unlikely that resident vegetation caused an increase in nutrient availability. It is possible that tillage and incorporation of organic matter during crop establishment approximately 21 months prior resulted in increased nutrient availability (SOM for example) during the study or affected the soil structure or microbial community in some other way, but we believe that the most plausible explanation is that reduced insulating capacity of the sparse grass in the partially established RCG plots (fig. 5) caused more intense and frequent freeze-thaw cycling of these soils and this created the conditions that allowed denitrification to proceed.
We found that N2O emissions over the course of this 12-day study were greatest and most frequent from the Low – RCG subplots and were comparable in magnitude, but generally lower than fluxes observed from manure amended corn and alfalfa. In the perennial grasses studied here, the Low - RCG areas sustain conditions of elevated soil moisture compared to High topographic positions and have reduced insulating plant cover (above-ground biomass) compared to the FGL, both factors that have previously been shown to be important to N2O production and could be useful for predicting emissions at the field scale (objective iii). Our results suggest that
by conversion to reed canarygrass, but it is not clear that this is the only critical element of the conversion process affecting N2O emissions.
Ultimately, year-round observation is necessary to determine total annual N2O losses from each treatment, but our results suggest that conversion of fallow grassland to perennial grass cropping systems for bioenergy or other uses could increase spring-thaw N2O emissions in wetness prone areas (objective ii).
We found that in perennial grass cropping systems during spring thaw, nitrous oxide emissions are related to landscape features and environmental conditions. This is not surprising and has been found to be the case in other studies, but here we confirm that the primary mechanisms and controlling factors in other situations during spring thaw seem to also be relevant in perennial grasses. The results of this study strengthen the previous research that it confirms. It also provides some information about the potential impact of using fallow land for bioenergy production and suggests ways that nitrous oxide emissions could be mitigated in a land conversion scenario.
There are still aspects of nitrous oxide production in soils at spring thaw that are not understood. Specifically, these seem to be dynamics of microbial populations under freeze-thaw conditions, the exact physical and biological mechanisms of nutrient liberation during thaw events, and the effect of variation within the plant-soil continuum and its atmospheric interface on the generation and transmission on nitrous oxide.
It was noted during the observation period that the chamber collars that were installed prior to the study and left in place throughout seemed to have an effect on the conditions within.
Most notably, the collars seemed to shade the enclosed area and caused the temperatures to be cooler than the surrounding areas which were more exposed to sunlight and surface air currents.
These effects could affect the outcome of an experiment if the actual conditions within the chamber are not accounted for. It would be a worthwhile advance to adopt a less intrusive chamber design in this regard.
It was also noted during the study that significant heterogeneity existed in the soil and plant arrangement, both within subplots and within single chambers themselves. It is possible
of the soil, a microsite, enclosed by the chamber. Thus, characterizing an emission observation based on the general characterization of the chamber location or conditions may overlook the actual conditions of nitrous oxide genesis. A study on the effect of the degree of surface soil topographic variability on emissions may provide insight on this front.
We saw that when emissions occurred they tended to be sudden, sporadic, and short-lived and this has been the trend in other studies. This suggests that the microbial action that results in nitrous oxide production may change on a temporal scale of hours or even minutes. The change in phase of water between liquid and solid also takes place at short time scales, and we suggest that a higher frequency of observation is appropriate to further investigate the dynamics of the processes and factors involved.
To summarize, additional research should look in finer detail at temporal fluctuations in soil temperature, and asess the effect of small-scale heterogeneity of soil conditions with respect to proximity to plants, micro-topography, and atmospheric exposure. Microbial community analysis may also prove useful in elucidating the mechanisms at play in nitrous oxide formation, and if chambers are used in the field, an improved design or temperature and moisture monitoring scheme employed.
into four stages: Preparation, sample collection, storage, and analysis.
1. Preparation Vial preparation: The vials are first left uncapped in the open air for at least 30 minutes to allow any previous contents to diffuse. The vials are then fitted with an un-used butyl rubber stopper and capped with an aluminum ring crimped by the “Crimpenstein” device, or the hand held crimper (lab crimper #2). The “Crimpenstein” crimping pressure is set to one less that the middle setting (one green and one yellow light), the hand held crimper is set to the tightest setting. After capping is complete the vials are randomized before evacuation.
The vials are evacuated the morning of the sampling campaign. Each vial is evacuated for 30 seconds at -90 kPa using a large (21 gage) needle, changed every 60 vials. Evacuate all vials in a single session and use a rotation technique, employing the white shut off valves to start and stop the evacuation of each vial. When removing an evacuated vial, pull smoothly straight off the needle at a steady pace, not too fast or too slow. The rate at which a vial is pulled from the needle impacts the leakage from the surrounding air back into the vial as it is removed, greatly affecting the quality of the vacuum. Randomize all vials by gently mixing them in a cardboard box or plastic bucket prior to evacuation.
The evacuated vials are arranged in trays and pre-labeled with the chamber and time of sample injection. Spare vials and calibration vials were bagged, stored and transported along with the sample vials.