|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 LONG TERM TRENDS OF CO2 EXCHANGE IN A TUNDRA ECOSYSTEM AFFECTED BY PERMAFROST THAW AND THERMOKARST FORMATION By CHRISTIAN TRUCCO A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DE GREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2011
2 2011 Christian Trucco
3 To the well being of the planet, Be conscientious about your decisions and actions, as each of them make a great impact in our precious environment
4 ACKNOWLEDGMENTS I would like to thank Drs. Edward Schuur, Timothy Martin, and Michelle Mack for their excellent guidance, mentoring, and the example they have set as researchers in t his important field of science. They gave me the opportunity to focus my academic pursuits on a field of scientific endeavors that I find fascinating and incredibly valuable. I would also like to thank Drs. Jason Vogel, Susan Natali, and the members of the Schuur lab for their support and help on the research project. Last but not least, I thank my family, and my beautiful novia Dr. Leda Nikola Kobziar for giving me encouragement, and support throughout the process.
5 TABLE OF CONTENTS page ACKNO WLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 8 LIST OF ABBREVIATIONS ............................................................................................. 9 ABSTRACT ................................................................................................................... 10 CHAPTER 1 INTRODUCTION .................................................................................................... 12 2 METHODS .............................................................................................................. 1 7 Site Description ....................................................................................................... 17 Net Ecosystem Carbon Exchange (NEE) ............................................................... 18 Static Carbon Flux Measurements .......................................................................... 19 Automatic Carbon Flux Measurements ................................................................... 20 Environmental Variables ......................................................................................... 20 Data Analysis .......................................................................................................... 22 Statistical Analysis .................................................................................................. 24 3 RESULTS ............................................................................................................... 26 Environmental Variables ......................................................................................... 26 Distance to Water Table (DWT) .............................................................................. 26 Active Layer Depth (ALD) ....................................................................................... 27 Ecosyste m Carbon Fluxes ...................................................................................... 28 Growing Season Net Ecosystem Exchange (NEE) .......................................... 28 Growing Season Gross Primary Productivity (GPP) ......................................... 30 Growing Season Ecosystem Respiration (Reco) ................................................ 31 Winter Respiration ............................................................................................ 32 Annual Net Ecosystem Exchange .................................................................... 33 Annual Ecosystem Respiration ......................................................................... 33 Above Ground Net Primary Productivity (ANPP) .............................................. 34 Long Term Carbon Exchange .......................................................................... 35 Carbon Response to Abiotic Factors ...................................................................... 36 4 DISCUSSION ......................................................................................................... 54 5 CONCLUSION ........................................................................................................ 61
6 LIST OF REFERENCES ............................................................................................... 62 BIOGRAPHICAL SKETCH ............................................................................................ 67
7 LIST OF TABLES Table page 3 1 Average photosynthetic active radiation (PAR), precipitation, air temperature, soil temperature and cumulative PAR from the weather station at Eight Mile Lake .................................................................................................................... 49 3 2 Aboveground net primary productivity (ANPP), average growing season thaw depth (Avg GS TD), active layer depth (ALD), and av erage growing season depth to water table (DWT) for three sites varying in degree of permafrost thaw .................................................................................................................... 50 3 3 Seasonal and annual carbon fluxes for three sites varying in degree of permafrost thaw .................................................................................................. 51 3 4 Mixed model parameter estimates with associated standard errors and P values for GPP, Reco, and NEE.. ......................................................................... 52 3 5 Results for the repeated measure ANOVA full model, and the ANCOVA full model and reduced model .................................................................................. 53
8 LIST OF FIGURES Figure page 3 1 Seasonal thaw for each site and water table depth averaged across the three sites over a period of seven years. ..................................................................... 39 3 2 G rowing season average depth to water table by sit e.. ...................................... 40 3 3 Active layer depth (ALD), and average growing season thaw depth (Avg GS TD) for three sites varying in degree of permafrost thaw.. .................................. 41 3 4 Growing season carbon exchange over seven consecutive years for three sites varying in degree of permafros t thaw. ........................................................ 42 3 5 Annual carbon exchange over six consecutive years for three sites varying in degree of permafrost thaw. ................................................................................ 43 3 6 Aboveground net primary productivity for three sites varying in degree of permafrost thawing.. ........................................................................................... 44 3 7 Long term carbon exchange average for three different sites varying in degrees of permafrost thaw.. .............................................................................. 45 3 8 Relationship between active layer depth (ALD) and growing season (May September) carbon exchange.. .......................................................................... 46 3 9 Relationship between g rowing season depth to water table (DWT) and growing season carbon exchange. ..................................................................... 47 3 10 Relationship between active layer depth (ALD) and growing season above ground net primary production (ANPP).. ............................................................. 48 3 11 Relationship between depth to water table (DWT) and aboveground net primary production (ANPP).. ............................................................................... 48
9 LIST OF ABBREVIATION S ALD active layer depth ALT annual long term ANPP above ground net primary production DWT distance to water table F cmax maximum photosynthesis GPP gross primary production GS growing season LTGS long term growing season NEE net ecosystem exchange PAR photosynthetic active radiation Ra autotrophic respiration Rd dark respiration Reco ecosystem respiration Rh heterotrophic respiration Ta air temperature Ts soil temperature
10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Par tial Fulfillment of the R equirements for the Degree of Master of Science LONG TERM TRENDS OF CO2 EXCHANGE IN A TUNDRA ECOSYSTEM AFFECTED BY PERMAFROST THAW AND THERMOKARST FORMATION By Christian Trucco May 2011 Chair: Edward A. G. Schuur Major: Botany Warming of the arctic has le d to the degradation of per mafrost ice reach organic soils, creating depressions on the soil surface called Thermokarst. The thawing of frozen soil organic mat ter due to thermokarst development can increase CO2 emissions to the atmosphere. Three areas of varying levels of thermokarst (Extensive Thaw Moderate Thaw and Minimal Thaw ) were studied on an upland tussock tundra ecosystem in Central Alaska. The purpose of thi s study was to detect inter annual differences in ecosystem r espiration (Reco), gross primary productivity (GPP), and net ecosystem exchange (NEE) between these three thermokarst levels over a period of seven years, and to learn how these differences, if a ny, could be determined using environmental variables such as photosynthetic active radiation (PAR), air temperature, precipitation, water table levels, and active layer depth. Ecosystem carbon dynamics (Reco and NEE) were measured using static and automatic chamber techniques, and GPP was consequently calculated. R esults showed an increasing growing season (May September) trend in GPP, NEE, ANPP, ALD and annual NEE at all three sites over the study period, but no change in annual and growing season Reco at the three sites. The long term average showed that Extensive and M oderate Thaw had significantly
11 greater growing season GPP and Reco and annual Reco than at M inimal Thaw In addition, the seven year growing season GPP at the three sites exceeded Reco, making these sites a carbon sink during the growing season. Interannual differences in growing season GPP, Reco and NEE are mainly controlled by active layer depth, water table depth of the soil, which varies with levels of thermokarst s.
12 CHAPTER 1 INTRODUCTION Permafrost regions, which comprise about 22% of the northern hemisphere land surface, are an important component of the global carbon cycle because they store an estimated 50% of the belowground global carbon pool (Zhang et al. 1999, Schuur et al. 2008) About 1672 Pg of carbon, roughly equivalent to twice the amount of carbon in the atmospheric pool, is currently sequestered in permafrost in the form of organic soil carbon (Post et al. 1982, Zimov et al. 1996, IPCC 2007, Schuur et al. 2008, Tarnocai et al. 2009) Increasing air temperature during the past century has caused permafrost in regions of the arctic and subarctic to warm (Burn & Smith 1993, Osterkamp et al. 1994, Osterkamp & Romanovsky 1999, Osterkamp 2007) and become susceptible to thawing, since small changes in heat energy input can have large impacts on the energy balance of permafrost. This susceptibility is especially pronounced in the subarctic region. Because subarctic permafrost temperatures are near thawing, with average temperatures of 0oC in central Alaska, a small change in air surface temperature could cau se irreversible degradation (Osterkamp & Romanovsky 1999, Jorgenson & Osterkamp 2005, Lawrence & Slater 2005) In addition, soil temperature in central Alaska can increase by approximately 3oC when snow cover is excessive (Natali et al. 2011) Permafrost thawing in the sub arctic is likely, as climate change predictions inclu de air temperature increases of up to 78C along with increased snow fall (ACIA 2004, IPCC 2007) Changes in arctic and subarctic permafrost have already been evidenced in landscape disturbances such as thermokarsts (Osterkamp & Romanovsky 1999, Jorgenson & Osterkamp 2005, Schuur et al. 2007) Thermokarsts are the result of
13 thawed icerich permafrost s oils, which leave the upper soil layers without support, causing a collapse and leaving a subsidenc e in the soil surface. This is exemplified in arctic and subarctic tundra ecosystems, where changes in microtopography due to subsidence cause changes in soil properties and hydrological cycles (McNamara et al. 1998, Jorgenson et al. 2001, Yoshikawa & Hinzman 2003, Schuur et al. 2007) In the summer months, and during snow melt, water drains into thermokarsts, creating several hydrolog ical and soil thermal scenarios depending on the overall topography of the landscape (e.g., thermokarst ponds or lakes, or thermokarst drainages or streams among others) (Jorgenson & Osterkamp 2005) At the microsite level, thermokarsts drain the adjacent soil layers, creating dry microsites, with shifts from moss layers to lichen dominated areas and from high to low soil moisture (Schuur et al. 2007) During the winter months, thermokarsts accumulate more snow, keeping the soil warmer than the air temperature above (Walker et al. 1999, Hinkel & Hurd 2006, Nowinski et al. 2010, Natali et al. 2011) The snows insulating effect acts as a positive feedback to ex acerbate the further degradation of permafrost. Since the soils heat energy is higher under deep snow cover, the energy required to warm and thaw the soil in the summer is greatly reduced (Seppala 1994, Hinkel & Hurd 2006) The altered thermal dynamics of the system accelerate the thickening of the active layer depth (ALD, the upper surface of the soil that thaws each summer), consequently increasing the water storing capacity of the soil and soil moisture (Hinkel et al. 2001, Jorgenson et al. 2001, Zhang 2005, Akerman & Johansson 2008) These physical changes have important biological consequences, as well as direct and indirect feedbacks to climate change. The thawing of century or millennia old
14 organic carbon, coupled with increased soil moisture, enhances microbial metabolism and nutrient availability, and hence, increases ecosystem respiration (Hobbie et al. 2002, Mack et al. 2004, Dutta et al. 2006) The increase in heterotrophic respiration can be offset by uptake of CO2 by plants in the growing season. But continued microbial activity during the winter months, however limited, occurs over a time span that may result in the net release of CO2 and/or CH4 carbon to the atmospheric pool, creating a positive feedback to climate warming (Vogel et al. 2009) others). Dutta et al. (2006) demonstrated, through laboratory soil incubations, that permafros t in northeastern Siberia has the potential to release 20 to 24 ka old carbon, and that if 10% of the soils thaw, at a soil temperature of 5oC, a release of 1Pg of labile carbon would be expected to be released in the first year followed by 40Pg of soil carbon over the next 40 years. Other studies in northern and central Alaska, have shown, through in situ measurements, the release of around 9 ka old CO2 carbon from thawing permafrost, demonstrating that a warmer climate has the potential to unlock ancient organic carbon and enhance soil organic carbon decomposition (Schuur et a l. 2009, Nowinski et al. 2010) Increased surface permafrost temperature and old organic carbon availabi lity, along with changes in soil moisture and nutrient availability, can shift the net ecosystem carbon balance to a new point of carbon sink or source, depending on permafrost degradation levels (Mack et al. 2004, Harden 2008, Schuur et al. 2008, Vogel et al. 2009) In addition to changes in the thermal regime, hydrology, microbial activity and organic matter availability in t undra ecosystems, thermokarst formations have also changed the plant community distribution and their relative abundance. Affected wetter
15 areas have shifted from graminoid dominance to increased woody plant dominance with greater abundance of hydrophilic mosses (Schuur et al. 2007) This shift from graminoids to deciduous woody plants is in part attributed to increased soil moisture and the release of nutrients as newly unfrozen organic matter decomposes (Chapin & Shaver 1996, Jonasson et al. 1999, Shaver et al. 2001, Schuur et al. 2007) Shrub expansion in the arctic has been associated with thermokarst formation, where ALD, nutrient and water availability i ncrease with thermokarst formation (Lloyd et al. 2003, Lantz et al. 2009) The more productive shrub species accumulate more carbon in long lived woody biomass in part due to high photosynthetic biomass and higher gross primary productivity (GPP) (Schuur et al. 2007, Vogel et al. 2009) Our objective was to understand the i mplications of a warmer climate on the carbon cycle of permafrost regions, and the potential feedbacks of a disturbed permafrost ecosystem to climate change. In order to achieve our objective, we compared carbon dynamics among three sites along a naturally degrading permafrost gradient in an upland tussock tundra ecosystem. Our study site has experienced various degrees of permafrost degradation in the last 30 years, altering the overall soil surface patterns and hydrology of the landscape (Osterkamp & Romanovsky 1999, Osterkamp 2007, Schuur et al. 2007) Specifically, we investigated whether there are any differences in ecosystem respiration, gross pri mary productivity (GPP) and net ecosystem exchange (NEE) among three levels of increasing thermokarst formation. We also determined how these differences could be modeled using environmental variables such as photosynthetic active radiation (PAR), air and soil temperature, water table levels, and thickness of the active layer. We quantified soil respiration over three
16 winters, which, along with the growing season measurements, provide a more accurate estimate of annual carbon exchange (Vo gel et al. 2009) Other studies have predicted carbon dynamics for this type of ecosystem based on one or two growing seasons (Vogel et al. 2009) but our study captures carbon exchange measurements for seven consecutiv e growing seasons (20042010). Given the int era nnual variability in environmental factors s uch as rainfall, snowfall and distribution, and soil and air temperatures, the combination of growing season and winter carbon exchange is key to deriving accurate predictive models Therefore, a more representative carbon exchange model can be derived from this work, allowing for the prediction of longer term impacts of thermokarst formation on the contribution of the tundra to the global carbon cycle.
17 CHAPTE R 2 METHODS Site Description The study was conducted within the Eight Mile Lake (EML) watershed, west of Healy, Alaska, USA ( 63o52 42.1 N, 149o15 12.W 700 m.a.s.l. (Schuur et al. 2007, Schuur et al. 2009, Vogel et al. 2009, L ee et al. 2010) This is within the discontinuous permafrost zone, but permafrost is widespread at the study site. The 30year annual mean air temperature is 1oC with a summer mean air temperature of 11.2oC. Annual mean precipitation is 378 mm with growing season mean precipitation of 245 mm. The watershed is defined by a valley carved by glaciers dating from the early Pleistocene (Wahrhaftig 1958) The study site lies on a gentle northfacing (<5o) slope, where three sites were selected which had different degrees of permafrost degradati on and thermokarst. The first sit e (hereafter called Minimal Thaw ) had the least permafrost degradation, with no defined water tracks or thermokarst subsidence (Lee et al. 2010) The vegetation was mainly dominated by the tussock forming sedge Eriophorum vaginatum along with dwarf shrubs and a moss understory, typical of moist acidic tundra (CAVM 2004, Schuur et al. 2007) The second site (herea fter called Moderate Thaw ) had intermediate permafrost degradation with shallow thermokarst subsidence and a deeper active layer (seasonally thawed ground). This site was co located with a permafrost bore hole that has monitored ground temperature since 1985 and has doc umented permafrost thaw over the past 20 years that initiated rapidly in 1989 (Osterkamp & Romanovsky 1999, Osterkamp et al. 2009) Here, there is a shift in vegetation composi tion as compared to Minimal Thaw with shrubs and mosses being more abund ant than at Minimal Thaw (Schuur et al. 2007) Finally, the third site on the
18 gradient (hereafter called Extensive Thaw ) had a high degree of permafrost degradation, with deeper thermokarst subsidence than Moderate Thaw and the deepest active layer of all the sites. Thawing at this site is thought to have initiated several decades prior to Moderate Thaw (Osterkamp & Romanovsky 1999) and the plant community was now mostly dominated by shrubs and hydrophilic mosses, with a significant decrease in abundance of E. vaginatum tussocks. Also, the redistribution of surface water created both saturated areas where the water table perched on the permafrost surface was at or near the soil surface, as well as dry patches with dead mosses and less vascular plant biomass on adjacent patches that had not subsided. Net Ecosystem Carbon Exchange (NEE) Net ecosystem carbon exchange ( NEE) is the net gain or loss of carbon by an ecosystem in a time interval. NEE is the balance of two large opposing processes: carbon uptake by primary producers (gross primary production, GPP ) and respiration losses (autotrophic Ra, and heterotrophic Rh): = ( ) ( 2 1) where ( GPP Ra) is the net primary production ( NPP ). We used the convention that positive values of NEE indicate net carbon uptake by the tundra ecosystem. A combination of automated and static chamber carbon flux measurements were carried out during the seven growing seasons. For 2004 through 2008, carb on fluxes were measured with a combination of the two methods, as described in Vogel et al. (2009). For the growing seasons of 2009 and 2010, only static chamber measurements were conducted.
19 Static Carbon Flux Measurements We measured ecosystem carbon flu x in six plots within each of the three sites using a clear Plexiglas chamber (70 x 70 x 40 cm) connected to an infrared gas analyzer (LI 820 Licor Inc., Lincoln, NB). The chamber was securely placed on top of a plastic base embedded permanently ~ 10cm into the moss/soil surface, but not sealed airtight, in order to minimize pressure differential that can affect CO2 flux inside the chamber (Lund et al. 1999) In order to make a good seal between the chamber and the base, a soft insulating pipe was attached to the base of the chamber Two small fans kept the air well mixed inside the chamber, and prevented high temperatures from altering photosynthesis of plants, and respiration of plants and soils. Air temperature inside the chamber was measured using a type T ther mocouple. Photosynthetic active radiation (PAR) inside the chamber was measured using a PAR sensor ( Li Cor 190SA, Li Cor Inc., Lincoln,NB ) located at the north side of the chamber to prevent shading to the plants. Ecosystem respiration (Reco) was first measured for 1.5 minutes, by covering the chamber with a reflective material that prevented light from get ting inside the chamber. The cover was then removed and NEE was measured for another 1.5 minutes. These two steps were done sequentially w ithout removing the chamber from the base. For the 2009 and 2010 growing season, an additional step was added to the CO2 flux measurements. The first two steps described above were performed as usual, but then the chamber was removed from the base for roug hly 45 seconds to flush the chamber. The chamber was then placed back onto the base, and a screen mesh was placed over the top and sides to reduce the amount of irradiance penetrating the chamber. This additional measurement was done to obtain measurements of ecosystem CO2 exchange over a wider range of PAR values in order to obtain a better
20 representation of a light response curve for each individual chamber flux base. Weekly measurements were taken concurrently (within four hours) at the three sites, and roughly starting soon after snow melt (~36 May), and continued until the third or fourth week of September. Automatic Carbon Flux Measurements Automated carbon flux measurements were done using an automated CO2 closed loop system similar to the static ch amber measurements done above. Six acrylic chambers were placed on the chamber bases within each site. These chambers differed from the static chamber, since the top of the chambers had articulated doors controlled by an automated pneumatic system. This automated system, which included a datalogger (CR10X Campbell Scientific) and an infrared gas analyzer (LI 820 LICOR Inc, Lincoln, NB ), was connected to the six chambers at a site and measured CO2 flux over a 2 minute interval, while the top doors were close d, sequentially for each individual chamber. Each chamber was equipped with a temperature sensor to monitor the temperature inside of the chamber, while two small computer fans mixed the air inside the chamber. The automated chamber system was moved among the three sites each 710 days. While the automated system was taking measurements at any one site, manual measurements were done at the other two sites using the static carbon flux method described above. The first three years (2004 2006) of data was reported in Vogel et al (2009) and we incorporated in this analysis four more years until 2010, for a total of seven years of carbon exchange measurements. Environmental Variables An On Set HOBO weather station was placed at the study site and measured phot osynthetic active radiation (PAR), air temperature, relative humidity, precipitation,
21 atmospheric pressure, wind speed and direction. The station was maintained year round for the duration of the study, and the data stream was gap filled for short periods when necessary using Remote Automated Weather Station (RAWS, Western Regional Climate Center, Stampede station) located in Denali National Park. Soil temperature was measured (10, 20, 30, and 40 cm depths, n = 3), using type T thermocouples controlled by a CR10X datalogger (Campbell Sci Inc., Logan Utah)at each of the sites from May 2004 until September of 2007. From this point on, the sensor depths were reconfigured at Moderate and Extensive Thaw sites to capture the temperature at maximum thaw depth, keeping the 10cm positions unchanged, and the two middle sensors equally spaced from its consecutive sensor. Minimal Thaw soil sensors were not reconfigured; keeping the original sensor depth distribution. Only the 10cm positions were used for comparisons for the seven year period. Thaw depth throughout the growing season was measured by pushing a thin metal rod through the unfrozen soil until it hit the ice surface. The length of the embedded portion of the rod was then measured with a ruler. Three measurements were taken around each base that was used for carbon flux measurements, and then averaged to get a value for an individual chamber flux base. Then the average value for each chamber flux base was furthered av eraged to obtain a site thaw depth average. Thaw depth measurements were taken once a week from the beginning of May through the end of September. Maximum seasonal thaw depth (defined as the active layer depth; ALD) was determined by averaging the last two weeks of measurements in September. Water table was monitored at each site by inserting three 6 inch diameter PVC pipes into the permafrost. The distance from the top of the pipe to the surface of the
22 once a week to derive the position of the water table in the soil throughout the growing season. Data Analysis We calculated carbon flux by regressing CO2 concentration inside the chamber against changes in time for the sampled time interval, and used the slope of the line as the CO2 flux rate (Vogel et al. 2009). Data were filtered to eliminate spurious values resulting from erratic fluctuations in CO2 concentrations, overheating inside the chamber (Temperature > 30C), wind speed > 5 m s1 or during raining conditions. Gaps in the flux data were filled as follows: First, if PAR was > 5 mol m2 s1 we assumed daytime conditions and NEE was gap filled using parameters obtained by fitting measured NEE at each chamber to PAR through a nonrectangular hyperbola (SAS proc nlin procedure) : NEE = ( ) ( ( ) ) ( 2 2) where is the initial linear slope of the light response curve (quantum yield, mol CO2 m2 s1), Fcmax is maximum pho tosynthesis ( mol CO2 m2 s1) at light saturation, and Rd is dark ecosystem respiration (e.g., NEE at PAR = 0). Parameter estimates for eq uation 2 2 were obtained for each individual plot at all three sites. Second, if PAR was < 5 mol m2 s1 dark conditions were assumed and gaps were filled using parameters relating NEE at night plus dark chamber measurements (Reco) from each chamber to air temperature (Ta) using an exponential relationship (SAS proc non linear procedure): Reco = ( 2 3)
23 May and September NEE parameters for eq uation 22 could not be determined (with the exception of 2008 and May of 2007), therefore interpolated values of NEE were estimated by averaging measured NEE values and adjusting for duration of day and night. Gross primary productivity (GPP) was calculated as the difference between NEE and Rec o: GPP = NEE Reco ( 2 4) Monthly, growing season and annual averaged carbon fluxes were obtained for each site using measured and gap filled values Winter carbon fluxes were estimated as in Vogel et al (2009), using the integrated temperature of the soil profile to model winter carbon flux in the form: ( < 2 ) = 0 23 ( ) ( 2 5) where Ts is the integrated soil temperature averaged across 10, 20, 30 and 40 cm. The model was used for integrated Ts ranging from 2oC and colder. Vogel et al. could not find any relationship to soil temperature and measured fluxes when the soil temperature was between 2oC and 0.5oC For integrated soil temperatures above 2oC, the first three winters average measured flux for each of the sites was used. The parameters for the exponential equation did not differ among the three sites; therefore a single model w as developed for all sites. We estimated above net primary productivity (ANPP) as an independent way to estimate ecosystem carbon uptake and to follow the effects of permafrost disturbance on plant productivity. We estimated ANPP by measuring plant biomass at peak growing season using point intercept methods (Shaver et al. 2001) that were specific to the study site (Schuur et al. 2007) Moss ANPP was estimated using a crank wire method
2 4 (C lymo 1970) where the vertical growth for the growing season was associated with moss biomass obtained from a destructive harvest, and then applied to the area covered by mosses from each flux chamber base (Schuur et al. 2007) The methods were established on the first year of the study, and thereafter, we measured vertical moss growth at the end of the growing season, and conducted point intercepts at peak biomass on each chamber base, with the exception of 2005 and 2010. Statistical Analysis All environmental and carbon flux variables were analyzed using repeated measures analyses of variance (ANOVA, SAS mixed procedure) with the three sites as treatments (fixed effect), and years as the repeated factor (fixed effect). The covariate structure used was auto regressive (1) and multiple comparisons were obtained by the least square mean (Tukeys method) for treatment and year for all pair wise comparisons unless otherwise stated. We further explored the use of analysis of covariance (ANCOVA, SAS mix ed procedure) to detect directional trends through time for all carbon flux variables, as well as ALD, Average GS_TD, DWT, and ANPP. We first tested for interaction effects (full model), and then used the results of the reduced model (without the interacti on term) to make inferences about the data. We explored the relationships among C flux (GPP, Reco, NEE) and abiotic factors using mixed effect multiple linear regression to detect which abi otic factors were important in explaining the variation in carbon flux We incorporated a random effect o f year because we knew (from the repeated measures analysis) that C flux varied among years. All potential explanatory variables were selected a priori and centered for ease of interpretation (Schielzeth & Holger 2010) We gap fill ed one year (2004) of missing data for DWT using its sevenyear mean. All models were fit using the nmle package (Pinheiro et al.
25 2010) using restri cted maximum likelihood in R (R Core Development Team 2010). Furthermore, we performed individual analysi s using single linear regression on the response of carbon flux to abiotic factors. All data were tested for normality using Kolmogorov Smirnov test, a nd by assessing normal probability plots. Data that did not show normal distributions w ere transformed and then retested to ensure normal distribution.
26 CHAPTER 3 RESULTS Environmental Variables Overall there was a twofold range in growing season precipitation across the time p eriod of this study (Table 31). Precipitation in 4 years out of the 7year dataset was below the long term 30year average, whereas 2 years were above average and the remaining year was near the average. Average growing season air temperature varied between 8.1oC and 11.5oC while cumulative PAR for the growing season also varied ~10% among years. Lower cumulative PAR did not reflect higher precipitation for some of the years; for instance, 2010 received about half of the preci pitation than 2008 during the growing season, but experienced about the same cumulative PAR than 2008 (3919 mol m2 and 3935 mol m2 respectively; Table 3 1). Distance to Water Table (DWT) There was an overall significant difference in depth to water table across the years (P < 0.0001), but no significant differences among sites ( P = 0.9), nor site by year interaction ( P = 0.9)( Figure 3 2). Depth to water table ( DWT ) is measured as the distance from the soil surface to the surface of the water t able; greater DWT corresponds to drier surface soi l conditions. Interesting intera nnual variation was observed ( Figure 3 2), where some years with low water table corresponded well with low precipitation (2004), and years with high water table corresponded with high precipitation (2008). Surprisingly, years with similar DWT (2007 = 17 1 cm and 2009 = 18 1 cm), had opposing precipitation, where 2007 and 2009 growing season rain fall were 331 mm and 178.2 mm respectively (Tables 3 1 & 3 2). For most of th e years, DWT at Minimal Thaw was higher than at the other two sites (Table 3 2), and when
27 years with low precipitation (e.g. 2004, 2009 and 2010), the difference became less pronounced. This pattern is further supported by field observations, where the wat er table in thermokarst was up to the soil surface in times of prolonged precipitation. After rain events, the water table remained high in thermokarst for prolonged periods of time. The variation across years was observed to have a trend towards wetter s oil conditions dominated by a greater 2004 DWT, twice as thick then that of 2010 for all sites (Figure 3 2). This trend was further tested using analysis of covariance (ANCOVA), where across sites, DWT decreased during the seven year sampling period (F = 23.57, P < 0.0001). However, we detected no difference in the rates of decrease among sites (F = 0.14, P = 0.869), nor differences among sites (intercepts, F = 1.01, P < 0.385; Figure 3 2 & Table 3 5 ). Active Layer Depth (ALD) Average t haw depth (May Sept ember ) and active layer depth (maximum seasonal thaw in September) together describe the timing and magnitude of seasonal surface soil thaw. There were no overall differences among the three sites for ALD ( P = 0.72), nor was there a significant year by sit e interaction ( P = 0.44), most likely due to high withinsite variability especially for Extensive and Moderate Thaw ( Figure 3 3A ). There was significant interannual variation across the sites (p < 0.001), largely driven by variation in the initial years o f the dataset with 2004 having s ome of the shallowest ALD and 2005 having some of the deepest (Table 3 2). Defining the maximum of this interannual variation, Moderate Thaw had a 28% increase in ALD between 2004 and 2005, but then returned to values in 2006 that were only 9% higher than 2004. High ALD values in 2005 coincided with low water table and low precipitation for that year and the year before. Some of the interannual variation was also caused by a trend of increasing ALD
28 through the time period of this dataset. Analysis of covariance (ANCOVA) revealed that across sites, there was a slight increase in ALD from 2004 through 2010 (F = 3.50, P = 0.0786) and significant intercept differences among sites (F = 12.76, P = 0.0004). Active layer depth was hig her in Extensive Thaw than Minimal Thaw (P < 0.001) and Moderate Thaw ( P = 0.011), but intercepts did not differ between Minimal and Moderate Thaw ( P = 0.241) That the three sites were continuing to increase in ALD fits the idea that the process of permaf rost thaw and ground subsidence is propagating across this hillslope with the largest current changes occurring at the site that was initially least affected at the outset of this gradient study. Interannual patterns of growing season average TD mirrored A LD in many respects, with the exception of 2009 and 2010 that had among the lowest TD values across years for all sites (Table 3 2, Figure 3 3B ). These low values helped produce a trend of decreasing TD ac ross sites (ANCOVA, F = 11.97, P = 0.003), with a r ate of decrease not significantly different among sites (ANCOVA, F = 0.14, P = 0.875; Figure 3 3B ), but significantly different intercepts among sites (ANCOVA, F = 6.34, P = 0.009; Figure 3 3B ). Decreasing TD shows that, on average, soil thaw was proceeding more slowly through the growing season across the time period of this dataset, but eventually reaching a similar ( Moderate and Extensive Thaw ) or increasing ( Minimal Thaw ) maximum depth of thaw. Ecosystem Carbon Fluxes Growing Season Net Ecosyst em Exchange (NEE) Repeated measure ANOVA on growing season (NEE) detected differences among years (p<0.0001), and site by year interaction ( p < 0.0001)( Figure 3 4A Table 33 Table 35 ). Overall site effect was not detected by the test ( P = 0.5). Interannual
29 variation for all sites was evident, where the lowest and greatest NEE occurred in 2005 and 2009, respectively (Table 3 3). Between these two years, Moderate Thaw had the greatest NEE variation among the three sites (43 24 to 173 65 g C m2). In 20 05, Extensive Thaw was a source of carbon ( 18 13 g C m2).The following year, Extensive Thaw became a carbon sink, with quantities of carbon sequestered seven times greater than that of 2005 (Table 3 3). This carbon sink persisted for the duration of th e study period. In contrast, Minimal and Moderate Thaw were weak carbon sinks in 2005, but doubled and tripled, respectively, the amount of carbon sequester ed by the following year. At Moderate Thaw NEE fluctuated from high to low from one year to the nex t, with the exception of 2009, where a doubled net uptake of carbon occurred. By the end of the study period, Minimal and Extensive Thaw doubled and tripled, respectively, the amount of carbon sequestered at the first year of the study. The full model coul d not detect differences among sites ( P = 0.5 ; Table 35). Although not significant, Moderate Thaw was a stronger carbon sink than both Minimal and Extensive Thaw for 2004, 2009, and 2010 (Table 3 3).Within 2009, Moderate Thaw sequestered about 37% more ca rbon than Minimal and Extensive Thaw No differences in NEE were detected between Minimal and Extensive Thaw (2005, 2006, and 2008), nor were overlapping ranges of mean growing season NEE evidenced (Table 3 3). This is most likely due to the highly variable net carbon exchange wit hin the chamber locations at Extensive Thaw (75 to 334 g CO2C m2). Furthermore, mean NEE for the three sites trended upwards over the seven years. An analysis of covariance (ANCOVA) revealed a significant increase in growi ng season NEE across the seven year measurement period (F = 16.16, P = 0.0009; Figure 3 4A ).
30 However, we detected no differences in the rate of change (slope) among the three sites (site by time: F = 0.22, P = 0.808), nor a difference in site represented by the intercepts (F = 1.51, P = 0.2483). The increasing trends were mainly driven by the large differences in NEE between 2005 and 2009. Growing Season Gross Primary Productivity (GPP) There were overall differences among the three sites for GPP ( Figure 3 3B; P = 0.09), but not a significant year by site interaction ( P = 0.15). Interannual variation differed across the sites (p < 0.001), where the greatest differences are shown between the early and final years of the data set (Table 3 3), suggesting an in crease of GPP over the years at the three sites. Across sites, GPP followed closely the patterns observ ed in NEE, where GPP was lowest in 2005, and greatest in 2009, with the rest of the years being inter mediate. In 2005 GPP was 62%, 64% and 55% lower than those of 2009, and 23%, 40% and 30% lower than those of 2010 for Minimal Moderate and Extensive Thaw respectively. Site differences were significant in years 2004 ( P = 0.08), 2006 ( P = 0.07), 2008 ( P = 0.6) and 2010 ( P = 0.07), where they were mainly dr iven by the large variation between Minimal Thaw and both Moderate and Extensive Thaw Within these years, GPP at Extensive and Moderate Thaw exceeded those of Minimal Thaws GPP by as much as 31% to 46%. There were no signi ficant differences in GPP detect ed between Moderate and Extensive Thaw for all years, although in 2007, the range in mean GPP at Extensive Thaw (527 58 g C m2) did not overlap the range in mean GPP of Moderate Thaw (431 19 g C m2) (Table 3 3). Throughout t he study period, mean GPP at Minimal Thaw never exceeded mean GPP at Moderate and Extensive Thaw GPP at Moderate and Extensive Thaw were more similar; remaining between 1% to 1 0% of each other (except 2007).
31 Similar to the interannual NEE trends, GPP at the three sites trended upwards through the time period of this study ( Figure 3 4B ). There was a significant linear increase in GPP across the seven year measurement period (ANCOVA, F = 19.10, P = 0.0004; Figure 3 4B ), where the variation between 2005 and 2009 seemed to have driven the upward increase. While the rate of increase in GPP over time was not significantly different among sites ( ANCOVA, F = 0.29, P = 0.753), intercepts of the regression lines were significantly different ( ANCOVA, F = 12.54, P = 0.0005), with roughly 37% l ower GPP in Minimal Thaw than either Moderate or Extensive Thaw (P < 0.05 for both). Growing Season Ecosystem Respiration (Reco) There were overall significant differences among sites for growing season Reco ( P = 0.01), and significant year by site interaction ( P = 0.02). There was also significant interannual variation across the sites ( P = 0.005), where the greatest difference between years differed according to site. The interannual variation at Minimal Thaw although significant, remained low thr oughout the study period ( Figure 3 3C ), where 2009 Reco was highest (321 16 g C m2), and 2006 lowest (266 18 g C m2,20% difference). Moderate and Extensive Thaw had a greater interannual variation, where the lowest values occurred in 2006 (313 30 g C m2), and 2010 (345 18 g C m2) respectively, and highest values occurred in 2008 (399 20 g C m2) and 2009 (410 23 g C m2) respectively. Individual comparisons among sites within years showed that Minimal Thaw had significantly lower Reco than E xtensive Thaw (Table 3; 0.004
32 Reco (0.004
33 Annual Net Ecosystem Exchange Repeated measure ANOVA on annual NEE detected differences among years (p<0.0001), and site by year interaction ( P = 0.001), but no site differences ( P = 0.4)( Figure 3 5A Table 3 3). Winter respiration had a profound effect on annual carbon exchange for all three sites, where some years went from car bon sink in the growing season ( Figure 3 4A ) to annual carbon neutral or source ( Figure 3 5A ). Minimal and Extensive Thaw were carbon sources the first two years of the data set, but then remained sinks for the remaining years (Table 3 3). In 2009, both Mi nimal and Extensive Thaw sequestered about 3 and 4 times the amount of carbon released in 2004 respectively. Moderate Thaw although remaining about neutral the first two years, was a much stronger sink of carbon in 2009, doubling the amount of carbon sequ estered by both Minimal and Extensive Thaw The interannual variation was driven by an increasing trend in annual NEE for all sites ( Figure 3 5A ). ANCOVA analysis showed that across sites, there was a significant increase in annual NEE from 2004 through 2009 (F = 31.05, P < 0.001; Figure 3 5A ). However, we detected no differences in t he rate of carbon uptake (site by time: F = 0.14, P = 0.868), nor differe nce among sites (F = 1.79, P = 0.203) Annual Ecosystem Respiration There were overall differences among the three sites for annual Reco ( P = 0.002), but not a significant year by site interaction ( P = 0.23). Interannual variation differed across the sites ( P = 0.007), even thought the range of mean Reco remained low after including winter res piration in the annual budget ( Figure 3 5B Table 3 3). A gentle downward trend across all sites was observed for the first three y ears, where Reco decreased by 12% by the third year, but later shifted to an upward trend, ending in
34 similar Reco of that from the beginning of the study period ( Figure 3 5B ). Differences in annual Reco were evident among sites, where at Extensive Thaw Re co was between 30% to 40% higher than Minimal Thaw with Moderate Thaw fluctuating i n between ( Figure 3 5B Table 3 3). This similar pattern was observed for both, growing season (28 to 40%, Figure 3 4C ), and winter respiration Reco rates (~40%, Figure 3 5C ). The small interan n ual fluctuation in annual Reco for all sites, contributed to the lack of directional trend over the study period. Further analysis (ANCOVA) revealed no linear change in annual Reco from 2004 through 2009 across sites (F = 0.09, P = 0.769), nor a time by site interaction (F = 0.43, P = 0.660); however, there were significant site (intercept) differences (F = 37.92, P < 0.001), where annual respiration differed among all three sites (p < 0.05 for all comparisons; Figure 3 5B ). These r esults mirror those of the repeated measure ANOVA. Above Ground Net Primary Productivity (ANPP) Repeated measure ANOVA on ANPP revealed differences among years (Table 3 2, p<0.0001), and site by year interaction ( P = 0.06). Overall site effect was not detected by the test ( P = 0.6). Minimal and Moderate Thaw had the greatest interannual variation with a peak ANPP in 2008 ( Figure 3 6,Table 3 2). By 2008, ANPP at Minimal Thaw had increased by 133% from the first year of the data series. Moderate and Extensive Thaw had a similar increase, but not as pronounced as Minimal Thaw (67% and 23% respectively). This trend in interannual variation reflected an upward trend over the duration of the dataset ( Figure 3 5). There was an overall significant increase in ANPP over the years (ANCOVA, F = 21.5, P = 0.0007), and no site by time interaction ( ANCOVA, F = 1.23, P = 0.3), nor differences in sites ( ANCOVA, F = 1.43, P = 0.3)( Figure 3 6).
35 Long Term Carbon Exchange Within the long term growing season (LTGS) average, non e of the sites significantly differ from one another in net ecosystem exchange ( Figure 3 7A; P = 0.6 and P = 0.4 ). Even though there is a lack of significant difference, LTGS NEE average at Minimal Thaw (80 14 g CO2C m2) was lower than Extensive (102 34 g CO2C m2), and Moderate Thaw (122 30 g CO2C m2) Both LTGS GPP and Reco differed within sites ( P = 0.01 and P = 0.09 respectively), where GPP overcame Reco by 27%, 33% and 25% for Minimal Moderate and Extensive Thaw GPP and Reco at Minimal Thaw were significantly lower than at Extensive Thaw (Table 3 3) by about 34%, with Moderate Thaw GPP and Reco being about 2% and 8% lower than Extensive Thaw (although not significant) (Table 3 3). In order to estimate the carbon exchange annual long ter m (ALT) average, we had to omit the growing season of 2010, since the annual cycle for 2010 started on May 1st, and ended on April 30th of the following year ( Figure 3 7B ). The seven year LTGS carbon exchange average varied by less than 1% of those estimat ed using six year average. When winter respiration was added to obtain an annual budg et, Extensive Thaw became carbon neutral (Figure 3 7; t = 0.71, P = 0.25) where as Minimal and Moderate Thaw remained sinks of carbon (Figure 3 7; t < 2.3, p < 0.05) Eve n though a one way ANOVA did not detect significant differences among sites ( P = 0.4), Moderate Thaw mean annual NEE more than doubled mean annual NEE at Minimal Thaw ( Figure 3 7B ). LTA Reco differed among the three sites ( P = 0.003), where Minimal Thaw had the lowest Reco rates, and Extensive Thaw the highest, with Moderate Thaw intermediate ( Figure 3 7B ) Long term winter respiration comprised about 1517% of the
36 annual carbon respired, where Extensive Thaw contributed more carbon release than at the other two sites (Table 3 3). Carbon Response to Abiotic Factors Using a mixed model approach we found that for GPP, the intercept of the model varied by site, and ALD stood out as the most important explanatory variable (Table 3 4). ALD was highly significant and the slope indicated that as ALD increased by 1 cm, GPP increased by 0.003 0.0006 log units g C m2. Also, ANPP was marginally significant and the model estimated that as ANPP increased, GPP increased (P = 0.001; Table 34) For GS Reco, the intercept of the model varied by site, and ALD was also highly significant. The slope estimate indicated that as ALD increase by 1 cm, GS Reco increased by 2.0 0.04 g C m2. DWT was found to be marginally significant, with a negati ve slope, indicating that as DWT increased (became more dry), GS Reco increase (P = 0.13; Table 34). For GS NEE, the intercept of the model also varied by site, and ALD came out to be the most i mportant explanatory variable. The slope estimate for ALD was highly significant and indicated that as ALD increased by 1 cm, GS NEE increased by 1.6 0.05 g C m2 (P = 0.004; Table 34) Furthermore, the sites intercepts varied significantly for all models, and removing the intercept increased the residual variance and AIC. This indicat ed there are other unmeasured factors that vary with site that influences carbon flux. In addition, the standard deviation of the random intercept among years was larger than zero for all models, which indicated that we were justified in using a mixed model approach. Based on these results, we further investigated the relat ionship between ALD, DWT and ANPP to growing season CO2 exchange by fitting linear regressions for each of the factors that best explained carbon flux.
37 Active layer dept h (ALD ) was a good predictor of growing season carbon exchange (NEE, GPP and Reco) within the growi ng seasons seven year period ( Figure 3 8). ALD measured at the chamber location was positively correlated with GS NEE (R2 = 0.13, p<0.0001; Figure 3 8A ), GS GPP ( R2 = 0.31, p<0.0001; Figure 3 8B ), and GS Reco (R2 = 0.24, p<0.0001; Figure 3 8C ). The correlation was mainly driven by chambers located in highly variable ALD sites ( Extensive Thaw ALDs of 45 cm to 122 cm and Moderate Thaw ALDs of 50 cm to 106 cm), where the active layer was deeper than areas of no, or slight thaw ( Minimal Thaw plus some chamber locations within Extensive and Moderate Thaw ). At the site level, Extensive Thaw had the highest correlation between ALD and GS NEE, GS GPP, and GS Reco, followed by Moderate Thaw with Minimal Thaw showing no significant correlation. Average g rowing season thaw depth (Avg GS TD) was not as good of a predictor of CO2 exchange as ALD was, although CO2 exchange responded similar to both ALD, and avg GS TD ( GS NEE R2 = 0.02 p<0.001; GS GPP R2 = 0.16 p<0.001 and GS Reco R2 = 0.2 6 p<0.001 ) (Figure not shown). A marginally significant relationship between DWT and GS GPP was evident, where GS GPP decrease as the depth to the water table increased (R2 = 0.17, P = 0.06; Fi gure 3 9B ). Simila rly, as DWT increased (became dryer) NEE decreased, although this trend was weak (R2 = 0.11, P = 0.1; Figure 3 9A ). There was no relationship between Reco and DWT (R2 = 0.03, P = 0.46; Figure 3 9C ) although a slight decrease in GS Reco can be seen with an increased DWT the mixed model results, mirrored those of the single linear relationship Reco increased with a greater DWT Both of these results were not significant, indicating that there was too much variability in the data to
38 detect the expected relationship. The month of May was necessarily excluded, as this months data masked growing season trends due to the freezing of water from May precipitation on the active layer. We found a relationship between ANPP and ALD ( Figure 3 10; R2 = 0.31, p<0.0001), where deeper thawed areas had a higher ANPP. This relationship follows closely to an increase in GS GPP with deeper thawing ( Figure 3 8B ). The relationship was mainly driven by areas within Extensive Thaw that had undergone deep thaw ( 100 cm and greater), and a plant species shift to deciduous shrubs and hydrophilic mosses. Another predictor of ANPP was DWT ( R2 = 0.51, P = 0.003 ; Figure 3 11), where as depth to the water table increased ANPP decrease d This relationship is further supp orted by a similar relationship between DWT and GS GPP ( Figure 3 9B ).
39 Fig ure 3 1. Seasonal thaw for each site and water table depth averaged across the three sites over a period of seven years. Soil surface is represented by zero on the x axis.
40 Fig ure 3 2. Growing s eason (June Sept ember ) average depth to water table by site. Soil surface is represented by zero on the x axis The larger the absolute values, the greater the distance of t he surface of the water table from the surface o f the soil P values are results for the ANCOVA test. T he single solid line indicates no significant differences in y intercept among sites
41 Fig ure 3 3 ( a ) A ctive layer depth (ALD) measured in September ( b ) ave rage growing season thaw depth (Avg GS TD) for three sites varying in degree o f permafrost thaw. P values are results for the ANCOVA test. S olid line s indicate sa me y intercept for Minimal and Moderate Thaw Different numbers of a sterisks indicate significant difference in y intercept s. year
42 Fig ure 3 4. Growing season (May September) (a) net ecosystem exchange (NEE), (b) gross primary production (GPP), and (c) ecosystem respiration (Reco) over seven consecutive years for three sites varying in degree of permafrost thaw. P values are resul ts for the ANCOVA test. Solid line in panel (a) indicates no significant differences in y intercept among sites. Solid line in panel (b) indicates no dif ference in y intercept betw een Moderate and Extensive Thaw Dashed line represent Minimal Thaw Differe nt numbers of asterisks indicate significant difference in y intercept s.
43 Fig ure 3 5. Annual (a) NEE, (b) Reco, and (c) winter Reco (Oct April) over six consecutive years for three sites varying in degree of permafrost thaw. P values are results for the ANCOVA test. Solid line indicates no significant difference in y intercept among sites Different numbers of a sterisks indicate significant difference in y intercept s.
44 Fig ure 3 6. Aboveground net primary productivity for t hree sites varying in degree of permafrost thawing. 2005 and 2010 were not measured. P values are results for the ANCOVA test. Solid line indicates no significant differences in y intercept among sites
45 Figure 37. Long term average of (a) growing season, (b) winter, and annual carbon exchange for three different sites varying in degrees of permafrost thaw = 0. 0 5 ). Asterisk denotes marginally signific ant difference ( P value 0.08).
46 Fig ure 3 8. Relationship between active layer depth (ALD) and gro wing season (May September) (a) net ecosystem exchange ( GS_ NEE), (b) gross primary production ( GS_ GPP), and (c) ecosystem respiration ( GS_ Reco). All rel ationships have an associated p value <0.0001. Each point represents a single carbon flux base in a given y ear and the regression line is fit through all data points.
47 Fig ure 3 9. Relationship between growing season (June September) depth to water table ( DWT ) and growing season (a) net ecosystem exchange (GS_NEE) (b) gross primary production (GS_GPP) and (c) ecosy stem respiration ( GS_ Reco). H igher DWT values represent drier conditions Data points are aggregated the site level, since the water wel l s are n ot directly paired with carbon flux bases. Each point represents a specific year and site, and the regression line is fit through all data points.
48 Fig ure 3 10. Relationship between ac tive layer depth (ALD) and growing season above ground net primary production (ANPP) Each point represents a single carbon flux base in a given y ear, and the regression line is fit through all data points. Fig ure 311. Relationship between depth to water table (DWT) and aboveground net primary production (ANPP ). Each point represents a specific year and site, and the regression line is fit through all data points.
49 Table 3 1 Average photosynthetic active radiation ( PAR ), precipitation, air t emperature, soil t emperature and cumulative PAR from the weather s tat ion at Eight Mile Lake Growing Season (May September) Year Cumulative PAR ( mol m -2 ) Average PAR ( mol m -2 sec -1 ) Precipitation ( mm ) Air Temp ( o C ) Soil Temp ( o C ) 2004 332 153.2 11.5 6.7 2005 3767 341 145.2 10.8 7.2 2006 3906 354 227.4 9.1 6.8 2007 4116 375 331.4 10.3 7 2008 3935 357 346.2 8.1 5.9 2009 4147 376 178.2 9.7 6.8 2010 3919 356 182 9.8 7.7 Soil temperature at a depth of 10cm from Minimal Thaw only
50 Table 32. Aboveground net primary productivity (ANPP), average growing season thaw depth (Avg GS TD), active layer depth ( ALD), and average growing season depth to water table (DWT) for three sites varying in degree of permafrost thaw Growing Season (May September) Site Year ANPP ( g C m -2 yr -1 ) Avg GS TD ( cm ) ALD ( cm ) DWT ( cm ) Minimal 2004 105 6 41 1 58 1 31 3 Moderate 2004 154 28 41 2 57 2 30 2 Extensive 2004 168 29 48 6 68 5 30 3 Minimal 2005 44 2 62 2 22 2 Moderate 2005 50 3 73 3 20 1 Extensive 2005 57 7 76 5 17 2 Minimal 2006 116 10 35 1 59 1 23 1 Moderate 2006 152 17 38 2 62 1 21 1 Extensive 2006 140 25 42 5 69 4 20 2 Minimal 2007 163 24 43 1 65 1 19 1 Moderate 2007 207 21 45 2 68 2 16 1 Extensive 2007 173 32 50 5 71 4 15 1 Minimal 2008 245 17 37 1 65 2 15 1 Moderate 2008 258 18 40 2 66 2 11 1 Extensive 2008 207 37 46 6 73 5 11 1 Minimal 2009 218 24 34 1 62 1 19 1 Moderate 2009 231 16 35 1 66 2 17 1 Extensive 2009 216 32 39 4 73 5 19 1 Minimal 2010 35 1 66 1 16 2 Moderate 2010 36 1 67 1 13 1 Extensive 2010 41 3 72 3 15 1 There were no significant differences among sites within each year.
51 Table 3 3. Seasonal and annual carbon fluxes for three sites varying in degree of permafrost t haw Growing Season (May September) Site Year GPP R eco NEE Winter (October April) R eco Annual R eco Annual NEE Minimal 2004 327 20 a 288 12 a 39 15 a 75 12 a 362 15 a 36 16 a b Moderate 2004 445 48 b 339 27 ab 105 26 a 79 14 a 418 25 a 26 26 b Extensive 2004 450 56 b 404 37 b 46 22 a 104 18 b 507 35 b 58 23 a Minimal 2005 300 14 a 276 9 a 29 6 a 48 10 a 324 14 a 19 04 a Moderate 2005 377 34 a 341 24 ab 43 24 a 50 11 a 390 23 ab 7 24 a Extensive 2005 376 28 a 393 36 b 18 13 a 61 16 a 454 29 b 78 13 a Minimal 2006 322 24 a 266 18 a 62 14 a 57 11 a 324 16 a 5 14 a Moderate 2006 424 51 a b 313 30 ab 117 35 a 60 12 a 373 24 ab 57 36 a Extensive 2006 474 83 b 359 58 b 121 31 a 81 17 b 440 39 b 40 31 a Minimal 2007 382 33 a 290 21 a 92 17 a 47 9 a 337 21 a 45 17 a Moderate 2007 431 19 ab 373 19 b 58 24 a 50 10 a 423 19 b 8 24 a Extensive 2007 527 58 b 406 34 b 121 31 a 66 14 b 472 34 b 55 31 a Minimal 2008 370 21 a 288 22 a 82 7 a 54 10 a 342 22 a 28 7 a Moderate 2008 514 22 b 399 20 b 115 26 a 58 12 a 457 20 b 57 26 a Extensive 2008 507 51 b 380 27 b 127 27 a 78 17 b 458 27 b 49 27 a Minimal 2009 486 37 a 321 16 a 164 35 a 45 9 a 366 16 a 119 36 b Moderate 2009 617 40 a 348 22 ab 268 33 a 49 10 a 397 21 a 220 33 a Extensive 2009 584 68 a 410 23 b 173 65 a 64 14 b 475 23 b 109 64 b Minimal 2010 370 24 a 282 22 a 88 30 a Moderate 2010 527 43 b 377 44 b 150 48 a Extensive 2010 487 83 ab 345 18 ab 142 80 a Minimal Average 364 15 a 287 10 a 80 14 a 53 4 a 342 9 a 23 12 a Moderate Average 475 30 b 356 18 ab 122 30 a 57 4 a 409 16 b 60 26 a Extensive Average 484 59 b 385 28 b 102 34 a 77 6 b 467 31 b 19 27 a Different letters denote significant thin year, based on unadjusted P values. Units are in grams CO2C m-2
52 Table 3 4. Mixed mode l parameter estimates with associated standard errors and P values for GPP, Reco, and NEE. All explanatory variables were centered before analysis and GPP was log10 transformed for normality. The standard deviation of the random intercept for year and the standard deviation of the residuals are also shown. Ex planatory variables: site (Min= Minimal Thaw Mod=Moderate Thaw Ext = E xt ensi ve Thaw ), precip = cumulative growing season precipitation (mm), Air t emp = average GS air temperature (C), Avg PAR = average GS PAR (mol m2 sec1), ALD = maximum active layer depth (cm); D WT= distance to water table (cm). Variables NEE GPP R eco estimate Stnd err p value estimate Stnd err p value estimate Stnd err p value Min (intercept) 87.048 24.749 0.001 2.5779 0.0274 0.000 300.784 10.234 0.000 Mod (intercept) 122.022 23.902 0.000 2.6586 0.0264 0.000 354.183 9.827 0.000 Ext (intercept) 94.704 23.875 0.000 2.6427 0.0264 0.000 373.131 9.971 0.000 ANPP 0.530 0.157 0.001 0.0006 0.0002 0.001 0.111 0.120 0.357 ALD 1.623 0.544 0.004 0.0030 0.0006 0.000 1.924 0.447 0.000 Precipitation 0.141 0.370 0.718 0.0001 0.0004 0.807 0.046 0.106 0.687 DWT 7.729 16.734 0.645 0.0199 0.0186 0.286 10.912 7.138 0.129 Air Temp 27.661 26.869 0.351 0.0188 0.0297 0.554 7.042 7.398 0.385 Random std dev 54.915 0.061 7.012 Residual std de v 71.10 0.079 59.8
53 Table 35 Results for the repeated measure ANOVA full model, and the ANCOVA full model and reduced model Repeated Measure ANOVA ANCOVA ** Variable Effect F df p value F df p valu e GS NEE year 17.96 6 <0.0001 16.15 1 0.0009 site 0.67 2 0.5 1.51 2 0.25 year x site 3.86 12 0.0001 0.22 2 0.81 GS GPP year 16.85 6 <0.0001 19.1 1 0.0004 site 2.85 2 0.09 12.54 2 0.0005 year x site 1.45 12 0.16 0.29 2 0.75 GS R eco year 3.31 6 0.005 0.76 1 0.39 site 5.65 2 0.01 29.7 2 <0.0001 year x site 2.14 12 0.02 1.91 2 0.18 ANPP year 21.94 4 <0.0001 21.54 1 0.0007 site 0.55 2 0.6 1.43 2 0.28 year x site 2.02 8 0.06 1.23 2 0.33 Annual NEE year 22.34 5 <0.0001 31.05 1 <0.0001 site 0.85 2 0.44 1.79 2 0.2 year x site 3.35 10 0.001 0.14 2 0.87 Annual R eco year 3.47 5 0.007 0.09 1 0.77 site 9.8 2 0.0019 37.9 2 <0.0001 year x site 1.33 10 0.23 0.43 2 0.66 Annual Winter year 8.04 1 0.13 site 6.89 2 0.008 year x site 0.02 2 0.98 ALD year 13.84 6 <0.0001 3.5 1 0.078 site 0.34 2 0.7 12.76 2 0.0004 year x site 1.02 12 0.43 0.27 2 0.77 DWT year 17.15 6 <0.0001 23.57 1 0.0001 site 0.11 2 0.89 1.01 2 0.38 year x site 0.56 12 0.86 0.14 2 0.86 R esults of the full model **Results of the full model (interaction term only), and reduced model (year and site only ) The signif icance level for the tests is = 0.05
54 CHAPTER 4 DISCUSSION Our objective was to understand the implications of a warmer climate on the C cycle in ecosy stems underlain by permafrost. As permafrost thaw and thermokarst formation is expected to accelerate with future climate warming (Jorgenson et al 2006, Osterkamp 2007) it is important to quantify how rapidly tundra ecosystems respond to such disturbances, and whether the historical role of these ecosystems as C sinks is changing (Post et al 1982, Oechel et al 1993, Tarnocai et al 2009) The combination of our sevenyear time series measured across sites on a perma frost thaw gradient addresses that question on both inter annual and decadal time scales. Our results show that during the seven year study period, sites on the permafrost thaw gradient that started with more widespread permafrost thaw and thermokarst had both higher Reco and higher GPP rates compared to the site least affected by thermokarst. This was true both during the growing season and on an annual basis when winter respiration was included. Interannual variation observed during the study period showed a linear increase in GPP over the sevenyear period across all study sites, but with no concomitant increase in Reco. Stable Reco combined with increasing GPP caused the three sites to accumulate increasingly larger amounts of C both during the growing se ason and annually across the duration of the study period. This increasing trend in NEE ultimately resulted in all three sites acting as strong C sinks in during the growing season and annually by the end of the seven year period, even though some sites we re C sources near the outset. The sites more affected by permafrost thaw and ground subsidence (Extensive and Moderate Thaw) had higher GPP compared to the site that was least affected
55 (Minimal Thaw) (Figure 3 4). This plant response was likely due to a co mbination of factors altered by permafrost thaw and thermokarst such that the GPP response was not just due to changes in temperature alone (Vourlitis et al 2000) Variation in GPP across sites and years was primarily explained by changes in ALD that occurred spatially within a site as well as among years of the study (Figure 3 8 and Table 3 4). Greater active layer depth reflects a larger volume of seasonally thawed soil where microbial and plant activity is expected to increase in response to increased temperature and nutrient availability. Due to the heterogeneous nature of thaw and ground subsidence, Extensive Thaw had the largest variance in ALD as well as in GPP (Figure 3 3, Figure 34). Deeply subsided microsites had higher maximum photosynthetic rates (F cmax) compared to elevated microsites that exis ted even within Extensive Thaw This illustrates the indirect effect that thaw and subsidence has on the redistribution of other resources such as water (Schuur et al 2007, Osterkamp et al 2009, Vogel et al 2009, Lee et al 2010) Elevated microsites were higher and drier with a shallower ALD, all which decreased GPP while at the same time subsided microsites had increased ALD, GPP, and plant biomass. Despite this bimodal effect on m icrosites, overall permafrost thaw at the Moderate and Extensive Thaw had a positive effect on GPP with those sites elevated in GPP as compared to Minimal Thaw. Variation in GPP was, not surprisingly, also explained by measurements of plant biomass (ANPP) that reflects differences across sites and years (Table 34). Thermokarst formation has led to a shift in plant community composition and structure, where shrubs are most abundant in Extensive and Moderate Thaw, while tussock forming sedges dominate in Minimal Thaw (Schuur et al 2007) This shift in plant
56 community composition towards increased shrub abundance has been documented in other tundra locations in Alaska although not necessarily linked specifically to ground subsidence (Sturm et al 2001, Epstein et al 2004, Tape et al 2006) The most likely mechanism for the shift in plant community composition and change in biomass and productivity is the increase in nutrient availability as a result of the thaw and decomposition of previously frozen organic matter (Chapin & Shaver 1996, Schuur et al 2007) In particular for upland tundra, N availability is considered to be a much more important driver of NPP and plant community composition as compared to temperature alone (Chapin et al 1995) The shift in plant community composition at EML appears to have taken decades, consistent with vegetation changes elsewhere in the arctic (Epstein et al 2004) Sedges still are codominant with shrubs at Moderate Thaw with 1520 years of thawing whereas Extensive Thaw with > 50 years of thawing is dominated by shrubs (Schuur et al 2007) Against this backdrop of plant community composition shifts across the different sites at the EML permafrost thaw gradient, the increasing trend of GPP (Figure 3 4) and ANPP (Figure 3 6) over the sevenyear study period was not associated with a further shift in communit y composition, but rather an increase in plant productivity of the existing communities (data not shown). And, while GPP was generally correlated with moisture availability when this variable (DWT) was considered alone (Figure 3 9), moisture was not a significant variable in the full model (Table 3 4) likely because its variation was correlated with differences in ALD. Similar to GPP, Reco variation across sites and years was strongly controlled by variation in ALD (Table 3 4). Ho wever, In contrast to GPP, there was no directional trend in GS Reco over the study period (Figure 3 4), while there were consistent
57 differences among sites. Minimal Thaw had lower GS Reco than Extensive and Moderate Thaw across the sevenyear period (Figure 3 4C and Table 3 3) corresponding to the shallower ALD at that site. Active layer depth by itself correlated with GS Reco across all sites (Fig 3 8C), but the weak trend of ALD across the sevenyear period (Figure 3 3) meant that there was no directional trend in GS Reco across the study period. The relationship of ALD and GS Reco is likely to be the result of increased thaw of previously frozen organic matter and the decomposition of old C stored at depth in the permafrost surface (Goulden et al 1998, Schuur et al 2007, Nowinski et al 2010) The pattern observed here mirrors trends found in the nearby CiPEHR (Carbon in Permafrost, Experimental Heating Research) warming experiment (Natali et al 2011) as well as observed across other permafrost ecosystem types (Goulden et al 1998, Lantz et al 2009) This control of ALD over Reco (and GPP) was demonstrated with the combination of the sevenyear record along with the decadal scale difference across the thaw gradient (full model). On its own, variation in ALD within the sevenyear period was too small to determine cont rols over C flux rates. In general, ALD is linked to warmer air temperatures (Akerman & Johansson 2008, Huds on & Henry 2009) but can also be related to snow depth in the winter (Walker et al 1999, Nowinski et al 2010, Natali et al 2011) This may represent an important positive feedback to localized permafrost thaw as initial ground subsidence caused by thawing permafrost accumulated more snow than nearby unsubsided areas, thus further warming permafrost and increasing ALD in those subsided microsites. Deeper ALD along with increased moisture in the subsided areas lead to a longer period where the soil remained unfrozen in the fall freezeup period. This lead to
58 increased winter Reco in Extensive T haw with the deepest active layer (Schuur et al 2009, Vogel et al 2009) Other studies have shown higher winter Reco due to warmer soil temperatures, and higher snow accumulation (Fahnestock et al 1999, Nobrega & Grogan 2007, Natali et al 2011) Winter respiration plays an important role in decreasing the growing season sink strength, and in the case of 2004 and 2005 even shifted these ecosystems into annual C sources (Table 33 ). While winter respiration was relatively high in 2004, there was no directional trend across the study period. In these upland tundra sites, we did not detect a relat ionship between DWT and Reco based on simple regression (Figure 3 9), however the mixed model did indicate a weak negative relationship between those two variables (P = 0.129). Increased DWT, indicative of drier conditions, decreased Reco. This response of upland tundr a differed from lowland tundra and peatlands where water table controls over C emissions are well documented (Bellisario et al 1998, Bubier et al 2003, Dunn et al 2007) Nevertheless, even in these upland sites, the water table perched on the permafrost surface creates the potential for anaerobic zones within the soil profile. Indeed, methane concentrations above atmospheric values (data not shown) suggest that there is the potential for negative water effects on decomposition although little to no methane is emitted to the atmosphere likely as a result of consumption in the surface soil layers (data not shown). Additionally, the water table wells at this site were not paired directly with the carbon flux bases, and this decoupling of measurements that decreased statistical power could have masked an overall relationship between Reco and water table. The combination of increasing GPP along with stable Reco produced an increasing trend of NEE over the duration of the study period, both for the growing season and
59 annually. Within the grow ing season, all sites acted as C sinks with the exception of 2005, where Extensive Thaw was actually a C source. Across sites, Moderate and Extensive Thaw trended toward greater average GPP (Figure 3 7), but this increased uptake was matched by increased Reco such that the site NEE did not differ statistically from one another. Other studies have shown similar results in different permafrost ecosystems, where the strength of the growing season NEE has increased within the recent past. A combination of plant acclimation to warmer temperatures, position of the water table, and temperature sensitivity of GPP seemed to play an important role in dictating the fate of net C exchange (Oechel et al 2000, Welker et al 2004, S onnentag et al 2010) Other studies have linked dry summers and low water table to low GS NEE rates (Bubier et al 2003, Aurela et al 2007) Annual NEE, which includes additional C release via winter respiration, overall decreased the growing season sink strength, such that some years (2004, 2005) were predominantly C sinks across sites. This again demonstrates the importance of winter respiration measurements in arctic ecosystems; while respiration in winter was only ~20% of the annual Reco this shifted the overall s ign of NEE from C sink to source in some years similar to previous studies (Fahnestock et al 1999) Averaged across years, the addition of winter respiration decreased the sink strength of Minimal a nd Extensive Thaw such that they were C neutral. Moderate Thaw remained an annual C sink, as annual Reco there was not as high as Extensive Thaw even though GPP levels of the two sites did not statistically differ (Figure 3 7). This pattern of annual NEE m irrored that demonstrated in earlier analyses of the 20042006 study period (Schuur et al 2009, Vogel et al 2009) Extensive Thaw had a higher winter respiration rate than
60 both Moderate and Minimal Thaw, which was mainly attributed to the greater ALD, increased water content, and warmer soil temperatures. This pattern held for the remaining four years of the study, where winter respi ratio n remained higher at Extensive T haw than at the other two sites. These findings imply that as the ALD increases, annual Reco eventually overcomes the higher GPP rates that are also associated with a greater ALD (Figure 3 7B).
61 CHAPTER 5 CONCLUSION Overall, permafrost thaw and associated ground subsidence lead to a shift in plant community composition with higher shrub abundance, which in turn lead to higher photosynthetic rates, increased active layer depth, and higher soil respiration. Across the decadal scale of the thaw gradient, the initial stimulation of C uptake by plant growth was greater than increases in respiration such that the tundra was an annual sink of C, but with more decades of thawing, plant growth does not keep up and increased respiration reduces the C sink strength and in some microsites even transforms tundra into a C source. Analysis of the first 3 years of this dataset showed that decades of thaw transformed the tundra into a C source, but the trend of increasing plant C uptake through the sevenyear study period with no changes in respiration had the effect of reducing the source strength of the most thawed site when averaged across years. Interestingly, this recent period of measurements corresponds to a decadal pattern of thaw which shows that warming and permafrost thaw initiated several decades ago has given way to a slight cooling trend tied to decadal scale climate oscillations. This short term reversal in the warming trend is not expected to last, and it will be interesting to document the trajectory of C uptake and release if and when warming continues to increase for these high latitude ecosystems. Given the role the arctic and subarctic regions have played in C sequestration for thousands of years, the significance of these changes extends from the tundra to the global environment.
62 LIST OF REFERENCES ACIA (2004) Impacts of a Warming Arctic, Cambridge University Press, Cambridge. Akerman HJ, Johansson M (2008) Thawing permafrost and thicker active layers in subarctic Sweden. Permafrost and Periglacial Processes, 19 279 292. Aurela M, Riutta T, Laurila T et al. (2007) CO2 exchange of a sedge fen in southern Finland The impact o f a drought period. Tellus Series B Chemical and Physical Meteorology, 59 826837. Bellisario LM, Moore TR, Bubier JL (1998) Net ecosystem CO2 exchange in a boreal peatland, northern Manitoba. Ecoscience, 5 534541. Bubier JL, Bhatia G, Moore TR, Roulet NT, Lafleur PM (2003) Spatial and temporal variability in growing season net ecosystem carbon dioxide exchange at a large peatland in Ontario, Canada. Ecosystems, 6 353367. Burn CR, Smith MW (1993) I ssues in canadian permafrost research. Progress in Physical Geography, 17 156 172. Chapin FS, Shaver GR (1996) Physiological and growth responses of arctic plants to a field experiment simulating climatic change. Ecology, 77, 822 840. Chapin FS, Shaver GR, Giblin AE, Nadelhoffer KJ, Laundre JA (1995) R esponses of arctic tundra to experimental and observed changes in climate. Ecology, 76 694711. Clymo RS (1970) G rowth of sphagnum methods of measurement. Journal of Ecology, 58, 13 &. Dunn AL, Barford CC, Wofsy SC, Goulde n ML, Daube BC (2007) A long term record of carbon exchange in a boreal black spruce forest: means, responses to interannual variability, and decadal trends. Global Change Biology, 13 577590. Dutta K, Schuur EAG, Neff JC, Zimov SA (2006) Potential carbon release from permafrost soils of Northeastern Siberia. Global Change Biology, 12 23362351. Epstein HE, Calef MP, Walker MD, Chapin FS, Starfield AM (2004) Detecting changes in arctic tundra plant communities in response to warming over decadal time scal es. Global Change Biology, 10 13251334. Fahnestock JT, Jones MH, Welker JM (1999) Wintertime CO2 efflux from arctic soils: Implications for annual carbon budgets. Global Biogeochemical Cycles, 13 775 779. Goulden ML, Wofsy SC, Harden JW et al. (1998) S ensitivity of boreal forest carbon balance to soil thaw. Science, 279 214 217.
63 Harden J (2008) The fate of terrestrial carbon following permafrost degradation: Detecting changes over recent decades. Proc. Ninth Int. Conf. Permafrost, edited by D. L. Kane and K. M. Kinekl, pg. 649 653, Inst. of North. Eng., Univ. of Alaska Fairbanks, Fairbanks, Alaska. Hinkel KM, Hurd JK (2006) Permafrost destabilization and thermokarst following snow fence installation, Barrow, Alaska, USA. Arctic Antarctic and Alpine Res earch, 38, 530539. Hinkel KM, Paetzold F, Nelson FE, Bockheim JG (2001) Patterns of soil temperature and moisture in the active layer and upper permafrost at Barrow, Alaska: 19931999. Global and Planetary Change, 29 293309. Hobbie SE, Nadelhoffer KJ, Hogberg P (2002) A synthesis: The role of nutrients as constraints on carbon balances in boreal and arctic regions. Plant and Soil, 242 163170. Hudson JMG, Henry GHR (2009) Increased plant biomass in a High Arctic heath communi ty from 1981 to 2008. Ecology, 90, 26572663. IPCC (2007) Climate Change 2007: The Physical Science Basis, Cambridge University Press, Cambridge. Jonasson S, Michelsen A, Schmidt IK, Nielsen EV (1999) Responses in microbes and plants to changed temperatur e, nutrient, and light regimes in the arctic. Ecology, 80, 18281843. Jorgenson MT, Osterkamp TE (2005) Response of boreal ecosystems to varying modes of permafrost degradation. Canadian Journal of Forest ResearchRevue Canadienne De Recherche Forestiere, 35 2100 2111. Jorgenson MT, Racine CH, Walters JC, Osterkamp TE (2001) Permafrost degradation and ecological changes associated with a warming climate in central Alaska. Climatic Change, 48 551 579. Jorgenson MT, Shur YL, Pullman ER (2006) Abrupt increas e in permafrost degradation in Arctic Alaska. Geophysical Research Letters, 33, Lantz TC, Kokelj SV, Gergel SE, Henry GHR (2009) Relative impacts of disturbance and temperature: persistent changes in microenvironment and vegetation in retrogressive thaw sl umps. Global Change Biology, 15, 16641675. Lawrence DM, Slater AG (2005) A projection of severe near surface permafrost degradation during the 21st century. Geophysical Research Letters, 32, Lee H, Schuur EAG, Vogel JG (2010) Soil CO2 production in upland tundra where permafrost is thawing. Journal of Geophysical ResearchBiogeosciences, 115
64 Lloyd AH, Yoshikawa K, Fastie CL, Hinzman L, Fraver M (2003) Effects of permafrost degradation on woody vegetation at arctic treeline on the Seward Peninsula, Alaska. Permafrost and Periglacial Processes, 14, 93 101. Lund CP, Riley WJ, Pierce LL, Field CB (1999) The effects of chamber pressurization on soil surface CO2 flux and the implications for NEE measurements under elevated CO2. Global Change Biology, 5 269 281. Mack MC, Schuur EAG, Bret Harte MS, Shaver GR, Chapin FS (2004) Ecosystem carbon storage in arctic tundra reduced by long term nutrient fertilization. Nature, 431, 440 443. McNamara JP, Kane DL, Hinzman LD (1998) An analysis of streamflow hydrology in the Kuparuk River basin, Arctic Alaska: A nested watershed approach. Journal of Hydrology, 206 39 57. Natali SM, Schuur EAG, Trucco C, Hicks Pries CE, Crummer KG, Baron Lopez A F (2011) Effects of experimental warming of air, soil and permafrost on carbon balance in Alaskan tundra. Global Change Biology, 17, 13941407. Nobrega S, Grogan P (2007) Deeper snow enhances winter respiration from both plant associated and bulk soil carbon pools in birch hummock tundra. Ecosystems, 10, 419431. Nowinski NS, Taneva L, Trumbore SE, Welker JM (2010) Decomposition of old organic matter as a result of deeper active layers in a snow depth manipulation experiment. Oecologia, 163 785 792. Oechel WC, Hastings SJ, Vourlitis G, Jenkins M, Riechers G, Grulke N (1993) R ecent change of arctic tundra ecosystems from a net carbondioxide sink to a source. Nature, 361 520 523. Oechel WC, Vourlitis GL, Hastings SJ, Zulueta RC, Hinzman L, Kane D (2000) Acclimation of ecosystem CO2 exchange in the Alaskan Arctic in response to decadal climate warming. Nature, 406 978 981. Osterkamp TE (2007) Characteristics of the recent warming of permafrost in Alaska. Journal of Geophysical ResearchEarth Surface, 112 Os terkamp TE, Jorgenson MT, Schuur EAG, Shur YL, Kanevskiy MZ, Vogel JG, Tumskoy VE (2009) Physical and ecological changes associated with warming permafrost and thermokarst in interior Alaska. Permafrost and Periglacial Processes, 20 235 256. Osterkamp TE, Romanovsky VE (1999) Evidence for warming and thawing of discontinuous permafrost in Alaska. Permafrost and Periglacial Processes, 10, 17 37.
65 Osterkamp TE, Zhang T, Romanovsky VE (1994) E vidence for a cyclic variation of permafrost temperatures in norther n alaska. Permafrost and Periglacial Processes, 5 137144. Pinheiro J, Bates D, DebRoy S, Sarkar D (2010) nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.197. Post WM, Emanuel WR, Zinke PJ, Stangenberger AG (1982) S oil carbon pools and world life zones. Nature, 298 156 159. Schielzeth, Holger (2010) Simple means to improve the interpretability of regression coefficents. Methods in Ecology and Evolution, 1 103113. Schuur EAG, Bockheim J, Canadell JG et al. (2008) Vulnerability of permafrost carbon to climate change: implications for the global carbon cycle. BioScience, 58 701 714. Schuur EAG, Crummer KG, Vogel JG, Mack MC (2007) Plant species composition and productivity following permafrost thaw and therm okarst in alaskan tundra. Ecosystems, 10, 280 292. Schuur EAG, Vogel JG, Crummer KG, Lee H, Sickman JO, Osterkamp TE (2009) The effect of permafrost thaw on old carbon release and net carbon exchange from tundra. Nature, 459, 556 559. Seppala M (1994) S now depth controls palsa growth. Permafrost and Periglacial Processes, 5 283288. Shaver GR, Bret Harte SM, Jones MH, Johnstone J, Gough L, Laundre J, Chapin FS (2001) Species composition interacts with fertilizer to control long term change in tundra produc tivity. Ecology, 82, 31633181. Sonnentag O, van der Kamp G, Barr AG, Chen JM (2010) On the relationship between water table depth and water vapor and carbon dioxide fluxes in a minerotrophic fen. Global Change Biology, 16, 17621776. Sturm M, Racine C, Ta pe K (2001) Climate change Increasing shrub abundance in the Arctic. Nature, 411 546 547. Tape K, Sturm M, Racine C (2006) The evidence for shrub expansion in Northern Alaska and the PanArctic. Global Change Biology, 12, 686 702. Tarnocai C, Canadell J G, Schuur EAG, Kuhry P, Mazhitova G, Zimov S (2009) Soil organic carbon pools in the northern circumpolar permafrost region. Global Biogeochemical Cycles, 23,
66 Vogel J, Schuur EAG, Trucco C, Lee H (2009) Response of CO2 exchange in a tussock tundra ecosyste m to permafrost thaw and thermokarst development. Journal of Geophysical Research Biogeosciences, 114 Vourlitis GL, Oechel WC, Hope A et al. (2000) Physiological models for scaling plot measurements of CO2 flux across an arctic tundra landscape. Ecologic al Applications, 10, 60 72. Wahrhaftig C (1958) Quaternary geology of the Nenana river valley and adjacent parts of the Alaska range. U.S. Geological Survey, 1 68. Walker MD, Walker DA, Welker JM, et al. (1999) Long term experimental manipulation of win ter snow regime and summer temperature in arctic and alpine tundra. Hydrological Processes, 13, 23152330. Welker JM, Fahnestock JT, Henry GHR, O'Dea KW, Chimner RA (2004) CO2 exchange in three Canadian High Arctic ecosystems: response to long term experim ental warming. Global Change Biology, 10, 19811995. Yoshikawa K, Hinzman LD (2003) Shrinking thermokarst ponds and groundwater dynamics in discontinuous permafrost near Council, Alaska. Permafrost and Periglacial Processes, 14, 151160. Zhang T (2005) Influence of the seasonal snow cover on the ground thermal regime: An overview. Reviews of Geophysics, 43 Zhang T, Barry R, Knowles K, Heginbottom J, Brown J (1999) Statistics and characteristics of permafrost and groundice distribution in the Northern Hemisphere. Polar Geography, 23 132 154. Zimov SA, Davidov SP, Voropaev YV, Prosiannikov SF, Semiletov IP, Chapin MC, Chapin FS (1996) Siberian CO2 efflux in winter as a CO2 source and cause of seasonality in atmospheric CO2. Climatic Change, 33 111 120.
67 BIOGRAPHICAL SKETCH Christian Trucco attended high school in a small pour town in the North co ast of Colombia (South America). He then moved to the United States to learn English, before pursuing higher education. He attended Miami Dade College (Miami, Fl), and Santa Fe Community College (Gainesville, Fl), and got his associates degree in 2000. Christian later transferred to the University of Florida, and graduated with a Bachelor in Science in the School of Forest Resource and Conservation (2005). He then moved to Alaska to work as a field technician for Dr. Edward Schuur (20062008). After three years of work, Christian returned to the University of Florida in 2008 to pursue a Master of Science with Dr. Edward Schuur, within the same field of study that he worked on. He graduated with a Master of Science in May 2011. His next goal is to apply his expertise in the field of ecology to the building construction sector. Christian is planning to obtain a second Master of Science in Building Construction at the University of Florida.