Forest Floor Carbon Exchange of a Boreal Black Spruce Forest in Eastern Canada

Forest Floor Carbon Exchange of a Boreal Black Spruce Forest in Eastern Canada O. Bergeron, H. A. Margolis, and C. Coursolle Centre d’étude de la forêt, Faculté de foresterie, de géographie et de géomatique, Université Laval, Québec, Canada now at: Department of Natural Resource Sciences, McGill University, Montréal, Québec, Canada Received: 14 April 2009 – Accepted: 14 May 2009 – Published: 5 June 2009 Correspondence to: O. Bergeron (onil.bergeron@mcgill.ca) Published by Copernicus Publications on behalf of the European Geosciences Union.


Introduction
Total ecosystem respiration (R e ) is a major determinant of the carbon (C) balance of northern forests (Valentini et al., 2000).R e includes respiration by above-ground plant parts (boles, branches, twigs, and leaves) and soil.Soil respiration (R s ) is a dominant component of C exchange in boreal ecosystems, accounting for at least half of R e (Black et al., 2005).The temporal variability of respiratory metabolism is influenced mostly by temperature and humidity conditions (Davidson et al., 1998;Gaumont-Guay et al., 2006a).Above-and below-ground processes contributing to R e can respond in different ways to the seasonal variation of air and soil temperature, to the availability of water and to substrate type (Davidson et al., 2006;Jassal et al., 2007).
Coniferous boreal forests typically have relatively open canopies that allow a significant portion of incoming radiation to reach the ground vegetation (Baldocchi et al., 2000).Forest floor photosynthesis is thus a potentially important process that can represent up to 50% of C assimilation in such ecosystems (Goulden and Crill, 1997).The productivity of the forest floor depends on favourable light, temperature and moisture conditions (Swanson and Flanagan, 2001;Kolari et al., 2006) and also on the composition and relative presence of different forest floor communities in the ecosystem (O'Connell et al., 2003;Heijmans et al., 2004).
The boreal forest in eastern Canada is subjected to climatic conditions that differ significantly from those of other boreal regions in North America and Eurasia.For example, precipitation is generally more abundant in this region than in central Canada, while air temperatures are much cooler than in boreal Scandinavia.Also, the markedly different latitudes of boreal forests in Scandinavia and Canada lead to different light regimes in regards to both photoperiod and intensity.Such differences in climatic conditions can influence the C exchange of boreal forest ecosystems (Bergeron et al., 2007).
Furthermore, climate change is expected to have different regional consequences in northern forests (Flannigan et al., 2001).Thus, it is important to characterize the response of C exchange to environmental conditions in the eastern Canadian region of Introduction

Conclusions References
Tables Figures

Back Close
Full the circumpolar boreal forest.
Information on the seasonal contribution of soil respiration to ecosystem respiration is still very limited (e.g., Davidson et al., 2006) since values have been primarily reported on an annual or growing season basis (e.g., Lavigne et al., 1997;Law et al., 1999;Janssens et al., 2001;Gaumont-Guay et al., 2006b).Furthermore, there is apparently little information available for the vast area of cool humid boreal forest that is characteristic of eastern North America.On the other hand, photosynthesis of the moss stratum has been studied in different ecosystems (e.g., Swanson and Flanagan, 2001;Heijmans et al., 2004;Botting and Fredeen, 2006;Kolari et al., 2006).However, the extent to which it might contribute to the interannual variability in gross ecosystem productivity has apparently not been studied for boreal forests of eastern North America.Moreover, the relative contribution of different boreal micro-sites (sphagnum, feathermoss, lichen) has not been well explored.Such information is important for partitioning the different C sinks within an ecosystem and characterizing their specific response to environmental conditions.Numerous automated chamber systems have been developed to produce continuous measurements that help gain insight into the seasonal variability of soil respiration and forest floor photosynthesis (e.g., Goulden and Crill, 1997;Pumpanen et al., 2001;Gaumont-Guay et al., 2008).Coupling automated chamber measurements with eddy covariance (EC) measurements can provide valuable information on the concomitant response of respiratory and photosynthetic processes of different ecosystem components to environmental variables at fine time scales.However, due to logistical and equipment constraints, automated chamber systems are typically confined to a small portion of the footprint measured by an EC tower.This can lead to a mismatch in the source area, resulting in an apparent discrepancy between measurement methods (Drewitt et al., 2002).This source of error can be controlled by assessing the spatial representativeness of soil respiration using a portable manual chamber system (Savage and Davidson, 2003) and by limiting ecosystem C exchange estimates to specific sectors of the tower footprint (e.g., Yuste et al., 2005)  This study presents continuous automated measurements of ecosystem and forest floor CO 2 exchange for the snow-free period over three years in a boreal black spruce forest ecosystem in eastern North America.The objectives of this study were to (1) quantify the relationship of soil respiration and photosynthesis of different forest floor micro-sites (sphagnum, feathermoss and lichen) to environmental factors and (2) contrast the seasonal contribution of soil respiration and forest floor photosynthesis to that of the entire ecosystem.The spatial representativeness of automated soil respiration measurements in the tower footprint area was also assessed.

Site description
Our study site (Eastern Old Black Spruce, EOBS; 49.692 • N, 74.342 • W) lies in the commercial boreal forest of Canada and is located about 30 km south of Chibougamau, QC.The study site corresponds to a 500 m radius centered on the tower where at least 90% of the flux footprint originates (Bergeron et al., 2007).Black spruce (Picea mariana) dominates the site and there are sparse jack pine (Pinus banksiana) and tamarack (Larix laricinia).The shrub stratum is comprised of sheep laurel (Kalmia angustifolia) and Labrador tea (Rhododendron groenlandicum) on dry micro-sites and alder (Alnus rugosa) on wet micro-sites.The forest floor is covered by feathermoss (Hylcomnium splendens, Pleurozium schreberi), sphagnum (Sphagnum spp.), and lichen (Cladina spp.; see Table 1).Most of the study area originates from fire disturbance that occurred between 1885 and 1915.EOBS is dominated by podzol soils with 15-40 cm organic layers lying on silty-sand parent material.Mean tree height is 13.8 m, mean DBH is 12.7 cm, tree density is 4,490 stems ha −1 , black spruce basal area is 22.8 m 2 m −2 , and hemispherical LAI is 3.7 m 2 m −2 .The 30-year average of mean annual temperature and total annual precipitation measured at the nearest weather station (15 km NW of the site) are 0.0

Automated measurements of forest floor CO 2 exchange
Soil CO 2 efflux, including daytime forest floor photosynthesis, was measured continuously during the 2004-2006 snow-free periods by a non-steady state, automated chamber system manufactured by the Biometeorology and Soil Physics Group (University of British Columbia, Vancouver, BC, Canada; see Gaumont-Guay et al. (2008) for a complete description).The system is comprised of three temperature-controlled housings enclosing data logging, pumping and gas measurement equipment, as well as 6 to 9 chambers.The chambers are made of a clear acrylic dome fixed with a hinge to a 13-cm high PVC collar inserted 8-12 cm in the forest floor.The chambers are about 50 L in volume and cover an area of 0.216 m 2 .A 50-cm long venting tube is inserted into the top of each chamber to allow pressure equalisation.A pneumatic system controls the opening and closing of the chambers.All chambers are deployed in a 15 m radius around the main equipment stand which is located about 80 m south of the flux tower.In June 2004, a total of six chambers were installed on feathermoss (3 chambers), lichen (2), and sphagnum (1).In June 2005, three more chambers were set up on feathermoss (2) and sphagnum (1).Shrubs were excluded from the collars.At the time of measurement, chambers were closed for 2.5 min or 3 min (when 6 or 9 chambers were in use, respectively), otherwise the chamber stayed in the open position (83 or 92% of the time based on a 15 or 30 min cycle, respectively).Air was sequentially circulated at 9 L min 1 through ∼35 m of tubing (Synflex 1300, 4.0 mm Internal Diameter, Saint-Gobain Performance Plastics, Wayne, NJ, USA) between the chambers and an infrared gas analyser (IRGA) (model LI-6262, LI-COR Inc., Lincoln, NE, USA).The IRGA was calibrated daily using the same procedure as the EC system (see below).
Data were sampled at 1 Hz and averaged every 5 s.
Soil CO 2 efflux (F cs ) was calculated on a half-hour basis using the equation where ρ a is the density of dry air in the chamber headspace (mol m −3 ), V e the effective volume of the chamber (m 3 ), A the area of ground covered by the chamber (m 2 ) and ds c /dt the time rate of change of the CO 2 mixing ratio in the chamber head space over a 60-sec interval beginning 5-7 s following lid closure (mol CO 2 mol −1 dry air s −1 ).V e was estimated daily using the dilution technique described in detail in Drewitt et al. (2002) and Gaumont-Guay et al. (2006a).A 5-day running mean with a one-day increment was computed for V e to minimize day-to-day variation.Multiple F cs measurements performed in the same half-hour were averaged.

Manual soil respiration measurements
In June 2004 and May 2005, 45 white PVC collars (height=10 cm; diameter=9.6 cm) were installed in a 80 m by 80 m systematic grid, covering about half of the study area (roughly 400 m×800 m) and corresponding to the portion of the footprint most often upwind from the flux tower.Collars were inserted about 8 cm deep in the forest floor and shrubs were excluded from the collars.Manual soil respiration (R s−man ) was measured on a monthly basis during the 2005 snow-free season using a LI-6400 portable system (LI-COR Inc.) coupled to a LI-6400-09 soil chamber (volume=991 cm 3 ; diam-eter=9.55cm; ground area exposed=71.6 cm 2 ).R s−man measurements were made according to the soil chamber manual.More specifically, ambient CO 2 concentration was first measured by laying the chamber on its side on the ground and the LI-6400 was set to measure CO 2 concentrations 2 to 10 ppm (or higher for greater fluxes on rare occasions) around the ambient concentration.The chamber was then placed on the collar and allowed 10-20 s to equilibrate.Then, three measurement cycles were performed and only the last two measurements were averaged for further analysis.
The distance between the forest floor and the top of the collar was measured to calculate the actual chamber volume during post-processing.The whole measurement sequence was completed within 10 min at each location.Measurements were made between 09:00 and 20:00 (LT) during one single day and measurements began at a Introduction

Conclusions References
Tables Figures

Back Close
Full different location for each measurement date in a latin square fashion as suggested by Davidson et al. (2002).

Ecosystem CO 2 flux measurements
CO 2 flux at the ecosystem level was measured half-hourly following the EC technique as described in detail in Bergeron et al. (2007).A closed-path IRGA (model LI-7000, LI-COR Inc.) enclosed in a thermostatic box (37.5±0.5 • C) was coupled to a 3-D sonic anemometer-thermometer (model CSAT3, Campbell Scientific Canada corp.(CSC), Edmonton, AB, Canada) to make 10 Hz measurements of CO 2 concentration and vertical wind velocity at a height of 24 m.The IRGA was calibrated daily by injecting dry, CO 2 free nitrogen and an air/CO 2 gas mixture with a CO 2 concentration of approximately 370 ppm (0.001 ppm precision, traceable to NOAA/CMDL standards).Net ecosystem exchange was computed as the sum of CO 2 flux at the ecosystem level and CO 2 storage in the air column below the EC sensors measured with a 5-height profile system.

Ancillary climate measurements
Air temperature (T a ) and vapour pressure deficit (VPD) were measured at a height of 24 m with a shielded thermistor and humidity sensor (model HMP45C, CSC).Soil temperature at 5 (T s5cm ), 50 (T s50cm ) and 100 cm (T s100cm ) and soil water content at 5 cm (SWC) below the active moss layer were measured in two soil pits using thermistors and shared the same quantum sensor.Wind direction was monitored at a 24 m height with a wind monitor (Model 05103-10, RM Young, Traverse City, MI, USA).

Data analysis
For each chamber of the automated system, soil respiration (R s−auto ) and photosynthesis of the forest floor (P f f ) were calculated for the snow-free season using a modified version of the Fluxnet-Canada Research Network (FCRN) standard partitioning and gap-filling algorithm described in detail in Barr et al. (2004).In our study, soil respiration refers to CO 2 efflux from autotrophic and heterotrophic respiration originating from below-ground as well as autotrophic respiration from ground cover plants.
First, an exponential temperature function (Eq.2) was fit to nighttime (mean chamber PAR 30 cm <5 µmol m −2 s −1 ) F cs data to estimate daytime and missing half-hourly R s−auto as follows: where Q 10 (= exp(10B)) is a temperature sensitivity coefficient and R 10 (= Q 10 exp(A)) is base soil respiration at 10 • C (µmol m −2 s −1 ).A time varying factor was calculated as the regression slope between measured and predicted values using a moving window (100 good measurement points, increment of 20) to adjust for any seasonal variability of the temperature response of R s−auto .T s5cm measured in the soil pit under feathermoss served to estimate daytime R s−auto of chambers installed on feathermoss or sphagnum, while T s5cm measured in the lichen soil pit was used for automated chambers over lichens.The response of R s−auto to T s5cm as presented in Table 3 was characterized using a log transformed Q 10 function (Eq. 3) as presented by Morgenstern et al. (2004).
The transformation provides homoscedasticity to perform linear least squares regression.Equation ( 3) was also used to characterize the response of R s−auto to T s5cm , to Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc

Printer-friendly Version
Interactive Discussion calculate R s−auto normalized for soil temperature (R s−auto /R s−auto (T s5cm )) and to assess the influence of T a , T s−50cm and SWC on R s−auto .In all cases, only nighttime non gap-filled measurements were used.P f f was then calculated as daytime F cs -R s−auto for each chamber.The P f f record was gap filled using a rectangular hyperbola function where α is the apparent quantum yield and P ffmax is the horizontal asymptotic value of P f f .As for R s−auto , a time varying adjustment factor (moving window) was included.R s−auto and P f f were then averaged by cover type.Equation ( 4) was also used with non gap-filled data to characterize the response of P f f to PAR 30 cm and calculate P f f normalized for light (P f f /P f f (PAR 30 cm )) to assess the influence of T a , VPD and SWC on P f f .For three 2-to-6-day periods in 2005 (see Fig. 1 caption), all automated chambers were darkened using Lumite shade fabrics (Synthetic Industries, Gainesville, GA, USA) to measure daytime R s−auto .Figure 1 presents the relationship of measured to estimated daytime R s−auto using the modified partitioning algorithm described above.Since both variables contain errors due to the measurement techniques, we used geometric mean regression as presented in Jassal et al. (2007).Daytime estimates agreed well with measurements and allowed us to calculate reliable estimates of daytime R s−auto and P f f .Estimates of soil respiration and forest floor photosynthesis scaled-up to the ecosystem level (R s−eco and P ff−eco , respectively), were computed as a weighted average of R s−auto and P f f , respectively, based on the surface area of each ground cover for the three main cover types in the study area (Table 1).In 2006, one chamber on feathermoss showed spurious results and its measurements were thus discarded for that year.Also, the IRGA yielded unstable measurements in September and October 2006, which led to the exclusion of automated chamber data for this period.Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion R s−man was aggregated by sector and for the whole study area.Sectors were delimited by azimuth angles from the flux tower to include approximately the same number of measurement points per sector (Table 1).Each sector included collars located 120 to 450 m from the tower.
The EC record was quality controlled as described in Bergeron et al. (2007).Nighttime data under calm conditions (u * <0.25 m s −1 ) were discarded.Net ecosystem exchange was partitioned into total ecosystem respiration (R e ) and photosynthesis (P eco ) using the FCRN standard partitioning and gap-filling algorithm (see Bergeron et al., 2007 for details).EC data were corrected for the lack of closure in the energy budget (82% closure), as suggested by Barr et al. (2006).
Monthly totals of R s−eco , P ff−eco , R e , and P eco were obtained by summing gap-filled values.Regressions were performed using SAS (version 9.1; SAS Institute Inc., Cary, NC, USA), SigmaPlot (version 8.02; SPSS Inc. Chicago, IL, USA) and/or MatLab (version 7.3.0;The MathWorks Inc., Natick, MA, USE) and its curve fitting toolbox (version 1.1.6).

Adjustment of R s−eco for spatial representativeness
Manual soil respiration measurements made within three sectors of the tower footprint showed the same general seasonal patterns as automated measurements made in the immediate vicinity of the flux tower (Fig. 2).Nonetheless, total gap-filled R s−eco was compared to total R s−man for the entire snow-free season in 2005 (16 June to 4 October) to assess the spatial representativeness of scaled-up automated soil respiration measurements.R s−eco totals were calculated using daily averages from (1) only nighttime measurements and (2) only daytime measurements, multiplied by 48, and (3) from the sum of both nighttime and daytime measurements to test for the effect of time of measurement.We did this since R s−man was measured during the daytime while daytime R s−auto (thus R s−eco ) was estimated using the partitioning algorithm described above.Total R s−man was estimated using two common techniques, i.e. linear interpolation between sampling dates (e.g., Davidson et al. 2006) and derivation from an Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion exponential temperature function (e.g., Law et al., 1999).R s−eco systematically overestimated soil respiration by 2 to 28% as compared to R s−man regardless of the period of the day used for the calculation (Table 2).As well, total soil respiration estimated from R s−man diverged between summation techniques by less than 10 g C m −2 for the 80-day period, corresponding to an uncertainty of less than 3% (Table 2).Hence, the method chosen to estimate snow-free season totals of soil respiration from manual measurements did not account for the higher total respiration derived from the automated measurements.Therefore, the average of all the ratios of automated to manual measurements for each sector presented in Table 2 (S: 1.24; SW: 1.04; NW: 1.20;All sectors: 1.16) was used as a correction factor to empirically adjust the spatial representativeness of our time series of automated soil respiration measurements.R s−eco adjusted for spatial representativeness (R s−adj ) was computed by decreasing each half-hour measurement of R s−eco by the correction factor corresponding to the appropriate sector.Half-hours when wind direction was from outside the three sectors or when the footprint length did not match the source area defined above were corrected using a correction factor averaged for all three sectors.The footprint length was calculated using an inverse Lagrangian model (Kljun et al., 2004).3) ranged from 3.22-4.36for feathermoss, 3.54-4.42for lichen, and 3.33-4.04for sphagnum micro-sites (Table 3) for soil temperatures varying between 0 and 16 • C (data not shown).These estimates are within the range of reported values for other boreal forest soils (e.g.Davidson et al., 1998;Rayment and  Jarvis , 2000;Gaumont-Guay et al., 2006a, 2008).Q 10 values of sphagnum micro-sites were lowest overall in 2005 and 2006 and were consistently lower than lichen microsites during all three years.Q 10 values estimated on a growing season basis represent the temperature sensitivity of enzymatic activity and other temperature-dependant processes (Davidson et al., 2006) but also include phenological effects (e.g.root growth stage) and microbial population shifts (Janssens and Pilegaard, 2003;Yuste et al., 2004).Hence, our results suggest that the dynamics of the processes influencing the temperature sensitivity of soil CO 2 efflux are to some degree affected by forest floor vegetation type, which is, to some extent, a reflection of underlying soil properties.
Base respiration (R 10 ) was consistently lowest under feathermoss, intermediate under sphagnum, and highest under lichen (Table 3).Accordingly, R s−auto was highest under lichen, intermediate under sphagnum, and lowest under feathermoss at any given soil temperature (Fig. 3a-c).The spatial variability of soil respiration has been related to the physical (micro-topography, porosity, organic horizon depth, temperature, humidity), chemical (nutrient status of mineral and organic horizons, organic matter quantity and quality) and biological (microbial and fine root biomass, microbial community composition) properties of the soil which can influence either the production of CO 2 , its transport to the surface or both (Fang et al., 1998;Longdoz et al., 2000;Rayment and Jarvis, 2000;Xu and Qi, 2001;Heijmans et al., 2004;Khomik et al., 2006;Saiz et al., 2006).These properties are linked to micro-site structural characteristics that are in turn related to the distribution and composition of mosses and lichens (Bisbee et al., 2001;Sulyma and Coxson ,2001).Our results provide evidence that the heterogeneity of the ground cover vegetation, which can represent the spatial variability of soil properties, should be taken into account when characterizing or simulating the response of soil CO 2 efflux to environmental factors.
Shallow soil temperature explained 67-86% of the temporal variation of R s−auto under all ground cover types (Table 3, Fig. 3a-c).R s−auto normalized for soil temperature (R s−auto /R s−auto (T s5cm )) showed a significant (p<0.0001)positive linear correlation with air temperature for all three ground cover types (r 2 =0.01-0.07 for feathermoss, 0.02-

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion 0.13 for lichen, and 0.18-0.28for sphagnum, Fig. 3d-f).Furthermore, normalized R s−auto of feathermoss micro-sites exhibited a significant (p<0.05),but very weak, positive correlation with deep soil temperature (T s50cm ) in 2004 and 2005 (r 2 =0.04 for both years, Fig. 3g-i).Temperature has been demonstrated to exert a major influence on soil respiration (Singh and Gupta, 1977;Raich and Schlesinger, 1992;Lloyd and Taylor, 1994).Studies have also shown that most soil respiration occurs in the upper soil layers in northern forest ecosystems (Drewitt et al., 2005;Jassal et al., 2005).These results again emphasize the need to account for different ground cover vegetation types in soil and ecosystem C exchange studies as they may reflect the spatial variability of soil properties and the distribution of respiratory processes along the soil profile.

Response to substrate moisture
Substrate moisture limitation on R s−auto was apparent only under sphagnum when soil water content near the surface reached values below 0.10 m 3 m −3 in 2005 (Fig. 3j-l).
Soil moisture has been reported to affect the soil respiration of temperate and boreal forest ecosystems (Davidson et al., 1998;Subke et al., 2003;Gaumont-Guay et al., 2006a).However, Gaumont-Guay et al. ( 2008) also reported no effect of soil moisture on soil respiration for a boreal black spruce site in Saskatchewan.This Saskatchewan site is located in a topographic depression where the water table and soil water content are generally high.Our site is less hydric than the site in Saskatchewan but the wetter climate prevailing in eastern Canada can induce more frequent rainfall throughout the growing season, in addition to larger amounts of water released by snowmelt, compared to ecosystems located in the drier regions of central Canada (Bergeron et al., 2007).As a result, our soils stayed relatively moist, at least in deeper soil horizons (data not shown), thereby partly explaining the lack of a relationship between soil respiration and soil humidity under feathermoss and lichen.Also, root respiration is potentially less sensitive to the drying of superficial soil layers than soil organic matter decomposition (Gaumont-Guay et al water that is not available to decomposers in the upper soil layer.In addition, microhabitats dominated by sphagnum in boreal black spruce forests are less favourable for tree root growth as they are associated with wetter and colder conditions (Bisbee et al., 2001), hence microbial decomposition presumably contributes more to soil respiration under sphagnum micro-sites.As a result, it is possible that the microbial populations on sphagnum microsites were more sensitive to soil moisture than those on feathermoss or lichen microsites The differences we observed in the response of R s−auto to SWC between ground cover types may therefore reflect different contribution levels of autotrophic (root) and heterotrophic (microbial) respiration between micro-sites.

Contribution of soil respiration to ecosystem respiration
During the study period, monthly values of ratios of R s−eco , not adjusted for spatial representativeness, to R e ranged from 82-120%, with values exceeding 100% on several occasions (Table 4), suggesting our scaling method overestimated soil respiration at the ecosystem level.On the other hand, adjusted R s−eco (R s−adj ) to R e ratios ranged from 72% to 103% during the snow-free season on a monthly basis and exceeded 100% on only one occasion.These results emphasize the importance of assessing the spatial representativeness of automated soil respiration measurements when scaled up to the ecosystem level to help resolve the mismatch in source area between chamber and EC measurements and thus produce comparable estimates.The ratios of R s−adj to R e on a snow-free season basis ranged from 85-87% (Table 4) over the study period and are within the range, but near the upper end, of other published values for Canadian boreal forest sites.Soil respiration accounted for 48-71% of total ecosystem respiration for six coniferous boreal sites (Lavigne et al., 1997), 70% in a boreal aspen forest (Gaumont-Guay et al., 2006b), and 80, 67, and 83% for boreal aspen, black spruce, and jack pine sites, respectively (Black et al., 2005).In other ecosystems, the soil to ecosystem respiration ratio was found to be 67% in a temperate mixed forest (Yuste et al., 2005), 62% in a costal Douglas-fir stand (Jassal et al., 2007), 76% in a ponderosa pine forest (Law et al., 1999), and 69% in European Introduction

Conclusions References
Tables Figures

Back Close
Full forests (Janssens et al., 2001).The somewhat higher R s−adj to R e ratio reported in our study, as compared to sites in the western Canadian boreal forest, could be a result of regional differences within the boreal forest in terms of the various component contributions to total ecosystem respiration.Bergeron et al. (2007) showed that total ecosystem C exchange differs between our eastern Canadian site and two other black spruce sites located in central Canada.This difference was partly attributed to higher soil temperatures at 5 and 50 cm.The higher winter temperatures were attributable to a thicker snowpack, at the eastern Canadian black spruce site and led to higher winter soil respiration, which may explain the proportionately higher contribution of soil respiration to total ecosystem respiration at this site.The eastern site also has a greater below-ground biomass (Bergeron et al., 2007) compared to the western coniferous sites which may have led to proportionately higher soil respiration as compared to ecosystem respiration at this site.Furthermore, Bergeron et al. (2007) also observed a lower water table at the eastern site during the second half of the growing season, which may have led to higher soil respiration rates due to the absence of anaerobic conditions.They also noted that soil moisture helped explain anomalies in the response of ecosystem respiration to temperature between sites at the monthly time scale, with our eastern site at the low end of soil water content.Thus, it is possible that the contribution of soil respiration to ecosystem respiration was greater at the eastern site.This finding, and the fact that little to no soil respiration restriction due to low soil moisture was found in our study, suggests that regional differences may exist in tree root phenology and physiology and/or in microbial community composition and dynamics and help explain the high soil to ecosystem respiration ratios reported here.However, further study is needed to shed light on these regional differences.It is also possible that the lower above-ground biomass at the eastern site or its associated respiration could help explain the regional differences in the soil to ecosystem respiration ratio.
The R s−adj /R e ratio had a significant negative linearly correlation with the difference between air and soil temperature for the May-June, July-August and September-Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version

Interactive Discussion
October periods using daily values (Fig. 5).Furthermore, the R s−adj /R e ratio calculated from 5-day means tended to increase throughout the snow-free season, showing values of approximately 60% in spring and close to 100% in autumn (Fig. 4).On a monthly basis, a similar general increase was also observed (Table 4).Few studies have described in detail the seasonal variation of the soil to ecosystem respiration ratio.Black et al. (2005) reported that the soil to ecosystem respiration ratio of a boreal black spruce site was greater in mid-summer and in winter and reached a minimum in spring and early summer.Davidson et al. (2006) showed that the soil to ecosystem respiration ratio was minimal in spring and increased to a maximum in autumn and winter for a temperate deciduous forest.Jassal et al. (2007) also showed a spring to autumn increase for a coastal coniferous forest.This seasonal variation was related to the different phenologies of above-and below-ground ecosystem components, to variations in substrate supply, and to lags between changes in air and soil temperature.
It is worth noting that the R s−auto /R e ratio close to one in autumn reported here returns to a much lower value by the following spring.This reset likely occurs in springtime when warm air stimulates above-ground respiration and the cold soil limits root growth and organic matter decomposition (Davidson et al., 2006), thus limiting soil CO 2 efflux and decreasing the contribution of soil respiration to total ecosystem respiration to a minimum.

Response to light
Maximum photosynthetic capacity (P ffmax ) was higher for feathermoss than sphagnum in 2004 and 2005 (Table 5).P ffmax was lowest for lichen in 2005 and 2006 but highest in 2004.Feathermoss consistently showed lower photosynthetic apparent quantum yield (α) values than sphagnum (Table 5).The same was true for lichen except in 2005 when α was not significantly lower than for sphagnum.Feathermoss and lichen showed significantly different α only in 2004.Introduction

Conclusions References
Tables Figures

Back Close
Full These results contrast with other published values of light response curve parameters.Swanson and Flanagan (2001) reported higher maximum photosynthetic capacity and quantum yield for sphagnum compared to feathermoss from measurements made at an old black spruce site in Saskatchewan.Goulden and Crill (1997) also observed higher maximum photosynthetic capacity values for sphagnum than feathermoss at the Northern Old Black Spruce flux site in Manitoba.In both studies, ecological differences between sphagnum and feathermoss micro-sites were greater than in the present study.At these central Canadian sites, sphagnum occupied open, wet microsites in lower elevations (hollows) where the water table is near or at the surface, while feathermoss was found in shady, dry upland areas (hummocks) where the water table depth is greater.Furthermore, black spruce ecosystems in central Canada are associated with wet sites where small topographic variations can have a large influence on vegetation composition (Trumbore and Harden, 1997).In eastern Canada, black spruce ecosystems are commonly found on less hydric sites where micro-habitat conditions do not differ as much with microtopography, as was the case at our site.
The micro-site differences in light and water table regimes described for central Canadian sites were much less pronounced at our site, hence the differences among the studies in question could be attributed to different environmental conditions between micro-sites.Therefore, the photosynthetic response of the ecosystem ground cover to environmental conditions appears to depend on the interaction between ground cover type (feathermoss, sphagnum and lichen) and microhabitat environmental conditions.
The response of lichen photosynthesis to light is not well documented and specific information about the physiology of Cladina spp. is rare.Lichens are considered to have photosynthetic rates similar to bryophytes when hydrated (Green and Lange, 1994).Coxson and Wilson (2004) reported values of maximum photosynthesis per mass unit of Cladina mitis similar to those reported for feathermoss and sphagnum.Our results provide evidence that bryophytes and lichens have photosynthetic rates per unit ground area that are similar in range.However, our study shows that lichen has distinct light response curve parameters and provides values for these parameters that can be used

Conclusions References
Tables Figures

Back Close
Full to simulate the photosynthetic response of lichen to light.The photosynthesis of feathermoss, sphagnum and lichen saturated at relatively low irradiance values (∼200 µmol m −2 s −1 ; Fig. 6a-c) that correspond to the maximum daytime irradiance below the canopy for most days of the snow-free season (data not shown).This was expected, since bryophytes and lichen would likely be well adapted to shade conditions (Green and Lange, 1994).The saturating light levels observed here are consistent with those measured by Whitehead and Gower (2001) and Kolari et al. (2006) for feathermoss and with those reported by Swanson and Flanagan (2001) for sphagnum.For lichen photosynthesis, Coxson and Wilson (2004) reported lower saturating light levels that decreased with temperature.Kolari et al. (2006) observed that light levels at the forest floor of a Scots pine forest in Finland were high enough for photosynthesis to saturate most of the time.On the other hand, Bisbee et al. (2001) suggested that the photosynthetic production of the forest floor is light limited.Given that P f f reached saturation at low light levels that are typical of the understory environment on our site, our results support the idea that factors other than light (i.e., temperature and/or water stress) limit the photosynthesis, and thus the carbon uptake, of forest floors composed of sphagnum, feathermoss or lichen.Furthermore, increased light availability due to non-stand replacing disturbances (e.g., windthrow, canopy dieback) generally increases the abundance of understory vascular plants thus leaving unchanged the light regime at the forest floor (Hart and Chen, 2006).

Response to other environmental variables
The response of P f f normalized for light (P f f /P f f (PAR 30 cm )) to air temperature and vapor pressure deficit was similar for all three ground cover types (Fig. 6d-i).P f f /P f f (PAR 30 cm ), peaked at T a =5-8 • C and decreased at temperatures below and above this range, suggesting that forest floor photosynthesis is optimal at temperatures ranging from 5-8 • C.These results are consistent with those of Goulden and Crill (1997) for feathermoss and sphagnum.Coxson and Wilson (2004) reported an optimal temperature for Cladina mitis photosynthesis of about 15 than reported here.Normalized P f f of sphagnum showed a sharper decline at temperatures above 8 • C compared to feathermoss and lichen, decreasing from 1.4 to 0.6, while normalized P f f of feathermoss and lichen decreased from 1.2 to 0.8.This indicates that light response curves can overestimate sphagnum photosynthesis by up to 40% under high air temperature conditions while this overestimation is limited to 20% for feathermoss and lichen.Sphagnum also showed a stronger decrease of normalized P f f when VPD values were above 1 kPa and P f f was overestimated by as much as 50% for sphagnum (normalized P f f =0.5) compared to about 25% for feathermoss and lichen (normalized P f f =0.75).No restriction of P f f was observed at low SWC for any of the three ground cover types (Fig. 6j-l).
Air temperature and vapor pressure deficit can be viewed as surrogates for desiccation as opposed to SWC that may not reflect the desiccation status of the ground cover vegetation (Fig. 6j-l).Bryophytes and lichen are poikilohydric plants and the reduction of their photosynthetic capacity under desiccating conditions is well documented (Green and Lange, 1994;Williams and Flanagan, 1996;Schipperges and Rydin, 1998).Our results suggest that sphagnum photosynthesis is more sensitive to desiccation than feathermoss and lichen.More direct measurements of the vegetation water status would likely help us better understand and predict bryophyte and lichen photosynthesis.

Contribution of P f f to P eco
The mean P f f /P eco ratios on a snow-free season basis were consistent between years at 0.17-0.18(Table 4).These results agree with published values for mature black spruce ecosystems in other regions of Canada.Swanson and Flanagan (2001) reported that moss photosynthesis accounted for 13% of gross ecosystem productivity (GEP) for the growing season.Goulden and Crill (1997) estimated the contribution of moss photosynthesis to GEP ranged between 10 and 50% on a daily basis with a greater contribution observed after rain events.In our study, the forest floor vegetation accounted for as much as 45% of daily ecosystem photosynthesis toward the end of 5526 Introduction

Conclusions References
Tables Figures

Back Close
Full the snow-free season (data not shown).Snow-free season totals of P f f were 23-24% that of R s−adj (Table 4).Drewitt et al. (2002) also measured a decrease in daytime soil CO 2 efflux of up to 25% due to moss photosynthesis.Mor én and Lindroth ( 2002) observed an offset of soil respiration of about 20% due to forest floor photosynthesis.
Our results demonstrate that C fixation by forest floor photosynthesis is a significant component of C exchange of boreal ecosystems and that mature black spruce ecosystems in the eastern portion of North America seem to obtain a similar proportion of their assimilated carbon from the forest floor compared to black spruce forests in other regions.
The contribution of P f f to P eco was not constant over the course of the snow-free season for any of the three years, varying from 13 to 24% (Fig. 7, Table 4).The seasonal variability of the P f f /P eco ratio was best related to changes in air temperature (r 2 =0.09,Table 6).This relationship, although weak, shows that the contribution of forest floor vegetation decreased as air temperature increased up to 20 • C and varied little for further increases up to 30 • C (data not shown).This observed temperature limitation of the P f f /P eco ratio is consistent with the limitation of photosynthesis at temperatures above 8 • C mentioned earlier (Fig. 6d, f).On the other hand, boreal forest GEP is dependent on light levels but also on air temperature (van Dijk et al., 2005;Bergeron et al., 2007;Fig. 8g-o) and thus shows a pronounced seasonal variation with maximum values reached during mid-summer when air temperature is high (Black et al., 2005, Fig. 8j-o).These two contrasting responses, along with the fact that the seasonal variability of P eco is almost an order of magnitude greater than that of P f f , can explain the seasonal pattern of the P f f /P eco ratio.

Conclusions
This study reports on C exchange of the forest floor within a boreal black spruce forest in eastern Canada and its seasonal contribution to ecosystem C exchange.We have shown that the response of soil respiration to environmental factors differs be-Introduction

Conclusions References
Tables Figures

Back Close
Full tween micro-sites.Soil respiration was a dominant component of ecosystem respiration during the study period, accounting for 85-87% of total ecosystem respiration on a snow-free season basis.However, the contribution of soil respiration to ecosystem respiration varied seasonally, ranging 72 to 103% from May to October and this variation was related to the difference between air and soil temperature.
This study also showed that C assimilation by the moss and lichen stratum can significantly impact ecosystem C exchange in the boreal forest of eastern North America, accounting for 17-18% of the total ecosystem C assimilation.The contribution of forest floor to ecosystem photosynthesis did not vary significantly between years but showed a pronounced seasonal variation, ranging 13 to 24% on a monthly basis and even more on a daily basis, indicating that the different vegetation strata do not respond similarly to environmental conditions.The three ground cover types showed some differences in photosynthetic responses to environmental conditions but light did not appear to limit photosynthesis of bryophytes and lichen during the snow-free season.
To better understand and predict the consequences of the modification of temperature and precipitation regimes under different climate change scenarios, process models could utilize the parameters and response functions described in this paper to better characterize the physiological processes governing C exchange of the soil and ground cover of boreal ecosystems in eastern North America.
(model 107, CSC) and reflectometers (model CS616, CSC), respectively.The two soil pits were located on different micro-sites, the first one under a relatively closed canopy with feathermoss covering the ground surface, the second one under a relatively open canopy with lichens on the ground surface.Photosynthetically active radiation was measured with quantum sensors (model LI-190SB, LI-COR Inc.) at a 24-m height (PAR 24 m ) on the tower and ∼30 cm above the forest floor (PAR 30 cm ) beside (<50 cm away) seven of the nine chambers.Three chambers were located within a 2-m radius 10 and base respiration Q 10 values derived from Eq. (

Fig. 1 .
Fig.1.Relationship between predicted (with the gap-filling algorithm) and measured daytime soil respiration (R s ) for three periods in 2004 when chambers were darkened (18-19 July , 27 August -1 September and 22-24 September).A log transformation was used on both x and y to provide homoscedasticity.Given the error associated with x, a geometric mean regression was used.

Fig. 2 .Fig. 4 .Fig. 5 .
Fig. 2. Time series of automated (Auto; daily mean of non-gap-filled nighttime measurements) and manual (Man) soil respiration (R s ) measurements for three sectors of the tower footprint.

Table 4 .
Monthly totals of R s−eco , R s−adj , R e , P ff−eco and P eco .

Table 4 .
Continued.−2 month −1 .Gap-filled values were used.See Methods for details.Introduction 1 Values are in g C m

Table 5 .
Parameter values derived from Eq. (4) relating P f f to PAR 30cm .

Table 6 .
Relationships of P f f /P eco with air and soil temperatures.