- Email: [email protected]
Changing temperature response of respiration turns boreal forest from carbon sink into carbon source
Agricultural and Forest Meteorology 223 (2016) 30–38
Contents lists available at ScienceDirect
Agricultural and Forest Meteorology journal homepage: www.elsevier.com/locate/agrformet
Changing temperature response of respiration turns boreal forest from carbon sink into carbon source David Hadden ∗ , Achim Grelle Department of Ecology, Swedish University of Agricultural Sciences, Sweden
a r t i c l e
i n f o
Article history: Received 8 September 2015 Received in revised form 23 March 2016 Accepted 30 March 2016 Available online 6 April 2016 Keywords: Eddy covariance Carbon dioxide exchange Temperature response Boreal forest Carbon balance Respiration
a b s t r a c t Seventeen years (1997–2013) of carbon dioxide (CO2 ) fluxes were measured in a boreal forest stand in northern Sweden using the eddy covariance technique. During the measurement period the forest turned from a net carbon sink into a net carbon source. The net ecosystem exchange (NEE) was separated using values from periods of darkness into the gross components of total ecosystem respiration (TER) and gross primary productivity (GPP), which was calculated as GPP = −NEE + TER. From the gross components we could determine that an increase in TER during the autumn (September to end of November) and spring (March to end of May) periods resulted in the forest becoming a net source of CO2. We observed no increase in the GPP from the eddy covariance measurements. This was further supported by measurements of tree growth rings. The increased TER was attributed to a change in the forest’s temperature response at lower temperatures (−5 to 10 ◦ C) rather than to a temperature increase. This study shows that changes in ecosystem functioning can have a larger impact on the carbon balance than climate warming per se. © 2016 Elsevier B.V. All rights reserved.
1. Introduction In 2013 the daily mean concentration of atmospheric carbon dioxide (CO2 ) surpassed 400 ppm globally (Showstack, 2013). Anthropogenic emissions of CO2 are on the rise, which is having a direct impact on global warming. With an increasing concentration of atmospheric CO2 the ability of terrestrial ecosystems to sequester carbon is of great importance. Global forests play a vital role in cycling CO2 between the atmosphere and terrestrial biosphere and are capable of sequestering 4.0 Pg C year−1 (Pan et al., 2011). The Boreal forest in particular has been acting as a carbon sink for thousands of years and contains some of the world’s largest soil carbon stocks (Harden et al., 1997). The net uptake of CO2 within the forest ecosystem is the difference between the gross primary productivity (GPP) and the total ecosystem respiration (TER). A small change in either of the gross fluxes is therefore crucial in determining whether a forest may act as a sink or source of CO2 (Lindroth et al., 1998). Air temperature has been shown to be one of the main drivers of both carbon uptake and ecosystem respiration (Barr et al., 2007; Bergeron et al., 2007). The onset of the growth period in the spring is largely regulated by air temperatures and has a direct impact on
∗ Corresponding author. E-mail address: [email protected] (D. Hadden). http://dx.doi.org/10.1016/j.agrformet.2016.03.020 0168-1923/© 2016 Elsevier B.V. All rights reserved.
net annual uptake of CO2 from the atmosphere (Krishnan et al., 2008; Tanja et al., 2003). An increase in air temperature however also increases ecosystem respiration which has been shown to result in a net loss of carbon from the terrestrial system during the autumn months (Piao et al., 2008; Ueyama et al., 2014; Vesala et al., 2010). Furthermore, there are indications that a change in ecosystem function has a large impact on the ecosystem carbon balance and dominates the interannual variations to a greater extent than climatic variability which appears to impact the carbon fluxes at short time scales (Wu et al., 2012). These findings have rendered it difficult to evaluate future consequences of climate change on the Boreal zone’s carbon balance. One of the keys to predicting future consequences of warming on the Boreal forest carbon balance is to understand the underlying processes that drive the forest’s response to temperature. Due to the stochastic nature of weather and climate combined with the long life cycle of the boreal forest and changes within ecosystem functioning, long term measurements are needed to gain a detailed understanding of the carbon cycle on a large temporal scale. We must also consider the forest dynamics and how these change as the forest ages when evaluating the carbon balance. There are relatively few datasets of continuous long term carbon fluxes from the boreal zone (more than 15 years in length) where the interannual variations of climate and fluxes can be evaluated. Most long term measurements have been done by the eddy covariance technique (e.g., Ge et al., 2011; Ueyama et al., 2014; Vesala
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
et al., 2010). There are numerous studies showing the impact that a changing climate may have on the boreal forest’s carbon balance. There are however no studies that we are aware of that evaluate a change in the forests functioning and the impacts that this has on the carbon balance with the prospects of a changing climate. The objectives of this study are to report the interannual variations of CO2 fluxes within the boreal forest ecosystem and identify the underlying processes controlling the fluxes. We present a 17 year record (1997–2013) of eddy covariance data measured in northern Sweden. During this period we observed a change in the ecosystem function resulting in the forest turning from a carbon sink to a carbon source. 2. Materials and methods 2.1. Study site The experimental site is located at Flakaliden in northern Sweden (64◦ 07 N, 19◦ 27 E, 310 m above sea level). Flakaliden has a boreal climate, experiencing long and cold, dark winters and short, cool summers with long daylight hours. The mean temperature during the measurement period was 2.4 ◦ C. The site is a managed forest on a shallow till soil that consists predominantly of Norwegian spruce (Picea abies) with an average age of ca. 50 years. The average tree height in 2013 was ca. 12 m with an average diameter of 13 cm. The forest stand had a leaf area index (projected LAI) of 3.40 as of September 2014.
31
18 m height using an LI 190 quantum sensor (LI-COR inc., Lincoln, Nebraska, USA). Leaf area index (LAI) was measured using an LAI-2000 plant canopy analyser (LI-COR inc., Lincoln, Nebraska, USA) on a regular basis throughout the measurement period. This was done by manual multi-point measurements at 10 m intervals in North, East, South and Westerly directions from the tower. 2.3. Temperature response There is a correlation between air temperature and total ecosystem respiration which has resulted in air temperature being largely regarded as a main driver for TER (Barr et al., 2007; Bergeron et al., 2007). Due to this we chose to study the response of TER to air temperature rather than soil temperatures. Furthermore, this study is based upon eddy covariance measurements which measure the carbon exchange at the ecosystem level. It could therefore be misleading to analyse the carbon exchange at the ecosystem level with only soil temperatures as a driver. Indeed, generally our CO2 fluxes correlated better with air temperature than with soil temperatures and we therefore chose to carry out our analysis of respiration response to air temperatures rather than soil temperatures. The gross respiration was calculated with the equation: TER = a + b · e c · T
(1)
with a representing the offset and b and c represent curve coefficients. 2.4. Growth period
2.2. Instrumentation Turbulent fluxes of CO2 were measured by an eddy-covariance system (In Situ Instrument AB, Ockelbo, Sweden), consisting basically of a solent 1012R2 sonic anemometer (Gill Instruments, Lymington, UK) and a closed path infrared gas analyzer LI-6262 (LI-COR inc., Lincoln, Nebraska, USA) as described by Grelle and Lindroth (1996). The eddy covariance instrumentation was mounted on an 18 m tall guyed mast with a triangular cross section of 28 cm width. The flux measurements were taken at 15 m height on a boom extending 2 m from the mast. Extensive flux source area analyses have been applied to confirm that the appropriate height for flux measurements was chosen to represent the surrounding forest both during day- and night-time (Grelle, 1997). To minimize damping of high-frequency concentration fluctuations in a long intake tube, the entire system was mounted at the tower close to measurement height. The tube inlet was placed 10 cm below the sampling volume of the sonic anemometer. Air was drawn from the inlet through a 6 mm diameter, 6 m long high-density polyethylene tube to the IRGA, which was mounted at the mast in a heated, ventilated, and insulated enclosure at a height of 13 m. The air flow rate was 12 ln/min (litre at 0 ◦ C, 1013 hPa), measured and controlled by a mass flow regulator (Brooks Instrument, Hatfield, PA, USA). The time lag between data samples from the anemometer and the gas analyser was continuously determined by autocovariance analysis on a half-hourly basis, and the signals were shifted accordingly prior to further analysis. Energy balance closure indicated adequate performance of the measurement system (Grelle, 1997). Overall calculation and correction of fluxes followed the EUROFLUX methodology (Aubinet et al., 2000) and Lee et al. (2004). Soil temperature was measured in two profiles from 5 cm to 50 cm depth by 10 type 107 thermistors (Campbell Scientific, Logan, UT, USA), and air temperature and -humidity by an MP103A sensor (ROTRONIC AG, Bassersdorf, Switzerland) placed in a ventilated radiation shield (In Situ Instrument AB, Ockelbo, Sweden) at 18 m height. Photosynthetically active radiation (PAR) was measured at
Following a widely used convention, the growth period is defined as starting when the daily average temperature is ≥5 ◦ C for 5 consecutive days and ending when the daily average temperature is <5 ◦ C for 5 consecutive days (e.g., Frich et al., 2002). 2.5. Tree growth Tree growth was measured using core samples taken from 40 trees. Each sample was taken at breast height and from the south side of the trunk. Each sample was photographed and the year rings were measured in width with the use of a pixel counting software (Pixel Ruler inc.). Each year ring was calculated into area as follows: Ai =
2 D i
2
−
D
i+1
2
2
(2)
Di = Di+1 + 2Wi i=
1, 2, 3, .., N
The index number of a given year ring, i, is counted from outside towards the pith of the stem. N is the total number of year rings measured. Ai is the portion of the whole stem area represented by the area of the year ring growing from i + 1 to i. Di+1 is the diameter of the whole stem area when growth of the year ring began. Di is the diameter of the whole stem area including the growth represented by the width, Wi , of the year ring produced from i + 1 to i. D1 is stem diameter under bark. 2.6. Gap filling After quality control, there was an average fraction of 24.9% gaps in the flux data and 6.4% in the climatic data throughout the 17 years of measurements. For periods when climatic data were available, gaps in the flux data were filled by a neural network (NNDT, Saxén
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
Gaps were filled in this fashion in order to limit a loss of any trend that may have been present within the time series data.
5 0
NEE (tonnes CO2 ha−1 yr−1)
1. Half-hourly gaps were filled by linear interpolation 2. Gaps of up to 2 h were filled with averaging 2 h before and after the gap 3. Gaps of more than 2 h up to 2 week in size were bin-averaged from corresponding period directly before and after the gap 4. Gaps of more than 2 weeks were filled using bin averages from previous and following years
10
& Saxén, 1995) using air temperature, PAR, and wind speed as input variables. When climatic data were missing, gaps in the fluxes were filled according to the following criteria:
−5
32
The separation of the NEE was done on a seasonal basis. December to the end of February was considered as winter. March to the end of May was considered as spring. June to the end of August was considered as summer. September to the end of November was considered as autumn. Flux data during periods of darkness, when photosynthesis effectively is zero, were used to estimate total ecosystem gross respiration (TER). We defined periods of darkness as half an hour after sunset until half an hour before sunrise. During well-mixed periods (friction velocity u∗ > 0.2 m s−1 , Aubinet et al., 2000), an exponential temperature response (eq. 1) was fitted to the measured ecosystem fluxes. This response was then used to estimate daytime respiration to complement night-time measurements, which resulted in total ecosystem gross respiration. We used air temperature as the dependent variable describing TER as an exponential function, since this had shown the best explanatory power for ecosystem respiration that is composed by autotrophic and heterotrophic respiration from biomass and soil. Based on this measured/modelled TER and measured NEE, the gross primary productivity (GPP) was then calculated as: GPP = −NEE + TER
−10
2.7. Partitioning of NEE into gross components (GPP and TER)
1997 1999 2001 2003 2005 2007 2009 2011 2013 Fig. 1. Annual Net Ecosystem Exchange (NEE) from 1997 to 2013. From 2007 to 2009 the forest ecosystem has had a declining net uptake followed by a net loss of C from 2010 to 2013.
3. Results During a 17 year period the forest ecosystem has swung from a carbon sink to a carbon source. From 1997 until 2009 – with the exception of 1998 when it was close to neutral – the forest was annually a continuous net carbon sink. The forest’s ability to sink carbon is seen to decline consistently from 2007 (−1.4 t C ha−1 y−1 ) until 2013 (1.2 t C ha−1 y−1 ) with the forest becoming a net carbon source in 2010 (Fig. 1). From 2010 until 2013 the forest ecosystem consistently had an annual loss of CO2 to the atmosphere. During both spring and autumn time there had been a trend towards increased TER, however no trend towards a change in GPP over the entire measurement period (Fig. 2). TER was seen to have a greater increase during the autumn period than the spring period (Figs. 2 and 6). This increasing trend in TER had a significant impact on NEE and resulted in the forest becoming a net carbon source on an annual basis from 2010. Tree growth on the other hand, had been consistent since 2004, while showing some annual variations from 1996 to 2003 (Fig. 3). This consolidates our finding that GPP was consistent (Fig. 2). Together with the change of NEE during the measurement period there has been a decline in the number of days per year with a net uptake of CO2 and an increase in the number of days per year with a net loss of CO2 (Fig. 4). This reduction in net uptake period was largely due to a greater level of respiration in the late summer and autumn months
Fig. 2. Annual sums of total ecosystem respiration (TER) and gross primary production (GPP) with linear trends during the spring and autumn periods between 1997 and 2013.
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
Fig. 3. Average yearly tree growth sampled from 32 trees within the forest stand. Values show the average area (mm2 ) of each year ring taken from cross sections at breast height. Bars show the standard error of the mean.
(Figs. 5 and 6). Between 2009 and 2013 there have been significantly greater net losses in the autumn period than in the previous years, thus resulting in an earlier end of the net CO2 uptake period. The start of the net uptake period has not changed
33
in time despite there being an increase in the respiration rate in late winter and in early spring (February–April) (Fig. 6). TER was seen to have increased both in the spring (March–May) and the autumn (September–November) during 2010–2013(Fig. 6). This corresponds with the years that there was a net loss from the forest system (2010–2013) (Fig. 1). From 2010 onwards there have been greater losses of CO2 (ca. 5 g m−2 d−1 ) during the winter periods whereas previous years were seen to have almost no respiration (ca. 0 g m−2 d−1 ) during the winter period (December–February). TER has increased resulting in an increase in the number of days per year with a net loss (Figs. 4 and 6). During the measurement period there has been no substantial long term trend towards a change in air temperature (Fig. 7a, right axis). Between 1996 and 2013 there was a non-significant trend towards autumn and spring time warming with a yearly temperature increase of 0.1 ◦ C. However between 2007 and 2013 there is a negative trend in both spring and autumn temperatures. This corresponds to the period where there is a decline in uptake and shift to net losses of CO2 . Autumn and spring time temperatures were relatively cool with 79% of measured spring temperatures and 81% of autumn temperatures ranging between −5 ◦ C to 10 ◦ C (Fig. 8). This range of temperatures is representative for all years as temperatures out of this range predominantly occurred in warmer and cooler years and are therefore considered to be outliers. The forest ecosystem has undergone a change in its response to temperature during both spring and autumn between 1997 and 2013 with an elevated respiration response during the period of 2010–2013 compared with that of 1997–2009 (Fig. 9).
Fig. 4. Number of days each year with a net uptake of CO2 . From 2007 to 2013 there has been a strong decline (linear regression as dashed line) in the number of days with a net uptake. P < 0.05, R2 = 0.74.
34
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
Fig. 5. Contour plot showing the annual net ecosystem exchange (NEE) from year 2000 to 2013. Blue and purple are showing a net uptake whilst yellow and red show net losses.
Furthermore, TER is more sensitive to temperature during the autumn period than spring period with greater carbon losses at the same temperature ranges (−5 to 10 ◦ C), which is shown by the autumn response having a higher elevation than that of spring. A change in temperature response being the main driver for the trend towards net losses was confirmed by applying the average temperature response from years with a net uptake (1997–2009) to years with a net loss (2010–2013) and recalculating the annual budgets. By this combination of measurements and modelling we estimated annual carbon budgets that we would have been able to observe if no change in temperature response had occurred. It resulted in the annual budgets for 2010–2013 (except for 2011) being a net sink instead of a source (Fig. 10). 2011, despite not becoming a net sink, showed substantially fewer losses after recalculation. 4. Discussion NEE is shown to be highly sensitive to small changes in the gross components (TER and GPP) that compose the NEE (Lindroth et al., 1998). A small change in one of these components can result in the forest ecosystem being either a source or a sink of CO2 . 4.1. Gross fluxes and temperature Increased levels of TER (Fig. 2) have driven the forest ecosystem into being a carbon source to the atmosphere. There was no significant trend towards a change in GPP over the measurement period however there was an increase in TER. This suggests that TER is the driver of the net CO2 exchange at Flakaliden. This supports
the findings of Ueyama et al. (2014), which showed that TER controlled the interannual variation of the CO2 balance and not GPP. Furthermore, levels of tree growth had been consistent since 2003 (Fig. 3), which strengthens the findings that GPP had not changed as the forest aged. Between 1998 and 2001 both TER and GPP were relatively high. The high levels of TER during this period can be accredited to increased levels of autotrophic respiration as a direct result of GPP. 1998 was seen to have a NEE close to 0 t C ha−1 (Fig. 1). 1998 was characterised as having relatively large levels of GPP during the spring and TER during the autumn (Fig. 2). 1997 was a particularly warm year (Fig. 7). The NEE during 1998 may therefore be the outcome of a lag effect of a warm period that may have affected forest dynamics through, e.g., litter fall and/or bud formation. The observed increase of TER was seen to occur during the spring and autumn periods (Figs. 5 and 6) with the autumn showing the strongest increase (Fig. 2). Furthermore, prior to 2009, NEE during winter periods was around 0 g CO2 m−2 day−1 when the temperatures were typically below freezing, however from 2009 we observed continual small losses of CO2 during the winter period (Fig. 5). An increased loss of CO2 during the autumn has in recent years been shown by a number of studies (Ueyama et al., 2014; Vesala et al., 2010; Piao et al., 2008) and has largely been accredited to warmer autumns. Piao et al. (2008) stated that an extended growth period may result in a net carbon loss from the ecosystem due to greater levels of autumn respiration than increased levels of uptake in spring. Ueyama et al. (2014) accredited their observed increases of TER to a trend in increasing autumn air temperature. With a 9 year data set of air temperatures they observed an (insignificant) increasing autumn trend of 0.22 ◦ C y−1 .
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
35
Fig. 6. Contour plot showing annual TER for the years 2000 to 2013.
From measured air temperature in Flakaliden (Fig. 7) autumn temperatures had an increasing trend of 0.1 ◦ C y−1 . We must consider however that these local long term measurements are likely to be site specific. Despite this however, the findings raise an important question: does a change in forest functioning have a larger impact on the carbon balance than a change in temperature of this scale? As a rough sensitivity analysis, we recalculated our autumn data with an increased air temperature of 0.22 ◦ C y−1 as observed by Ueyama et al. (2014) in order to demonstrate the effect that this temperature rise would have on the TER at Flakaliden. TER increased by 2.19%. This increase in TER alone could not account for the net losses observed. Accordingly, the actual observed autumn trend at Flakaliden (0.1 ◦ C y−1 ) hardly had the potential to affect our carbon balance at all. Taking into account changes in temperature distribution as well would generate slightly different results, but the order of magnitude would still be relatively small. Furthermore, it can be seen that between 2007 and 2013 there is a negative trend in both spring and autumn air temperatures. This period corresponds to when NEE began to decline and eventually become positive. We conclude that small trends in warming likely have a very little impact on the carbon balance at Flakaliden and consequently, a change in ecosystem functioning will have a greater effect on TER than a temperature change of this magnitude. When the average response from years with a net uptake was applied to the data the annual net losses were greatly reduced and during most years the forest would act as a carbon sink rather than source (Fig. 10). This strongly indicates that the net losses are driven predominantly by a change in temperature response rather than a climatic change.
During the spring and autumn periods 80% of temperatures recorded were in the range of −5 to 10 ◦ C (Fig. 8). A change in the forest’s respiration response to those temperatures (Fig. 9) was observed since 2005 with no change in temperature (Fig. 7). The change in the respiration response was strong for both the autumn and the spring time periods. At the same temperature range (−5 to 10 ◦ C), the autumn respiration was seen to be larger than the spring respiration. This can be seen by the temperature response of autumn TER having a higher elevation (Fig. 9) than that of the spring. Reasons for that difference could be the annual litterfall pattern of Boreal forests (Zhang et al., 2014) or different soil frost dynamics in spring and autumn, respectively, or both. This is in accordance with the findings of Vesala et al. (2010) and Ueyama et al. (2014) who showed a higher increase of autumn TER than that of spring. Although we cannot identify the reason for the increase in temperature response we did observe an increased rate of respiration at temperatures typically occurring during the autumn and spring periods. With the boreal zone having increasingly warmer seasons (Xu et al., 2013) the observed increased carbon losses at Flakaliden may be further amplified if current autumn temperatures are extended into the winter period. 4.2. Biomass and LAI Although there was an observed change in the forests response to temperature, an explanation of what factors may have driven this change has been hard to identify. One possibility is that there has been an increase in dead biomass on the forest floor, which would have contributed towards an increase in TER and therefore could cause an elevation of the temperature responses. The temper-
36
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
Fig. 7. (a) Half hourly means of air temperature (◦ C) during the period 1996 to 2013 (left axis). There has been no substantial trend in annual air temperature (black line and right hand axis, observe the different scales) during the measurement period (y = 3.77E − 10 x + 2.356). (b) Average temperatures for spring and autumn between 1996 and 2013. A non-significant trend (solid line) towards warming was observed for spring and autumn with p-values of 0.48 and 0.29 respectively. Dashed lines show a negative trend in spring and autumn temperature between 2007 and 2013. Spring slope = −0.16 ◦ C yr−1 . Autumn slope = −0.2 ◦ C yr−1 .
ature response was seen clearly to have elevated, with respiration being at a higher level after 2009 (Fig. 9). This may suggest that there was an increased availability of material for decomposition. Correspondingly, regular web-cam pictures and ocular inspections of the trunk space indicated a trend towards increasing amounts of dead wood and litter in the understory. According to our measurements and reports by Lindroth et al. (2008) and Bergh et al. (2003) the projected LAI at Flakaliden had increased from 2.0 in
1994 to 3.4 in 2014, with a major rise (2.1–3.2) between 2002 and 2009. This was not associated with an increase of GPP or biomass increment, which indicated that the photosynthetic capacity of the stand remained unchanged, while the projected biomass increased. This might imply an increase of living and dead needle and branch biomass that contributed to the measured LAI and increased both autotrophic and heterotrophic respiration, but not photosynthesis. We should therefore consider LAI as a measurement of vegetation
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
Fig. 8. Density plot showing autumn and spring temperature distribution for 1997–2013.
37
Fig. 10. Annual sums showing measured and modelled values where the average temperature response for period 1997–2009 is applied to years with a net loss (2010–3013).
4.3. Water availability We analysed monthly precipitation data which was recorded 20 km from our flux site. There was no trend towards a change in precipitation levels during the measurement period (trend of 9.25E–09 mm per year). Studies have shown that the ecosystem carbon balance is partly driven by water availability (Flexas et al., 2006; Hu et al., 2010). The results presented here show that the greatest changes in CO2 losses occurred during the spring and autumn periods (Figs. 2, 5 and 6). Although we observed no change in precipitation, the effects of moisture availability cannot be ruled out. A change in forest stand dynamics, such as a denser canopy, and more dead biomass may lead to a change in moisture retention of the understory and soil. 4.4. Concluding remarks
Fig. 9. Average temperature response for the spring and autumn during 1997–2009 and 2012–2013. The error bars denote standard deviations.
area index (VAI). Furthermore, the LAI measurements were carried out by several different persons and at different points, which can lead to errors and makes it difficult to identify trends with sufficient accuracy. This is however contradictory to older studies which find negative correlations between litterfall and forest age (e.g., Berg et al., 1999). The study site at Flakaliden will undergo a thinning regime as part of normal forest management procedures, which will result in a substantial part of the current biomass being removed. This will allow for future analysis in understanding the role that forest management plays on the carbon balance and to what extent thinning regimes impact the TER.
In conclusion we have shown that a productive managed forest has swung from being a net carbon sink to a net carbon source within a short time scale, caused by an increase of total ecosystem respiration in autumn and spring. The increase in TER is explained by the forest’s response to temperature changing during the measurement period, especially during the autumn. We did not observe a substantial trend towards a changing temperature during the 17 years of measurements, and therefore cannot readily associate the increased levels of TER to climatic changes. A change in forest ecosystem functioning is a larger driver of the carbon balance than a change in climate at Flakaliden. This finding is significant with regards to modelling ecosystem carbon balances and thus future climate. Changes in forest dynamics should therefore be considered in coupled climate models. Acknowledgements This study was supported by EU projects Euroflux (EU-FP4) and Carboeuroflux (EU-FP5), and by the environmental monitoring program Fomaflux at the Swedish University of Agricultural Sciences. We gratefully acknowledge the technical assistance by the staff of the Flakaliden research station.
38
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
References Aubinet, M., Grelle, A., Ibrom, A., Rannik, ü., Moncrieff, J., Foken, T., Kowalski, A.S., Martin, P.H., Berbigier, P., Bernhofer, C., Clement, R., Elbers, J., Granier, A., Grünwald, T., Morgenstern, K., Pilegaard, K., Rebmann, C., Snijders, W., Valentini, R., Vesala, T., 2000. Estimates of the annual net carbon and water exchange of forests: the EUROFLUX methodology. Adv. Ecol. Res. 30, 113–175. Barr, A.G., Black, T., Hogg, E., Griffis, T., Morgenstern, K., Kljun, N., Theede, A., Nesic, Z., 2007. Climatic controls on the carbon and water balances of a boreal aspen forest, 1994–2003. Global Change Biol. 13, 561–576. Berg, B., Albrektson, A., Berg, M.P., et al., 1999. Amounts of litter fall in some pine forests in a European transect, in particular Scots pine. Ann. For. Sci. 56, 625–639. Bergeron, O., Margolis, H.A., Black, T.A., Coursolle, C., Dunn, A.L., Barr, A.G., Wofsy, S.C., 2007. Comparison of carbon dioxide fluxes over three boreal black spruce forests in Canada. Global Change Biol. 13, 89–107. Bergh, J., Freeman, M., Sigurdsson, B., et al., 2003. Modelling the short-term effects of climate change on the productivity of selected tree species in Nordic countries. For. Ecol. Manage. 183, 327–340. Flexas, J., Bota, J., Galmes, J., Medrano, H., Ribas-Carbó, M., 2006. Keeping a positive carbon balance under adverse conditions: responses of photosynthesis and respiration to water stress. Physiol. Plant. 127, 343–352. Frich, P., Alexander, L.V., Della-Marta, P., Gleason, B., Haylock, M., Klein Tank, A.M.G., Peterson, T., 2002. Observed coherent changes in climatic extremes during the second half of the twentieth century. Clim. Res. 19, 193–212. Ge, Z.M., Kellomäki, S., Zhou, X.A., Wang, K.Y., Peltola, H., 2011. Evaluation of carbon exchange in a boreal coniferous stand over a 10-year period: an integrated analysis based on ecosystem model simulations and eddy covariance measurements. Agric. For. Meteorol. 151, 191–203. Grelle, A., Lindroth, A., 1996. Eddy-correlation system for long-term monitoring of fluxes of heat, water vapour and CO2. Global Change Biol. 2, 297–307. Grelle, A., 1997. Long-term Water and Carbon Fluxes from a Boreal Forest; Methods and Applications. Acta Universitatis Agriculturae Sueciae—Silvestria, pp. 80. Harden, J.W., O’Neill, K.P., Trumbore, S.E., Veldhuis, H., Stocks, B.J., 1997. Moss and soil contributions to the annual net carbon flux of a maturing boreal forest. J. Geophys. Res. D: Atmos. 102, 28805–28816. Hu, J., Moore, D.J., Burns, S.P., Monson, R.K., 2010. Longer growing seasons lead to less carbon sequestration by a subalpine forest. Global Change Biol. 16, 771–783. Krishnan, P., Black, T.A., Barr, A.G., Grant, N.J., Gaumont-Guay, D., Nesic, Z., 2008. Factors controlling the interannual variability in the carbon balance of a southern boreal black spruce forest. J. Geophys. Res.: Atmos. 113, 1984–2012.
Lee, X., Massman, W., Law, B., 2004. Handbook of micrometeorology – a guide for surface flux measurement and analysis – introduction. Atmos. Oceanogr. Sci. Lib. 29, 1–5. Lindroth, A., Grelle, A., Morén, A.S., 1998. Long-term measurements of boreal forest carbon balance reveal large temperature sensitivity. Global Change Biol. 4, 443–450. Lindroth, A., Lagergren, F., Aurela, M., et al., 2008. Leaf area index is the principal scaling parameter for both gross photosynthesis and ecosystem respiration of Northern deciduous and coniferous forests. Tellus 60B, 129–142. Pan, Y., Birdsey, R.A., Fang, J., Houghton, R., Kauppi, P.E., Kurz, W.A., Phillips, O.L., Shvidenko, A., Lewis, S.L., Canadell, J.G., Ciais, P., Jackson, R.B., Pacala, S.W., McGuire, A.D., Piao, S., Rautiainen, A., Sitch, S., Hayes, D., 2011. A large and persistent carbon sink in the world’s forests. Science 333, 988–993. Piao, S., Ciais, P., Friedlingstein, P., Peylin, P., Reichstein, M., Luyssaert, S., Margolis, H., Fang, J., Barr, A., Chen, A., Grelle, A., David, Y., Hollinger, T.Y., Lindroth, A., Richardson, A.D., Vesala, T., 2008. Net carbon dioxide losses of northern ecosystems in response to autumn warming. Nature 451, 49–52. Showstack, R., 2013. Carbon dioxide tops 400 ppm at mauna loa, Hawaii.E os. Trans. Am. Geophys. Union 94 (192-192). Tanja, S., Berninger, F., Vesala, T., Markkanen, T., Hari, P., Mäkelä, A., Ilvesniemi, H., Hänninen, H., Nikinmaa, E., Huttula, T., Laurila, T., Aurela, M., Grelle, A., Lindroth, A., Arneth, A., Shibistova, O., Lloyd, J., 2003. Air temperature triggers the recovery of evergreen boreal forest photosynthesis in spring. Global Change Biol. 9, 1410–1426. Ueyama, M., Iwata, H., Harazono, Y., 2014. Autumn warming reduces the CO2 sink of a black spruce forest in interior Alaska based on a nine-year eddy covariance measurement. Global Change Biol. 20, 1161–1173. Vesala, T., Launiainen, S., Kolari, P., Pumpanen, J., Sevanto, S., Hari, P., Nikinmaa, E., Kaski, P., Mannila, H., Ukkonen, E., Piao, S.L., Ciais, P., 2010. Autumn temperature and carbon balance of a boreal Scots pine forest in Southern Finland. Biogeosciences 7, 163–176. Wu, J., Linden, L., Lasslop, G., Carvalhais, N., Pilegaard, K., Beier, C., Ibrom, A., 2012. Effects of climate variability and functional changes on the interannual variation of the carbon balance in a temperate deciduous forest. Biogeosciences 9, 13–28. Xu, L., Myneni, R., Chapin Iii, F., Callaghan, T., Pinzon, J., Tucker, C., Zhu, Z., Bi, J., Ciais, P., Tømmervik, H., Euskirchen, E.S., Forbes, B.C., Piao, S.L., Anderson, B.T., Ganguly, S., Nemani, R.R., Goetz, S.J., Beck, P.S.A., Bunn, A.G., Cao, C., Stroeve, J.C., 2013. Temperature and vegetation seasonality diminishment over northern lands. Nat. Clim. Change 3, 581–586. Zhang, H., Yuan, W., Dong, W., Liu, S., 2014. Seasonal patterns of litterfall in forest ecosystem worldwide. Ecol. Complexity 20, 240–247.
Contents lists available at ScienceDirect
Agricultural and Forest Meteorology journal homepage: www.elsevier.com/locate/agrformet
Changing temperature response of respiration turns boreal forest from carbon sink into carbon source David Hadden ∗ , Achim Grelle Department of Ecology, Swedish University of Agricultural Sciences, Sweden
a r t i c l e
i n f o
Article history: Received 8 September 2015 Received in revised form 23 March 2016 Accepted 30 March 2016 Available online 6 April 2016 Keywords: Eddy covariance Carbon dioxide exchange Temperature response Boreal forest Carbon balance Respiration
a b s t r a c t Seventeen years (1997–2013) of carbon dioxide (CO2 ) fluxes were measured in a boreal forest stand in northern Sweden using the eddy covariance technique. During the measurement period the forest turned from a net carbon sink into a net carbon source. The net ecosystem exchange (NEE) was separated using values from periods of darkness into the gross components of total ecosystem respiration (TER) and gross primary productivity (GPP), which was calculated as GPP = −NEE + TER. From the gross components we could determine that an increase in TER during the autumn (September to end of November) and spring (March to end of May) periods resulted in the forest becoming a net source of CO2. We observed no increase in the GPP from the eddy covariance measurements. This was further supported by measurements of tree growth rings. The increased TER was attributed to a change in the forest’s temperature response at lower temperatures (−5 to 10 ◦ C) rather than to a temperature increase. This study shows that changes in ecosystem functioning can have a larger impact on the carbon balance than climate warming per se. © 2016 Elsevier B.V. All rights reserved.
1. Introduction In 2013 the daily mean concentration of atmospheric carbon dioxide (CO2 ) surpassed 400 ppm globally (Showstack, 2013). Anthropogenic emissions of CO2 are on the rise, which is having a direct impact on global warming. With an increasing concentration of atmospheric CO2 the ability of terrestrial ecosystems to sequester carbon is of great importance. Global forests play a vital role in cycling CO2 between the atmosphere and terrestrial biosphere and are capable of sequestering 4.0 Pg C year−1 (Pan et al., 2011). The Boreal forest in particular has been acting as a carbon sink for thousands of years and contains some of the world’s largest soil carbon stocks (Harden et al., 1997). The net uptake of CO2 within the forest ecosystem is the difference between the gross primary productivity (GPP) and the total ecosystem respiration (TER). A small change in either of the gross fluxes is therefore crucial in determining whether a forest may act as a sink or source of CO2 (Lindroth et al., 1998). Air temperature has been shown to be one of the main drivers of both carbon uptake and ecosystem respiration (Barr et al., 2007; Bergeron et al., 2007). The onset of the growth period in the spring is largely regulated by air temperatures and has a direct impact on
∗ Corresponding author. E-mail address: [email protected] (D. Hadden). http://dx.doi.org/10.1016/j.agrformet.2016.03.020 0168-1923/© 2016 Elsevier B.V. All rights reserved.
net annual uptake of CO2 from the atmosphere (Krishnan et al., 2008; Tanja et al., 2003). An increase in air temperature however also increases ecosystem respiration which has been shown to result in a net loss of carbon from the terrestrial system during the autumn months (Piao et al., 2008; Ueyama et al., 2014; Vesala et al., 2010). Furthermore, there are indications that a change in ecosystem function has a large impact on the ecosystem carbon balance and dominates the interannual variations to a greater extent than climatic variability which appears to impact the carbon fluxes at short time scales (Wu et al., 2012). These findings have rendered it difficult to evaluate future consequences of climate change on the Boreal zone’s carbon balance. One of the keys to predicting future consequences of warming on the Boreal forest carbon balance is to understand the underlying processes that drive the forest’s response to temperature. Due to the stochastic nature of weather and climate combined with the long life cycle of the boreal forest and changes within ecosystem functioning, long term measurements are needed to gain a detailed understanding of the carbon cycle on a large temporal scale. We must also consider the forest dynamics and how these change as the forest ages when evaluating the carbon balance. There are relatively few datasets of continuous long term carbon fluxes from the boreal zone (more than 15 years in length) where the interannual variations of climate and fluxes can be evaluated. Most long term measurements have been done by the eddy covariance technique (e.g., Ge et al., 2011; Ueyama et al., 2014; Vesala
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
et al., 2010). There are numerous studies showing the impact that a changing climate may have on the boreal forest’s carbon balance. There are however no studies that we are aware of that evaluate a change in the forests functioning and the impacts that this has on the carbon balance with the prospects of a changing climate. The objectives of this study are to report the interannual variations of CO2 fluxes within the boreal forest ecosystem and identify the underlying processes controlling the fluxes. We present a 17 year record (1997–2013) of eddy covariance data measured in northern Sweden. During this period we observed a change in the ecosystem function resulting in the forest turning from a carbon sink to a carbon source. 2. Materials and methods 2.1. Study site The experimental site is located at Flakaliden in northern Sweden (64◦ 07 N, 19◦ 27 E, 310 m above sea level). Flakaliden has a boreal climate, experiencing long and cold, dark winters and short, cool summers with long daylight hours. The mean temperature during the measurement period was 2.4 ◦ C. The site is a managed forest on a shallow till soil that consists predominantly of Norwegian spruce (Picea abies) with an average age of ca. 50 years. The average tree height in 2013 was ca. 12 m with an average diameter of 13 cm. The forest stand had a leaf area index (projected LAI) of 3.40 as of September 2014.
31
18 m height using an LI 190 quantum sensor (LI-COR inc., Lincoln, Nebraska, USA). Leaf area index (LAI) was measured using an LAI-2000 plant canopy analyser (LI-COR inc., Lincoln, Nebraska, USA) on a regular basis throughout the measurement period. This was done by manual multi-point measurements at 10 m intervals in North, East, South and Westerly directions from the tower. 2.3. Temperature response There is a correlation between air temperature and total ecosystem respiration which has resulted in air temperature being largely regarded as a main driver for TER (Barr et al., 2007; Bergeron et al., 2007). Due to this we chose to study the response of TER to air temperature rather than soil temperatures. Furthermore, this study is based upon eddy covariance measurements which measure the carbon exchange at the ecosystem level. It could therefore be misleading to analyse the carbon exchange at the ecosystem level with only soil temperatures as a driver. Indeed, generally our CO2 fluxes correlated better with air temperature than with soil temperatures and we therefore chose to carry out our analysis of respiration response to air temperatures rather than soil temperatures. The gross respiration was calculated with the equation: TER = a + b · e c · T
(1)
with a representing the offset and b and c represent curve coefficients. 2.4. Growth period
2.2. Instrumentation Turbulent fluxes of CO2 were measured by an eddy-covariance system (In Situ Instrument AB, Ockelbo, Sweden), consisting basically of a solent 1012R2 sonic anemometer (Gill Instruments, Lymington, UK) and a closed path infrared gas analyzer LI-6262 (LI-COR inc., Lincoln, Nebraska, USA) as described by Grelle and Lindroth (1996). The eddy covariance instrumentation was mounted on an 18 m tall guyed mast with a triangular cross section of 28 cm width. The flux measurements were taken at 15 m height on a boom extending 2 m from the mast. Extensive flux source area analyses have been applied to confirm that the appropriate height for flux measurements was chosen to represent the surrounding forest both during day- and night-time (Grelle, 1997). To minimize damping of high-frequency concentration fluctuations in a long intake tube, the entire system was mounted at the tower close to measurement height. The tube inlet was placed 10 cm below the sampling volume of the sonic anemometer. Air was drawn from the inlet through a 6 mm diameter, 6 m long high-density polyethylene tube to the IRGA, which was mounted at the mast in a heated, ventilated, and insulated enclosure at a height of 13 m. The air flow rate was 12 ln/min (litre at 0 ◦ C, 1013 hPa), measured and controlled by a mass flow regulator (Brooks Instrument, Hatfield, PA, USA). The time lag between data samples from the anemometer and the gas analyser was continuously determined by autocovariance analysis on a half-hourly basis, and the signals were shifted accordingly prior to further analysis. Energy balance closure indicated adequate performance of the measurement system (Grelle, 1997). Overall calculation and correction of fluxes followed the EUROFLUX methodology (Aubinet et al., 2000) and Lee et al. (2004). Soil temperature was measured in two profiles from 5 cm to 50 cm depth by 10 type 107 thermistors (Campbell Scientific, Logan, UT, USA), and air temperature and -humidity by an MP103A sensor (ROTRONIC AG, Bassersdorf, Switzerland) placed in a ventilated radiation shield (In Situ Instrument AB, Ockelbo, Sweden) at 18 m height. Photosynthetically active radiation (PAR) was measured at
Following a widely used convention, the growth period is defined as starting when the daily average temperature is ≥5 ◦ C for 5 consecutive days and ending when the daily average temperature is <5 ◦ C for 5 consecutive days (e.g., Frich et al., 2002). 2.5. Tree growth Tree growth was measured using core samples taken from 40 trees. Each sample was taken at breast height and from the south side of the trunk. Each sample was photographed and the year rings were measured in width with the use of a pixel counting software (Pixel Ruler inc.). Each year ring was calculated into area as follows: Ai =
2 D i
2
−
D
i+1
2
2
(2)
Di = Di+1 + 2Wi i=
1, 2, 3, .., N
The index number of a given year ring, i, is counted from outside towards the pith of the stem. N is the total number of year rings measured. Ai is the portion of the whole stem area represented by the area of the year ring growing from i + 1 to i. Di+1 is the diameter of the whole stem area when growth of the year ring began. Di is the diameter of the whole stem area including the growth represented by the width, Wi , of the year ring produced from i + 1 to i. D1 is stem diameter under bark. 2.6. Gap filling After quality control, there was an average fraction of 24.9% gaps in the flux data and 6.4% in the climatic data throughout the 17 years of measurements. For periods when climatic data were available, gaps in the flux data were filled by a neural network (NNDT, Saxén
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
Gaps were filled in this fashion in order to limit a loss of any trend that may have been present within the time series data.
5 0
NEE (tonnes CO2 ha−1 yr−1)
1. Half-hourly gaps were filled by linear interpolation 2. Gaps of up to 2 h were filled with averaging 2 h before and after the gap 3. Gaps of more than 2 h up to 2 week in size were bin-averaged from corresponding period directly before and after the gap 4. Gaps of more than 2 weeks were filled using bin averages from previous and following years
10
& Saxén, 1995) using air temperature, PAR, and wind speed as input variables. When climatic data were missing, gaps in the fluxes were filled according to the following criteria:
−5
32
The separation of the NEE was done on a seasonal basis. December to the end of February was considered as winter. March to the end of May was considered as spring. June to the end of August was considered as summer. September to the end of November was considered as autumn. Flux data during periods of darkness, when photosynthesis effectively is zero, were used to estimate total ecosystem gross respiration (TER). We defined periods of darkness as half an hour after sunset until half an hour before sunrise. During well-mixed periods (friction velocity u∗ > 0.2 m s−1 , Aubinet et al., 2000), an exponential temperature response (eq. 1) was fitted to the measured ecosystem fluxes. This response was then used to estimate daytime respiration to complement night-time measurements, which resulted in total ecosystem gross respiration. We used air temperature as the dependent variable describing TER as an exponential function, since this had shown the best explanatory power for ecosystem respiration that is composed by autotrophic and heterotrophic respiration from biomass and soil. Based on this measured/modelled TER and measured NEE, the gross primary productivity (GPP) was then calculated as: GPP = −NEE + TER
−10
2.7. Partitioning of NEE into gross components (GPP and TER)
1997 1999 2001 2003 2005 2007 2009 2011 2013 Fig. 1. Annual Net Ecosystem Exchange (NEE) from 1997 to 2013. From 2007 to 2009 the forest ecosystem has had a declining net uptake followed by a net loss of C from 2010 to 2013.
3. Results During a 17 year period the forest ecosystem has swung from a carbon sink to a carbon source. From 1997 until 2009 – with the exception of 1998 when it was close to neutral – the forest was annually a continuous net carbon sink. The forest’s ability to sink carbon is seen to decline consistently from 2007 (−1.4 t C ha−1 y−1 ) until 2013 (1.2 t C ha−1 y−1 ) with the forest becoming a net carbon source in 2010 (Fig. 1). From 2010 until 2013 the forest ecosystem consistently had an annual loss of CO2 to the atmosphere. During both spring and autumn time there had been a trend towards increased TER, however no trend towards a change in GPP over the entire measurement period (Fig. 2). TER was seen to have a greater increase during the autumn period than the spring period (Figs. 2 and 6). This increasing trend in TER had a significant impact on NEE and resulted in the forest becoming a net carbon source on an annual basis from 2010. Tree growth on the other hand, had been consistent since 2004, while showing some annual variations from 1996 to 2003 (Fig. 3). This consolidates our finding that GPP was consistent (Fig. 2). Together with the change of NEE during the measurement period there has been a decline in the number of days per year with a net uptake of CO2 and an increase in the number of days per year with a net loss of CO2 (Fig. 4). This reduction in net uptake period was largely due to a greater level of respiration in the late summer and autumn months
Fig. 2. Annual sums of total ecosystem respiration (TER) and gross primary production (GPP) with linear trends during the spring and autumn periods between 1997 and 2013.
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
Fig. 3. Average yearly tree growth sampled from 32 trees within the forest stand. Values show the average area (mm2 ) of each year ring taken from cross sections at breast height. Bars show the standard error of the mean.
(Figs. 5 and 6). Between 2009 and 2013 there have been significantly greater net losses in the autumn period than in the previous years, thus resulting in an earlier end of the net CO2 uptake period. The start of the net uptake period has not changed
33
in time despite there being an increase in the respiration rate in late winter and in early spring (February–April) (Fig. 6). TER was seen to have increased both in the spring (March–May) and the autumn (September–November) during 2010–2013(Fig. 6). This corresponds with the years that there was a net loss from the forest system (2010–2013) (Fig. 1). From 2010 onwards there have been greater losses of CO2 (ca. 5 g m−2 d−1 ) during the winter periods whereas previous years were seen to have almost no respiration (ca. 0 g m−2 d−1 ) during the winter period (December–February). TER has increased resulting in an increase in the number of days per year with a net loss (Figs. 4 and 6). During the measurement period there has been no substantial long term trend towards a change in air temperature (Fig. 7a, right axis). Between 1996 and 2013 there was a non-significant trend towards autumn and spring time warming with a yearly temperature increase of 0.1 ◦ C. However between 2007 and 2013 there is a negative trend in both spring and autumn temperatures. This corresponds to the period where there is a decline in uptake and shift to net losses of CO2 . Autumn and spring time temperatures were relatively cool with 79% of measured spring temperatures and 81% of autumn temperatures ranging between −5 ◦ C to 10 ◦ C (Fig. 8). This range of temperatures is representative for all years as temperatures out of this range predominantly occurred in warmer and cooler years and are therefore considered to be outliers. The forest ecosystem has undergone a change in its response to temperature during both spring and autumn between 1997 and 2013 with an elevated respiration response during the period of 2010–2013 compared with that of 1997–2009 (Fig. 9).
Fig. 4. Number of days each year with a net uptake of CO2 . From 2007 to 2013 there has been a strong decline (linear regression as dashed line) in the number of days with a net uptake. P < 0.05, R2 = 0.74.
34
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
Fig. 5. Contour plot showing the annual net ecosystem exchange (NEE) from year 2000 to 2013. Blue and purple are showing a net uptake whilst yellow and red show net losses.
Furthermore, TER is more sensitive to temperature during the autumn period than spring period with greater carbon losses at the same temperature ranges (−5 to 10 ◦ C), which is shown by the autumn response having a higher elevation than that of spring. A change in temperature response being the main driver for the trend towards net losses was confirmed by applying the average temperature response from years with a net uptake (1997–2009) to years with a net loss (2010–2013) and recalculating the annual budgets. By this combination of measurements and modelling we estimated annual carbon budgets that we would have been able to observe if no change in temperature response had occurred. It resulted in the annual budgets for 2010–2013 (except for 2011) being a net sink instead of a source (Fig. 10). 2011, despite not becoming a net sink, showed substantially fewer losses after recalculation. 4. Discussion NEE is shown to be highly sensitive to small changes in the gross components (TER and GPP) that compose the NEE (Lindroth et al., 1998). A small change in one of these components can result in the forest ecosystem being either a source or a sink of CO2 . 4.1. Gross fluxes and temperature Increased levels of TER (Fig. 2) have driven the forest ecosystem into being a carbon source to the atmosphere. There was no significant trend towards a change in GPP over the measurement period however there was an increase in TER. This suggests that TER is the driver of the net CO2 exchange at Flakaliden. This supports
the findings of Ueyama et al. (2014), which showed that TER controlled the interannual variation of the CO2 balance and not GPP. Furthermore, levels of tree growth had been consistent since 2003 (Fig. 3), which strengthens the findings that GPP had not changed as the forest aged. Between 1998 and 2001 both TER and GPP were relatively high. The high levels of TER during this period can be accredited to increased levels of autotrophic respiration as a direct result of GPP. 1998 was seen to have a NEE close to 0 t C ha−1 (Fig. 1). 1998 was characterised as having relatively large levels of GPP during the spring and TER during the autumn (Fig. 2). 1997 was a particularly warm year (Fig. 7). The NEE during 1998 may therefore be the outcome of a lag effect of a warm period that may have affected forest dynamics through, e.g., litter fall and/or bud formation. The observed increase of TER was seen to occur during the spring and autumn periods (Figs. 5 and 6) with the autumn showing the strongest increase (Fig. 2). Furthermore, prior to 2009, NEE during winter periods was around 0 g CO2 m−2 day−1 when the temperatures were typically below freezing, however from 2009 we observed continual small losses of CO2 during the winter period (Fig. 5). An increased loss of CO2 during the autumn has in recent years been shown by a number of studies (Ueyama et al., 2014; Vesala et al., 2010; Piao et al., 2008) and has largely been accredited to warmer autumns. Piao et al. (2008) stated that an extended growth period may result in a net carbon loss from the ecosystem due to greater levels of autumn respiration than increased levels of uptake in spring. Ueyama et al. (2014) accredited their observed increases of TER to a trend in increasing autumn air temperature. With a 9 year data set of air temperatures they observed an (insignificant) increasing autumn trend of 0.22 ◦ C y−1 .
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
35
Fig. 6. Contour plot showing annual TER for the years 2000 to 2013.
From measured air temperature in Flakaliden (Fig. 7) autumn temperatures had an increasing trend of 0.1 ◦ C y−1 . We must consider however that these local long term measurements are likely to be site specific. Despite this however, the findings raise an important question: does a change in forest functioning have a larger impact on the carbon balance than a change in temperature of this scale? As a rough sensitivity analysis, we recalculated our autumn data with an increased air temperature of 0.22 ◦ C y−1 as observed by Ueyama et al. (2014) in order to demonstrate the effect that this temperature rise would have on the TER at Flakaliden. TER increased by 2.19%. This increase in TER alone could not account for the net losses observed. Accordingly, the actual observed autumn trend at Flakaliden (0.1 ◦ C y−1 ) hardly had the potential to affect our carbon balance at all. Taking into account changes in temperature distribution as well would generate slightly different results, but the order of magnitude would still be relatively small. Furthermore, it can be seen that between 2007 and 2013 there is a negative trend in both spring and autumn air temperatures. This period corresponds to when NEE began to decline and eventually become positive. We conclude that small trends in warming likely have a very little impact on the carbon balance at Flakaliden and consequently, a change in ecosystem functioning will have a greater effect on TER than a temperature change of this magnitude. When the average response from years with a net uptake was applied to the data the annual net losses were greatly reduced and during most years the forest would act as a carbon sink rather than source (Fig. 10). This strongly indicates that the net losses are driven predominantly by a change in temperature response rather than a climatic change.
During the spring and autumn periods 80% of temperatures recorded were in the range of −5 to 10 ◦ C (Fig. 8). A change in the forest’s respiration response to those temperatures (Fig. 9) was observed since 2005 with no change in temperature (Fig. 7). The change in the respiration response was strong for both the autumn and the spring time periods. At the same temperature range (−5 to 10 ◦ C), the autumn respiration was seen to be larger than the spring respiration. This can be seen by the temperature response of autumn TER having a higher elevation (Fig. 9) than that of the spring. Reasons for that difference could be the annual litterfall pattern of Boreal forests (Zhang et al., 2014) or different soil frost dynamics in spring and autumn, respectively, or both. This is in accordance with the findings of Vesala et al. (2010) and Ueyama et al. (2014) who showed a higher increase of autumn TER than that of spring. Although we cannot identify the reason for the increase in temperature response we did observe an increased rate of respiration at temperatures typically occurring during the autumn and spring periods. With the boreal zone having increasingly warmer seasons (Xu et al., 2013) the observed increased carbon losses at Flakaliden may be further amplified if current autumn temperatures are extended into the winter period. 4.2. Biomass and LAI Although there was an observed change in the forests response to temperature, an explanation of what factors may have driven this change has been hard to identify. One possibility is that there has been an increase in dead biomass on the forest floor, which would have contributed towards an increase in TER and therefore could cause an elevation of the temperature responses. The temper-
36
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
Fig. 7. (a) Half hourly means of air temperature (◦ C) during the period 1996 to 2013 (left axis). There has been no substantial trend in annual air temperature (black line and right hand axis, observe the different scales) during the measurement period (y = 3.77E − 10 x + 2.356). (b) Average temperatures for spring and autumn between 1996 and 2013. A non-significant trend (solid line) towards warming was observed for spring and autumn with p-values of 0.48 and 0.29 respectively. Dashed lines show a negative trend in spring and autumn temperature between 2007 and 2013. Spring slope = −0.16 ◦ C yr−1 . Autumn slope = −0.2 ◦ C yr−1 .
ature response was seen clearly to have elevated, with respiration being at a higher level after 2009 (Fig. 9). This may suggest that there was an increased availability of material for decomposition. Correspondingly, regular web-cam pictures and ocular inspections of the trunk space indicated a trend towards increasing amounts of dead wood and litter in the understory. According to our measurements and reports by Lindroth et al. (2008) and Bergh et al. (2003) the projected LAI at Flakaliden had increased from 2.0 in
1994 to 3.4 in 2014, with a major rise (2.1–3.2) between 2002 and 2009. This was not associated with an increase of GPP or biomass increment, which indicated that the photosynthetic capacity of the stand remained unchanged, while the projected biomass increased. This might imply an increase of living and dead needle and branch biomass that contributed to the measured LAI and increased both autotrophic and heterotrophic respiration, but not photosynthesis. We should therefore consider LAI as a measurement of vegetation
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
Fig. 8. Density plot showing autumn and spring temperature distribution for 1997–2013.
37
Fig. 10. Annual sums showing measured and modelled values where the average temperature response for period 1997–2009 is applied to years with a net loss (2010–3013).
4.3. Water availability We analysed monthly precipitation data which was recorded 20 km from our flux site. There was no trend towards a change in precipitation levels during the measurement period (trend of 9.25E–09 mm per year). Studies have shown that the ecosystem carbon balance is partly driven by water availability (Flexas et al., 2006; Hu et al., 2010). The results presented here show that the greatest changes in CO2 losses occurred during the spring and autumn periods (Figs. 2, 5 and 6). Although we observed no change in precipitation, the effects of moisture availability cannot be ruled out. A change in forest stand dynamics, such as a denser canopy, and more dead biomass may lead to a change in moisture retention of the understory and soil. 4.4. Concluding remarks
Fig. 9. Average temperature response for the spring and autumn during 1997–2009 and 2012–2013. The error bars denote standard deviations.
area index (VAI). Furthermore, the LAI measurements were carried out by several different persons and at different points, which can lead to errors and makes it difficult to identify trends with sufficient accuracy. This is however contradictory to older studies which find negative correlations between litterfall and forest age (e.g., Berg et al., 1999). The study site at Flakaliden will undergo a thinning regime as part of normal forest management procedures, which will result in a substantial part of the current biomass being removed. This will allow for future analysis in understanding the role that forest management plays on the carbon balance and to what extent thinning regimes impact the TER.
In conclusion we have shown that a productive managed forest has swung from being a net carbon sink to a net carbon source within a short time scale, caused by an increase of total ecosystem respiration in autumn and spring. The increase in TER is explained by the forest’s response to temperature changing during the measurement period, especially during the autumn. We did not observe a substantial trend towards a changing temperature during the 17 years of measurements, and therefore cannot readily associate the increased levels of TER to climatic changes. A change in forest ecosystem functioning is a larger driver of the carbon balance than a change in climate at Flakaliden. This finding is significant with regards to modelling ecosystem carbon balances and thus future climate. Changes in forest dynamics should therefore be considered in coupled climate models. Acknowledgements This study was supported by EU projects Euroflux (EU-FP4) and Carboeuroflux (EU-FP5), and by the environmental monitoring program Fomaflux at the Swedish University of Agricultural Sciences. We gratefully acknowledge the technical assistance by the staff of the Flakaliden research station.
38
D. Hadden, A. Grelle / Agricultural and Forest Meteorology 223 (2016) 30–38
References Aubinet, M., Grelle, A., Ibrom, A., Rannik, ü., Moncrieff, J., Foken, T., Kowalski, A.S., Martin, P.H., Berbigier, P., Bernhofer, C., Clement, R., Elbers, J., Granier, A., Grünwald, T., Morgenstern, K., Pilegaard, K., Rebmann, C., Snijders, W., Valentini, R., Vesala, T., 2000. Estimates of the annual net carbon and water exchange of forests: the EUROFLUX methodology. Adv. Ecol. Res. 30, 113–175. Barr, A.G., Black, T., Hogg, E., Griffis, T., Morgenstern, K., Kljun, N., Theede, A., Nesic, Z., 2007. Climatic controls on the carbon and water balances of a boreal aspen forest, 1994–2003. Global Change Biol. 13, 561–576. Berg, B., Albrektson, A., Berg, M.P., et al., 1999. Amounts of litter fall in some pine forests in a European transect, in particular Scots pine. Ann. For. Sci. 56, 625–639. Bergeron, O., Margolis, H.A., Black, T.A., Coursolle, C., Dunn, A.L., Barr, A.G., Wofsy, S.C., 2007. Comparison of carbon dioxide fluxes over three boreal black spruce forests in Canada. Global Change Biol. 13, 89–107. Bergh, J., Freeman, M., Sigurdsson, B., et al., 2003. Modelling the short-term effects of climate change on the productivity of selected tree species in Nordic countries. For. Ecol. Manage. 183, 327–340. Flexas, J., Bota, J., Galmes, J., Medrano, H., Ribas-Carbó, M., 2006. Keeping a positive carbon balance under adverse conditions: responses of photosynthesis and respiration to water stress. Physiol. Plant. 127, 343–352. Frich, P., Alexander, L.V., Della-Marta, P., Gleason, B., Haylock, M., Klein Tank, A.M.G., Peterson, T., 2002. Observed coherent changes in climatic extremes during the second half of the twentieth century. Clim. Res. 19, 193–212. Ge, Z.M., Kellomäki, S., Zhou, X.A., Wang, K.Y., Peltola, H., 2011. Evaluation of carbon exchange in a boreal coniferous stand over a 10-year period: an integrated analysis based on ecosystem model simulations and eddy covariance measurements. Agric. For. Meteorol. 151, 191–203. Grelle, A., Lindroth, A., 1996. Eddy-correlation system for long-term monitoring of fluxes of heat, water vapour and CO2. Global Change Biol. 2, 297–307. Grelle, A., 1997. Long-term Water and Carbon Fluxes from a Boreal Forest; Methods and Applications. Acta Universitatis Agriculturae Sueciae—Silvestria, pp. 80. Harden, J.W., O’Neill, K.P., Trumbore, S.E., Veldhuis, H., Stocks, B.J., 1997. Moss and soil contributions to the annual net carbon flux of a maturing boreal forest. J. Geophys. Res. D: Atmos. 102, 28805–28816. Hu, J., Moore, D.J., Burns, S.P., Monson, R.K., 2010. Longer growing seasons lead to less carbon sequestration by a subalpine forest. Global Change Biol. 16, 771–783. Krishnan, P., Black, T.A., Barr, A.G., Grant, N.J., Gaumont-Guay, D., Nesic, Z., 2008. Factors controlling the interannual variability in the carbon balance of a southern boreal black spruce forest. J. Geophys. Res.: Atmos. 113, 1984–2012.
Lee, X., Massman, W., Law, B., 2004. Handbook of micrometeorology – a guide for surface flux measurement and analysis – introduction. Atmos. Oceanogr. Sci. Lib. 29, 1–5. Lindroth, A., Grelle, A., Morén, A.S., 1998. Long-term measurements of boreal forest carbon balance reveal large temperature sensitivity. Global Change Biol. 4, 443–450. Lindroth, A., Lagergren, F., Aurela, M., et al., 2008. Leaf area index is the principal scaling parameter for both gross photosynthesis and ecosystem respiration of Northern deciduous and coniferous forests. Tellus 60B, 129–142. Pan, Y., Birdsey, R.A., Fang, J., Houghton, R., Kauppi, P.E., Kurz, W.A., Phillips, O.L., Shvidenko, A., Lewis, S.L., Canadell, J.G., Ciais, P., Jackson, R.B., Pacala, S.W., McGuire, A.D., Piao, S., Rautiainen, A., Sitch, S., Hayes, D., 2011. A large and persistent carbon sink in the world’s forests. Science 333, 988–993. Piao, S., Ciais, P., Friedlingstein, P., Peylin, P., Reichstein, M., Luyssaert, S., Margolis, H., Fang, J., Barr, A., Chen, A., Grelle, A., David, Y., Hollinger, T.Y., Lindroth, A., Richardson, A.D., Vesala, T., 2008. Net carbon dioxide losses of northern ecosystems in response to autumn warming. Nature 451, 49–52. Showstack, R., 2013. Carbon dioxide tops 400 ppm at mauna loa, Hawaii.E os. Trans. Am. Geophys. Union 94 (192-192). Tanja, S., Berninger, F., Vesala, T., Markkanen, T., Hari, P., Mäkelä, A., Ilvesniemi, H., Hänninen, H., Nikinmaa, E., Huttula, T., Laurila, T., Aurela, M., Grelle, A., Lindroth, A., Arneth, A., Shibistova, O., Lloyd, J., 2003. Air temperature triggers the recovery of evergreen boreal forest photosynthesis in spring. Global Change Biol. 9, 1410–1426. Ueyama, M., Iwata, H., Harazono, Y., 2014. Autumn warming reduces the CO2 sink of a black spruce forest in interior Alaska based on a nine-year eddy covariance measurement. Global Change Biol. 20, 1161–1173. Vesala, T., Launiainen, S., Kolari, P., Pumpanen, J., Sevanto, S., Hari, P., Nikinmaa, E., Kaski, P., Mannila, H., Ukkonen, E., Piao, S.L., Ciais, P., 2010. Autumn temperature and carbon balance of a boreal Scots pine forest in Southern Finland. Biogeosciences 7, 163–176. Wu, J., Linden, L., Lasslop, G., Carvalhais, N., Pilegaard, K., Beier, C., Ibrom, A., 2012. Effects of climate variability and functional changes on the interannual variation of the carbon balance in a temperate deciduous forest. Biogeosciences 9, 13–28. Xu, L., Myneni, R., Chapin Iii, F., Callaghan, T., Pinzon, J., Tucker, C., Zhu, Z., Bi, J., Ciais, P., Tømmervik, H., Euskirchen, E.S., Forbes, B.C., Piao, S.L., Anderson, B.T., Ganguly, S., Nemani, R.R., Goetz, S.J., Beck, P.S.A., Bunn, A.G., Cao, C., Stroeve, J.C., 2013. Temperature and vegetation seasonality diminishment over northern lands. Nat. Clim. Change 3, 581–586. Zhang, H., Yuan, W., Dong, W., Liu, S., 2014. Seasonal patterns of litterfall in forest ecosystem worldwide. Ecol. Complexity 20, 240–247.