10 A.-S. Mor´en, A. Lindroth Agricultural and Forest Meteorology 101 2000 1–14
Fig. 8. Estimated instantaneous gross forest floor CO
2
exchange, F
G
, from Fig. 7b, plotted against a incident light, PPFD, and b chamber temperature, T
ch
for 7–13 June 1996.
of CO
2
released through respiration by about 28 Fig. 7c. Of the 96.6 g m
− 2
respired, 27.3 g m
− 2
was used in photosynthesis, which resulted in a net CO
2
efflux over 7 days of 69.3 g m
− 2
. F
G
showed a large scatter, and only a weak light-response Fig. 8a. Fur-
Fig. 9. Average net forest floor exchange rate, F
N
, based on measurements from the two chambers, accumulated per
day, and plotted against soil temperature, T
s
, at 5 cm. Regres- sion was made for three periods: 11 May–16 June circles,
F
N
= 5.236 + 0.483T
s
, R
2
= 0.42, n = 20; 17 June–8 July squares,
F
N
= − 19.77 + 3.227T
s
, R
2
= 0.24, n = 33; 9 July–7 October tri-
angles, F
N
= − 1.046 + 2.150T
s
, R
2
= 0.69, n = 43. Arrows indi-
cate the seasonal course. Fig. 10. Net forest floor exchange rate, F
N
, accumulated from 1 May–31 October 1996. Daily net CO
2
efflux, F
N
, was based on measurements from the two chambers, accumulated per day and
missing data filled in by linear regression with soil temperature, T
s
, at 5 cm cf. Fig. 9. Bars show F
N
accumulated per month.
thermore, F
G
increased as temperature increased. The temperature optimum was difficult to identify, but the
response in relation to chamber temperature appeared to level out at about 18
◦
C Fig. 8b. From 11 May–7 October, chamber measurements
covered 96 complete days. Soil temperatures were available for 179 days, from 6 May–31 October.
A linear relationship was established between daily accumulated net CO
2
efflux of the forest floor and mean soil temperature. The model was improved by
dividing data into three groups, 11 May–16 June, 17 June–8 July and 9 July–7 October Fig. 9. The third
period showed the strongest correlation with tem- perature, describing 69 of the variation in F
N
. To cover 6 months, 1 May–31 October, F
N
was modelled from soil temperature where no data were available
from soil chambers. Over the 6-month period, about 3.1 kg m
− 2
of CO
2
was released to the atmosphere. On a monthly basis, F
N
was lowest in May, with 6 of the total F
N
, and highest in August, with 30 of the 6-month total F
N
Fig. 10.
4. Discussion
4.1. Respiration rates and seasonal variation Among more recent studies in boreal forests,
Goulden and Crill 1997, Lavigne et al. 1997 and Rayment and Jarvis 1997 reported respiration rates
A.-S. Mor´en, A. Lindroth Agricultural and Forest Meteorology 101 2000 1–14 11
in the range 0.05−0.15 mg m
− 2
s
− 1
at 10
◦
C soil temperatures at 4–10 cm depth. Rayment and Jarvis
1997 found Q
10
in the range 1.8–3.8 for soil tem- peratures at 5 cm depth and Goulden and Crill 1997
and Lavigne et al. 1997 found Q
10 e
in the range 1.7–3.7 for soil temperatures in the range 4–10 cm.
Moreover, Pajari 1995 reported average respiration rates in the range 0.006–0.046 mg m
− 2
s
− 1
for mean air temperatures in the range −17 to 17
◦
C. The respi- ration rate of 0.19 mg m
− 2
s
− 1
at T
s
= 10
◦
C found in this study Fig. 6a, was at the higher end of the range
of values encountered in these studies, while the Q
10 e
value 4.8 was outside the range. Lindroth et al. 1998 showed for the present stand that respiration
rates scaled to stand level agreed well with respiration rates measured above the stand. At the ecosystem
level, they reported a Q
10 e
of 2.63 and a respiration rate at 10
◦
C of 0.19 mg m
− 2
s
− 1
, using air temper- ature above the canopy as predictor. Corresponding
figures for forest floor respiration, with chamber tem- perature as predictor, were 1.89 and 0.21 mg m
− 2
s
− 1
Fig. 6c, respectively, indicating a higher base respi- ration rate, but a less steep increase in respiration rate
with increasing temperature. Thus respiration rates reported here were large in comparison with those
from other studies, but within reasonable limits in relation to respiration rates at the ecosystem level. As
was pointed out by Rayment and Jarvis 1997, com- parison of respiration rates and Q
10
values is difficult, because both depend on the temperature from which
they are derived. Derivation of Q
10
from air tem- perature generates a lower Q
10
, because of the large diurnal temperature variation, than Q
10
derived from soil temperatures, where the range of diurnal varia-
tion is smaller Fig. 6a–c. Therefore modelling of the respiration component of stand fluxes requires that
Q
10
or Q
10 e
be used with the temperatures at which they were derived. Other use of Q
10
may result in erroneous estimates of respiration and consequently
of photosynthetic uptake. Some of the scatter in the relationship between
respiration and temperature could be explained by seasonal variation in the base respiration rate, and pos-
sibly also in Q
10 e
. Lavigne et al. 1997 found signifi- cant seasonal variation in the respiration–temperature
relationship at three boreal forest sites out of six. At the forest ecosystem level, Goulden et al.
1997 found that the base respiration rate varied throughout the season, while Q
10
was constant. In both cases, seasonal variation was added to an em-
pirical model as a day-number dependent parameter. Such a model is a valuable tool for filling in missing
data, and for analysing data for the period during which measurements were made, but is of less, or no
use for temporal and spatial extrapolation. Goulden et al. 1998 divided soil respiration into shallow and
deep-soil respiration. They found a strong correlation both between shallow-soil respiration and the tem-
perature of the uppermost soil layers, and between deep-soil respiration and soil temperature at 50 cm.
This implies that soil respiration depends on the tem- perature of the whole soil volume, which cannot be
described by the temperature of a single layer. It is beyond the scope of this paper to go deeper into
these questions but there is no question that there is room for a lot of development in this field of
research.
The net CO
2
efflux for the forest floor, accumu- lated from May to October, was ca. 3.1 kg m
− 2
. Dur- ing 1 week in June, photosynthetic uptake reduced
simultaneously respired CO
2
by ca. 28 Fig. 7c. This proportion probably did not remain constant
throughout the growing season, but was most likely controlled by day length, and therefore decreased
from June to October. Assuming that photosynthetic uptake from May to October reduced, on average,
respired CO
2
by 20, respiration would be 3.7 kg and photosynthetic uptake 0.6 kg. If we further as-
sume that the respiration rate of the forest floor during the winter equals the efflux rate at the beginning of
May, i.e. 0.05 mg m
− 2
s
− 1
, then the respiration from January to April and November to December would
be ca. 0.8 kg m
− 2
of CO
2
. Although respiration in November may be higher, this is probably an over-
estimate, because of the low temperatures in winter; chamber measurements in February–March indicated
soil CO
2
efflux for temperatures below 0
◦
C was less than 0.05 mg m
− 2
s
− 1
. Nevertheless, this assumption gives a total annual forest floor respiration of 4.5 kg
CO
2
m
− 2
. This is in agreement with Lindroth et al. 1998, where the total respiration soil, branch, and
needle for this forest in 1995 was 5.5 kg m
− 2
CO
2
. Compared to the mean annual respiration for boreal
forests, of 1.2 range 0.4–2.0 kg CO
2
m
− 2
per year, as reviewed by Raich and Schlesinger 1992, the
respiration rates in this study are, once again, large.
12 A.-S. Mor´en, A. Lindroth Agricultural and Forest Meteorology 101 2000 1–14
Lindroth et al. 1998, however, showed that the present forest ecosystem acted, during a 2-year pe-
riod, as a source of carbon and not as a sink, contrary to what is usually believed of boreal forests.
4.2. Within-day dynamics The soil chambers were transparent, and measured
CO
2
exchange over moss and Vaccinium surfaces. Therefore, the diurnal variation included photosyn-
thetic uptake during daytime. The respiration term was always larger than the uptake term, thus there
was a continuous efflux of CO
2
from the forest floor to the atmosphere. Goulden and Crill 1997
found similarly, for feather-moss sites, a continuous efflux of CO
2
, while sphagnum-moss sites exhib- ited larger uptake rates, which at mid-day frequently
offset respiration rates. Gross mid-day photosyn- thesis ranged from 0.5 to 1.0
m
mol m
− 2
s
− 1
at the feather-moss sites and from 0.5 to 2.5
m
mol m
− 2
s
− 1
at the sphagnum-moss sites. In the present study, gross mid-day photosynthetic rates for the short pe-
riod in June were in the range 0.05–0.19 mg m
− 2
s
− 1
i.e. 1.1–4.3
m
mol m
− 2
s
− 1
Fig. 8, in other words, higher rates, but still of the same magnitude as those
from the feather-moss and sphagnum-moss sites. In the present stand the dry mass of moss per unit area
of ground varied on average between 50−150 g m
− 2
. Hence, the photosynthetic rate of the moss layer
would have to be in the range 7–22, and 29–86 nmol CO
2
g photosynthesising moss
− 1
s
− 1
, in order to correspond to a photosynthetic uptake rate of 0.05
and 0.19 mg m
− 2
s
− 1
, respectively. Stålfelt 1937, as referred to by Kallio and Kärenlampi 1975 found
maximum values of net photosynthetic rates in forest mosses in southern Sweden in the range 13–22 nmol
CO
2
g
− 1
s
− 1
. Furthermore, the contribution from the vascular plants in relative terms can be considerable.
Wielgolaski 1975, reported for V. myrtillus leaves a net assimilation rate of 69 nmol CO
2
g
− 1
s
− 1
. There- fore, the uptake rate under light-saturated conditions,
of 0.19 mg m
− 2
s
− 1
, is not an unrealistic estimate of gross photosynthesis.
As was pointed out in Section 2 Fig. 2b, a peak, of varying size and not directly related to temper-
ature, often occurred near sunset. Baldocchi et al. 1986, measuring CO
2
efflux from a forest floor by eddy-covariance, similarly found that soil respiration
exhibited a burst at dusk, which often was greater than the highest respiration rates measured in the
afternoon. A likely explanation for this was thought to be the rapid catabolism of translocated carbohydrate
Edwards and McLaughlin, 1978. Later, Baldocchi and Meyers 1991 suggested that their measure-
ments were not reliable, because of non-steady-state conditions caused by a rapid build-up of carbon con-
centrations in the trunk space, and should have been discarded. However, it is interesting to note that the
two inherently different measurement systems re- vealed a similar diurnal pattern. Since measurements
with the soil chamber within 30 cm of the soil surface showed a similar dusk burst of CO
2
, the phenomenon cannot satisfactorily be dismissed as the result of
unreliable measurements. To improve understanding of within-day CO
2
exchange at the forest floor, this process therefore must be studied further.
4.3. Chamber system Compared to the static and dynamic chambers com-
monly used for soil respiration studies, the continu- ously measuring open-chamber system applied in the
present study had several advantages: i it could be left unattended for extended periods, ii the cham-
bers were transparent, which allowed measurement of the net CO
2
exchange of the forest floor, and iii the chambers covered a relatively large surface area,
thereby minimising ‘edge effects’ and could, in con- trast to small chambers, cover impermeable areas such
as rocks or larger roots near the surface cf. Norman et al., 1997. An advantage of this system, compared to
eddy-covariance systems, was that the chambers cov- ered a well-defined surface area, but at the price of
slight disturbance of the climate within the chamber. The canopy was, however, practically closed, and the
light penetrating to the forest floor only occasionally exceeded 200
m
mol m
− 2
s
− 1
Fig. 3, so that temper- ature and the vapour-pressure deficit inside the cham-
bers were not noticeably affected. But it remains to study how the chamber design affects natural pres-
sure fluctuations, which are of great importance to the transfer of CO
2
from the soil to the atmosphere e.g. Rayment and Jarvis, 1997.
A.-S. Mor´en, A. Lindroth Agricultural and Forest Meteorology 101 2000 1–14 13
5. Concluding remarks