Open Journal of Soil Science, 2012, 2, 187-195
http://dx.doi.org/10.4236/ojss.2012.22023 Published Online June 2012 (http://www.SciRP.org/journal/ojss)
187
Field Soil Respiration Rate on a Sub-Antarctic Island:
Its Relation to Site Characteristics and Response to
Added C, N and P
Andrea Lubbe, Valdon R. Smith
Department of Botany & Zoology, Stellenbosch University, Stellenbosch, South Africa.
Received February 13th, 2012; revised March 14th, 2012; accepted March 30th, 2012
ABSTRACT
Botanical, soil chemistry and soil microbiology variables were tested as predictors of in situ soil respiration rate in the
various terrestrial habitats on sub-Antarctic Marion Island (47˚S, 38˚E). Inorganic P and total N concentration were the
best predictors amongst the chemistry variables and bacteria plate count the best of the microbiology variables. How-
ever, while these chemistry and microbiology variables could accurately predict soil respiration rate for particular habi-
tats, they proved inadequate predictors across the whole range of habitats. The best suite of predictors comprised only
botanical variables (relative covers of five plant guilds) and accounted for 94% of the total across-habitat variation in
soil respiration rate. Mean field soil respiration rates (2.1 - 15.5 mmol CO2 m–2·h–1) for habitats not influenced by sea-
birds or seals are similar to rates in comparable Northern Hemisphere tundra habitats. Seabird and seal manuring en-
hances soil respiration rates to values (up to 27.6 mmol CO2 m–2·h–1) higher than found at any tundra site. Glucose, N, P
or N plus P were added to three habitats with contrasting soil types; a fellfield with mineral, nutrient-poor soil, a mire
with organic, nutrient-poor soil and a shore-zone herbfield heavily manured by penguins and with organic, nutrient-rich
soil. Glucose addition stimulated soil respiration in the fellfield and mire (especially the former) but not in the coastal
herbfield soil. N and P, alone or together, did not stimulate respiration at any of the habitats, but adding glucose to fell-
field soils that had previously been fortified with P or NP caused a similar increase in respiration rate, which was
greater than the increase when adding glucose to soils fortified only with N. This suggests that fellfield soil respiration
is limited by P rather than N, and that there is no synergism between the two nutrients. For the mire and coastal
herbfield, adding glucose to soils previously fortified with N, P or NP did not enhance rates more than adding glucose
to soils that had received no nutrient pre-treatment.
Keywords: Soil Respiration; Sub-Antarctic Island; Soil Moisture Content; Soil Nutrient Status; N Limitation;
P Limitation; C Limitation; Seal and Seabird Manuring
1. Introduction
The fauna of sub-Antarctic Marion Island (47˚S, 38˚E)
comprises few grazer or predator species so most energy
flow and nutrient cycling occurs in a detritus, rather than
a grazing, foodweb. Decomposition is thus crucial to
ecosystem functioning on the island. Consequently, sub-
stantial effort has gone into studies of decomposition-
related phenomena, such as the size and activity of
island’s soil microorganism populations [1-3], rates of
cellulose decomposition [4], the influence of invertebrate
detritivores on rates of carbon and nutrient mineralisation
[5,6] and soil respiration [7,8].
Measurements of CO2 evolution from soil samples in-
cubated in the laboratory [7] showed a pattern of soil
respiration rate across the island’s terrestrial habitats that
correlate well with the patterns of variation in the soil,
botanical and ecological attributes used to define the ha-
bitats [9]. Dry, mineral fellfield soils possess the lowest,
and organic soils of habitats heavily influenced by sea-
bird or seal manuring the highest, respiration rates. Whe-
ther manuring stimulates soil respiration by improving
the inorganic nutrient status of the soil or by supplying it
with easily-respirable carbon sources, or both, was tested
by incubating soil samples with added N, P and/or glu-
cose [8]. Glucose markedly stimulated soil respiration
rate in all the soils, suggesting that the primary factor li-
miting soil microbial activity on the island is labile car-
bon substrate. However, soil N and P status was also im-
portant, since adding N and P to soils with especially low
endogenous N and P concentrations stimulated respira-
tion, and adding glucose plus N and P to soils with low N

Field Soil Respiration Rate on a Sub-Antarctic Island: Its Relation to Site Characteristics and Response to
Added C, N and P
188
and P status resulted in a significantly greater stimulation
of respiration rate than adding glucose alone. For all soil
types, respiration rate increased with moisture content up
to full moisture holding capacity [8].
All the above findings were from laboratory incuba-
tions and thus subject to the usual limitations concerning
conclusions based on soil respiration measurements
made on excised, stored and homogenised soil samples.
Here, we report the across-habitat variation in field-mea-
sured rates of soil respiration and, for three habitats with
different nutrient statuses, how in situ soil respiration
responds to N, P and glucose addition. A goal of the re-
search program at Marion Island is to model carbon ex-
change for the various terrestrial habitats and for the is-
land as a whole. This requires an ability to predict in situ
soil respiration rate from the abiotic and biotic factors
that affect it. We thus relate the field-measured soil re-
spiration rates to site factors such as botanical compo-
sition, soil chemistry and soil microorganism counts, to
see whether any of these factors, or combinations of
them, can successfully predict soil respiration rate.
2. Sites, Materials and Methods
2.1. Field Measurements of Soil Respiration at
the Different Habitats
Sampling was carried out over about 100 km2 of the is-
land’s western, northern and eastern lowland plains, at 5
to 11 representative examples (sites) of 19 of the island’s
23 habitats. A detailed description of the vegetation and
soils of the habitats may be found in [9] and a synopsis
of their main characteristics is given in Table 1. At each
site, four areas large enough to accommodate a SRC-1
soil respiration chamber (PP Systems, U.K.) were cleared
by hand-plucking the vegetation. The wet, peaty soils
resulted in most roots coming out attached to the above-
ground shoots. Conspicuous remaining roots (mostly
these were ones attached to surrounding vegetation) were
excised with a scalpel blade. Root respiration would have
contributed to the measured CO2 evolution rates at all the
sites (except possibly the fellfield ones which had a very
sparse vegetation cover) but, with the exception dis-
cussed later, it was likely a minor component. About 30
minutes after clearing the vegetation, the respiration
chamber, connected to an EGM-2 CO2 analyzer (PP Sy-
stems UK), was placed on each cleared space, pushed
about 1 cm into the soil, and the increase in CO2 concen-
tration in the chamber monitored for 2 minutes. Soil re-
spiration rate was calculated from the increase in con-
centration, the system volume and the area of soil en-
closed by the chamber. The mean value for the four
cleared areas was taken as the soil respiration rate for the
particular site. Soil temperature of each cleared area was
also measured (chromel-alumel thermocouple; values
were from 3.4˚C to 12.7˚C, 90% were between 6˚C and
11˚C) and the respiration rate converted to a rate at 10˚C
using the average Q10 of 2 found [7] for soil respiration at
the island.
2.2. Prediction of Soil Respiration Rate from
Edaphic and Botanical Variables
Stepwise multiple regression of the habitat-mean field
soil respiration rates in Table 1 against the habitat-means
of a set of 23 soil chemistry variables, 17 botanical vari-
ables (relative covers of plant guilds, total vegetation
cover), two topographical variables (altitude and distance
from the shore), and three soil microorganism variables
(total and plate counts of bacteria and plate counts of
fungi) was used to identify a set of predictors of soil res-
piration rate. The habitat-means of these variables are
given in [7]. The plant guilds (based mainly on plant
growth form but also on taxonomic and ecological attri-
butes) are described in [9].
2.3. Assessment of the Effect of Added Nutrients
and Glucose on Soil Respiration Rate
Three sites representing contrasting habitats were se-
lected; a Mesic fellfield (inland/dry/mineral, low nutrient
status soil/low soil respiration rate), a Mesic mire (in-
land/wet/organic, low nutrient status soil/medium respi-
ration rate), and a Cotula herbfield (coastal/wet/organic,
high nutrient status soil/high respiration rate). At each
site on day 1, 40 areas just large enough to accommodate
a SRC-1 chamber were cleared of vegetation. Soil respi-
ration rate was measured every day on each cleared area.
After the measurement on day 5, the cleared areas were
randomly assigned into 5 groups. One group was watered
with a solution of NH4NO 3 (200 μg N per cm2 cleared
surface), one with a solution of KH2PO4 (200 μg·P·cm–2),
one with NH4NO3 and KH2PO4 (200 μg·N & 200
μg·P·cm–2), one with glucose (5 mg·G·cm–2) and the re-
maining group served as a control that received only wa-
ter. A total of 25 ml solution was added to each cleared
area. Respiration rates were monitored daily. After the
measurement on day 8, all the groups were watered with
glucose solution (5 mg·G·cm–2) and respiration rate mo-
nitored for 3 more days (2 days only at the Cot ula her-
bfield). All rates were corrected to a 10˚C value using a
Q10 of 2. The significance of the differences in respire-
tion rate before and after nutrient additions were assessed
by Analysis of Variance and Tukey’s Honest Signifi-
cance Difference Tests).
2.4. Soil Moisture and Chemical Analysis
Four 5 cm deep soil cores were taken from each of the

Field Soil Respiration Rate on a Sub-Antarctic Island: Its Relation to Site Characteristics and Response to
Added C, N and P
189
Table 1. Synopsis of the main features of the terrestrial habitats on Marion Island and their mean (±standard error) soil res-
piration rate. In the first column, an asterisc* indicates that the habitat is heavily influenced by seabird and/or seal manuring
and the number in brackets shows at how many examples of the particular habitat respiration measurements were taken.
The habitats are assigned to a high (H), medium (M) or low (L) soil respiration group based on Anova and the Tukeys honest
significant difference test. Tot. (cation) = total concentration of the particular cation in the soil; exch. (cation). = exchangeable
concentration; sol. (cation) = soil solution concentration.
Habitat General Dominant plant species Soil Chemical characteristic
of top soil layer
Soil respiration rate
(mmol·CO2·m–2·h–1)Group
1.1 Coastal
herbfield (9) Shore-zone Crassula moschata, sometimes Cotula plumosa Fibrous peat
Very high tot. Na, exch.
Na & Mg, sol. Na &
Mg
5.0 ± 0.6 M
1.2 Coastal
fellfield (6)
Shore-zone,
exposed C. moschata, Azorella selago Fibrous peat, volcanic ash As 1.1 7.1 ± 2.1 M
2.1 Xeric
fellfield (8)
Exposed; Sparse
vegetation;
Generally > 300
m altitude
A. selago, cushion and
ball-forming mosses, lichens
Skeletal, Volcanic
ash and rock
Mineral, basic, high
bulk density, high tot.
Ca and Mg, low
organic C & N, dry
2.1 ± 0.3 L
2.2 Mesic
fellfield (11)
Exposed; Sparse
vegetation;
Generally < 200
m altitude
A. selago, often Blechnum
p
enna-marina, Agrostis
magellanica, cushion and
ball-forming mosses, lichens
Volcanic ash and rock As 2.1 but slightly
more organic and moist 2.4 ± 0.4 L
3.1 Open
fernbrake (7)
Succession
between 2.2
and 3.2
A. selago, B. penna-marina
Ag. magellanica, Acaena
magellanica
Organic surface
layer, below that
similar to 2.2
Considerably more
organic and moist
than 2.2
6.3 ± 1.0 M
3.2 Closed
fernbrake (7)
Dominant habitat
on slopes
Continuous B. penna-marina
carpet, occasional Ag.
magellanica, Ac. magellanica
Poa cookii
Deep, well-developed
horizons; Highly
organic surface layer
Moist, high organic C,
high C.E.C., moderately
high inorganic P
5.5 ± 0.5 M
3.3 Mesic
fernbrake (5)
Similar to 3.2 but
on wetter,
less-steep slopes
Continuous B. penna-marina
carpet; Ag. magellanica,
Uncinia compacta, bryophyte
species common
Deep, poorer horizon
differentiation than 3.2;
Highly organic surface
layer
Wet, high organic C,
high C.E.C., low
inorganic P
8.0 ± 1.7 M
3.4 Dwarf
shrub fernbrake
(8)
More sheltered
and wet than 3.2
Ac. magellanica,
B. penna-marina, mosses
(especially Brachythecium sp.)
Similar to 3.3 Similar to 3.3 7.1 ± 2.0 M
3.5 Slope
drainage line
and streambank
(6)
Drainage lines on
slopes and on
banks of streams
Ac. magellanica, mosses
(especially Brachythecium
rutabulum and Sanionia
uncinatus)
Similar to 3.3
Higher pH, tot. &
exch.Ca & Mg, lower
organic C than other
fernbrake habitats
15.5 ± 2.6 H
4.1 Coastal
tussock
grassland* (6)
Coastal slopes
heavily
influenced by
penguins, petrels,
seals.
P. cookii on low peat pedestals;
frequent but low cover C.
pl
umosa, Callitriche antarctica,
Montia fontana, Poa annua
Compact fibrous peat,
decomposing tussock
bases
Acid, organic, very
high tot. N, inorganic
N & P
7.7 ± 1.7 M
5.1 Cotula
herbfield* (10)
Most common
coastal area
habitat, heavily
influenced by
seabirds and seals
C. plumosa, often with P. cookie
co-dominant, other species
infrequent
Compact peat
Very high tot. and
inorganic N and P. High
tot. Na, xch. Na & Mg,
sol. Na & Mg
27.6 ± 2.9 H
5.2 Biotic
mud* (8)
In and around
seal wallows,
penguin
rookeries, also
sometimes around
albatross nests
C. antarctica, sometimes
also M. fontana
Eutrophic, very wet,
generally anaerobic
mud
Highly organic; very
high inorganic N & P 20.8 ± 2.2 H
5.3 Biotic
lawn* (7) As 5.2
Poa annua dominant; P. cookii,
C. plumosa common;
sometimes C. antarctica and
M. fontana
Thin, well-drained
fibrous peat underlayed
by scoria; sometimes
just scoria
Less organic and lower
inorganic N & P than
5.2;
b
ut dryer and highe
r
tot. Ca & Mg.
23.2 ± 3.6 H

Field Soil Respiration Rate on a Sub-Antarctic Island: Its Relation to Site Characteristics and Response to
Added C, N and P
190
Continued
6.1 Dry mire
(8)
Transition
between 6.2 and
3.3, or between
6.2 and 2.2
Ag. magellanica, U. compacta,
B. penna-marina, bryophtes
(mainly Racomitrium
lanuginosum, Jamesoniella
colorata, Ptychomnion
ringianum)
Dry oligotrophic peats
Dryer, less organic, and
higher tot. Ca & Mg and
exch. Ca than other mire
habitats
6.0 ± 0.8 M
6.2 Mesic mire
(8)
Boggy grassland
vegetation
Greater dominance of
graminoids (Juncus
scheuchzerioides, Ag.
magellanica, U. compacta),
lesser of bryophytes,
compared with 6.3
Wet dystrophic
peats, deeper than 6.1
Wetter, more organic
than 6.1 5.3 ± 0.7 M
6.3 Wet mire
(6) Bog Bryophytes overwhelmingly dominant
Waterlogged peat,
water table mostly at
or above the surface
Extremely wet, organic,
inorganic N & P higher
than 6.1 or 6.2
7.4 ± 1.7 M
6.4 Mire
drainage line
(11)
Bog in water
tracks
Bryophytes overwhelmingly
dominant
More mineral than
other mire habitats,
large and rapid
fluctuations moisture
content
Extremely low inorganic
P; highest pH for mire
habitats
5.6 ± 0.9 M
6.5 Biotic
mire* (7)
Bog influenced
by manuring
Clasmatocolea vermicularis
dominant; Ag. magell an ic a and
P. cookii common; M. Fontana
frequent but low cover
Eutrophic, very
wet peat
Highest inorganic N & P
for mire habitats 17.0 ± 3.8 H
6.6 Saline mire
(7)
Bog influenced
by salt-spray
C. vermicularis dominant;
C. moschata and Ag.
magellanica common
Very wet peat
Highest tot., exch. And
sol Na, exch. and sol.
Mg of all habitats
except 1.1 and 1.2,
relatively high
inorganic N & P
7.0 ± 1.2 M
three sites where the nutrient addition experiments were
carried out. Half of each core was weighed, dried at
105˚C for 48 hour and reweighed to assess moisture con-
tent. The rest of the core was air dried and used to deter-
mine total carbon and nitrogen (TruSpec CHN analyser,
Leco Corporation, MI, USA) and total phosphorus (by
dry-ashing a subsample, dissolving the ash in dilute HCl,
and measuring the P concentration in the solution with a
Vista ICP-Optical Emission Spectrometer (Varian Inc.,
CA, USA).
3. Results
3.1. Across-Habitat Variation in Soil Respiration
Rate
Mean field soil respiration rate varies by an order of
magnitude across habitats (Table 1). Lowest values are
for inland fellfields and highest are for habitats heavily
manured by seabirds or seals. The various fernbrake and
mire habitats (excluding biotic mire) all have quite simi-
lar, moderately-low respiration rates. The Cotula herb-
field showed the highest mean rate but root respiration
might have contributed significantly since not all of the
large, fleshy Cotula plumosa rhizomes could be removed
without disturbing the soil to an unacceptable degree.
However, laboratory measurements on soil samples from
which all roots were removed showed Cotula herbfield
soils to be very active, with the second highest mean in
vitro respiration rate amongst the habitats [7]. Overall,
the ranking of habitats on field respiration rate is similar
to the ranking on laboratory measured rates. The only
serious discrepancy is for the Slope drainage line habitat,
which is in the upper part of the range of mean field res-
piration rates but was found to be in the lower part of the
range of laboratory respiration rates. The surface layer of
slope drainage line soils is a deep loose mat of decom-
posing litter, rather than soil proper. This layer would
have contributed substantially to the field measurements
whereas the laboratory measurements were made on the
underlying soil.
The Saline mire, Coastal herbfield and Coastal tussock
grassland exhibit surprisingly low field soil respiration
rates considering that all three habitats occur in the shore
zone and are thus to some extent affected by seals and
penguins (most stands of the last mentioned habitat are
generally very heavily affected). Possibly, unfavourable
soil moisture content is responsible. Mean rates for all
the coastal habitats are plotted against soil moisture con-
tent in Figure 1(a) and it is clear that the Coastal herb-
field (1.1) and Tussock grassland (4.1) have suboptimal,

Field Soil Respiration Rate on a Sub-Antarctic Island: Its Relation to Site Characteristics and Response to
Added C, N and P
191
400 600 8001000 1200 1400 1600 1800
Soil moisture content (%)
0
5
10
15
20
25
30
1.1
6.6
6.5
5.2
5.1
5.3
4.1
(a)
Soil respiration rate (mmol CO
2
m
–2
·h
–1
)
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Soil moisture content (%)
0
2
4
6
8
10
12
14
16
2.1
6.3
6.2
6.4
6.1
3.5
3.3
3.4
1.2
3.1
3.2
2.2
(b)
Soil respiration rate (mmol CO
2
m
–2
·h
–1
)
Figure 1. Habitat-mean field soil respiration rate versus
habitat-mean soil moisture content for (a) coastal habitats
and (b) inland habitats. Habitat numbers are as in Table 1.
Mean soil moisture contents are from [17].
and the Saline mire (6.6) supraoptimal, moisture contents
for soil respiration.
3.2. Prediction of Soil Respiration Rate from
Botanical and Soil Characteristics
Figure 1(a) shows that soil moisture content is a strong
determinant of field soil respiration rate for the coastal
habitats; rates increase sharply with moisture up to an
optimal moisture content between 650% and 750% and
then decline quite rapidy at higher moisture levels.
Amongst the non-coastal habitats (Figure 1(b)) respira-
tion also increases exponentially with moisture content,
again up to maximum values between 650% and 750%,
but the decline in rate above optimum moisture content is
much less pronounced; in fact, the wet mire habitat (6.3)
shows a mean respiration rate about 1/3 higher than most
of the other inland mire habitats (6.1, 6.2, 6.4), all of
which had drier soils.
Despite the clear relationship between respiration rate
and soil moisture, moisture content is a poor predictor of
respiration rate across all the habitats, mainly because
across the whole range of moisture contents there is a
disparity in rates between coastal (manured) and inland
(non-manured) habitats. Of the other soil variables, inor-
ganic P concentration is the most useful predictor of soil
respiration rate in the field, followed by total N concen-
tration. Multiple linear regression analysis showed that
the two together accounted for 56% (P < 0.001) of the
across-habitat variation in respiration rate. Respiration
rate is also significantly correlated with (log) plate count
of soil bacteria (r2 = 0.44, P = 0.001). However, while
regression models with only soil chemistry and/or soil
microorganism variables accurately predict soil respira-
tion rate for particular habitats, they prove inadequate
when applied across all the habitats, and perform par-
ticularly poorly in the case of the habitats with very low
respiration rates.
The suite of variables that best predict field respiration
rate across the whole suite of habitats comprises only
botanical characteristics—the relative covers of mat-
forming dicotyledons, rosette-forming dicotyledons, de-
ciduous shrubs, pteridophytes and cushion-forming di-
cotyledons. Mat-forming dicot (log) cover alone accounts
for 50 % of the across-habitat variation in soil respiration
rate but is not a useful predictor since that plant guild is
absent from some habitats. The regression model based
on the five plant guilds mentioned above accounts for
94% of the variation in respiration rate and successfully
predicts rates for most of the habitats (Figure 2). It un-
derpredicts rates for Mesic fernbrake (3.3) by about 33%
051015 20 25 30
0
5
10
15
20
25
30
Observed mmol CO
2
m
-2
h
-1
)
5.1
5.3
3.3
2.1 & 2.2
3.4 & 6.6
4.1
6.5
5.2
Re sp
10
= 9.4 93 log Matdi cot
+ 0.248 Rosedicot
+ 0.201 Decshrub
+ 0.028 Pterido
- 0.039 Cushdico
+ 3.570
r
2
= 0.94
Observed rat e (mmol CO
2
m
–2
·h
–1
)
Predicted rate (mmol CO2 m–2·h–1)
Figure 2. Predicted and observed habitat mean soil respire-
tion rates for a regression model using the relative covers of
five plant guilds as predictors. The solid line indicates per-
fect prediction and the dashed lines the 95% confidence
limits of the predictions. The slope, intercept and determi-
nation (r2) coefficients are shown. Habitats mentioned in the
discussion of the model are identified by their number (Ta-
ble 1).

Field Soil Respiration Rate on a Sub-Antarctic Island: Its Relation to Site Characteristics and Response to
Added C, N and P
192
and for Biotic mire (6.5) by about 22%. It overpredicts
rates for tussock grassland (4.1) by about 20% and for
Dwarf shrub fernbrake (3.4) and Saline mire (6.6) by
about 30%. Overall, however, the plant guild-based mo-
del performs well, predicting rates for the low fertility,
low respiration rate habitats (2.1. 2.2) as well as for the
manured/ high respiration rate habitats (5.1, 5.2, 5.3)
quite accurately. Adding soil chemistry or microbiology
variables to the suite of botanical variables in Figure 2
did not strengthen the predictive ability of the model.
3.3. Influence of Added N, P and Glucose on
Respiration Rate
Soil moisture and nutrient status differed considerably
between the three habitats chosen to assess the effect of
N, P and glucose (G) addition on soil respiration rate
(Table 2). The fellfield soil was driest, least organic and
contained the lowest contents of total N and P. The mire
soil was the wettest, had about twice as much organic
matter and total N, but only a slightly higher total P con-
centration, than the fellfield soil. The Cotula herbfield
was very heavily influenced by Gentoo Penguins and its
soil was intermediate in moisture, but had significantly
higher C, N and P concentrations than the other two soils.
C:N ratios were similar for the three soils but the mire
soil had significantly greater C:P and N:P ratios than the
fellfield or Co tul a herbfield soils.
Table 3 shows the soil respiration rates at the three
habitats before and after adding H2O, N, P, NP or G to
the surface of the particular treatment site on day 5, and
Table 2. Moisture, total carbon, total nitrogen and total phosphorus at the three habitats in which the effect of adding N, P
and glucose on soil respiration rate was assessed. Values are means ± standard deviations (N = 4) and are on a dry soil mass
basis. Different superscripts indicate that the habitat means are significantly (P ≤ 0.05) different (Anova and Tukey’s honest
significant difference test).
Habitat Moisture (%) C (%) N (%) P (%) C:N C:P N:P
Mesic fellfield a257 ± 39 a21 ± 1.5 a1.1 ± 0.12 a0.10 ± 0.013 a20 ± 1.1 a214 ± 14 a11 ± 0.7
Mesic mire c1417 ± 112 b44 ± 2.8 b2.4 ± 0.31 a0.15 ± 0.026 a18 ± 1.5 b291 ± 36 a16 ± 0.7
Cotula herbfield b660 ± 151 c55 ± 1.8 c3.2 ± 0.21 b0.31 ± 0.043 a17 ± 1.1 a181 ± 28 a10 ± 1.1
Table 3. Mean (± standard error, N = 8) soil respiration rates (mmol CO2 m–2·h–1) before and after adding nutrients or glu-
cose. On day 1 the vegetation was cleared from the localities at which respiration was measured. After measuring respiration
on day 5, the nutrient indicated in the column heading was added. Day 6 was 24 hours after, and day 8 was 72 hours after,
that addition. After the measurement on day 8, glucose was added to all the soils. Day 9 was 24 hours after, and day 11 was 72
hours after, the glucose addition. Different superscripts indicate that the means before and after nutrient/glucose addition are
significantly (P ≤ 0.05) different from each other (Anova and Tukey’s honest significant difference test).
Habitat Day Control (H2O) N P NP Glucose
Fellfield 5 1.7 ± 0.39 2.1 ± 1.22 2.1 ± 0.83 2.0 ± 0.93 a1.8 ± 0.45
6 1.6 ± 0.37 2.4 ± 0.93 1.6 ± 0.67 1.4 ± 0.92 b5.0 ± 1.11
8 2.4 ± 0.49 2.5 ± 1.12 2.2 ± 0.90 1.8 ± 0.69 ab2.8 ± 0.71
8 a2.4 ± 0.49 a2.5 ± 1.12 a2.2 ± 0.90 a1.8± 0.69 a2.8 ± 0.71
9
b5.7 ± 0.85 b6.3 ± 0.70 b8.1 ± 2.11 b6.6 ± 0.89 b7.0 ± 1.32
11
ab3.3 ± 0.81 ab3.9 ± 1.04 ab3.3 ± 1.13 ab3.5 ± 1.36 ab4.3 ± 1.34
Mire 5 5.9 ± 0.99 6.8 ± 1.74 6.5 ± 1.35 6.7 ± 1.36 6.2 ± 1.59
6 5.2 ± 0.76 5.1 ± 1.18 5.6 ± 0.78 5.3 ± 1.21 8.5 ± 1.31
8 6.3 ± 1.07 6.5 ± 0.84 8.2 ± 1.33 7.6 ± 0.90 7.6 ± 0.88
8 6.3 ± 1.07 6.5 ± 0.84 8.2 ± 1.33 7.6 ± 0.90 7.6 ± 0.88
9 10.4 ± 1.44 9.0 ± 1.31 10.5 ± 1.74 9.3 ± 1.00 9.7 ± 0.65
11 9.5 ± 1.10 7.8 ± 1.27 9.4 ± 1.51 9.0 ± 1.06 8.3 ± 1.74
Cotula 5 22.0 ± 2.97 24.5 ± 2.43 23.9 ± 2.88 25.2 ± 3.69 21.8 ± 1.62
herbfield 6 23.1 ± 3.12 21.6 ± 2.60 25.4 ± 2.82 21.9 ± 2.80 24.3 ± 2.08
8 24.7 ± 2.93 26.2 ± 2.72 27.8 ± 2.81 29.3 ± 4.12 24.7 ± 3.25
8 24.7 ± 2.93 26.2 ± 2.72 27.8 ± 2.81 29.3 ± 4.12 24.7 ± 3.25
9 26.8 ± 3.10 30.1 ± 2.23 29.9 ± 2.78 31.2 ± 3.35 26.2 ± 2.40

Field Soil Respiration Rate on a Sub-Antarctic Island: Its Relation to Site Characteristics and Response to
Added C, N and P
193
before and after adding G to all the treatment sites on day
8. Adding N, P or NP on day 5 did not significantly
change respiration rate after 24 or 72 hours at any of the
habitats. Mean respiration rate 24 hours after adding
adding G to the fellfield soil was nearly 3 times higher
than prior to the addition, and 48 hours after that it had
still not returned to the pre-addition value. Adding G on
day 5 did not significantly enhance respiration rate of the
mire or Cotula herbfield soils. However, for the mire, the
fact that mean rate 24 h after G addition was nearly 40%
greater than prior to the addition, whereas mean rates 24
h after adding H2O, N, P or NP were all lower than be-
fore, suggests that glucose might have had a stimulatory
effect on mire soil respiration.
At the fellfield site, adding glucose on day 8 resulted
in a strong stimulation of respiration by day 9, regardless
of whether the soils had previously received H2O, N, P,
NP or G, and the effect was still present after 72 hours.
At the mire too, adding G on day 8 resulted in an in-
creased respiration rate 24 hours later, although the dif-
ferences for individual pre-treatment groups were not
significant at P ≤ 0.05. For all pretreatments together the
effect was highly significant; glucose addition on day 8
increased mean respiration rate by nearly 40% at the mire,
from 7.2 ± 0.45 mmol CO2 m
–2·h–1 to 9.8 ± 0.55 mmol
CO2 m–2·h–1 (N = 40, P < 0.001). At the Cotula herbfield
the overall mean rate on day 9 (28.8 ± 1.23 mmol CO2
m–2·h–1) was not significantly (P ≤ 0.05) different to that
on day 8 (26.5 ± 1.39 mmol CO2 m–2·h–1).
4. Discussion
This is the first report of in situ soil respiration rates for a
sub-Antarctic island. Rates found for the non-manured
habitats are similar to those reported for comparable
Northern Hemisphere tundra vegetation types. For in-
stance, the rates for inland fellfields (2.1 and 2.4 mmol
CO2 m–2·h–1) are similar to those found for an Arctic li-
chen heath (2.5 mmol CO2 m–2·h–1; [10]) and Arctic po-
lar deserts (1.6 to 5.2 mmol CO2 m
–2·h–1; [11]. Inland
habi- tats with a closed vegetation cover comprised of
grami- noids, forbs and bryophytes and that are not in-
fluenced by animal manuring, such as the mires and
fernbrakes, have respiration rates (5.3 to 15.5 mmol CO2
m–2·h–1) that are within the range (2 - 17 mmol CO2
m–2·h–1) re- ported for physiognomically-similar tundra
vegetation types such as dwarf shrub tundra, wet and dry
tundra mea- dows, tussock tundra, forest tundra and taiga
([10,12-14]. Habitats on the island that are influenced by
seabird or seal manuring mostly have respiration rates
considerably higher than what has been reported for
Northern Hemi- sphere tundra.
The field respiration rates presented here are on a soil
surface area basis, whereas the laboratory rates given by
[7] are per soil mass. Since there are big differences in
soil bulk density between habitats, the across-habitat dif-
ferences in rates measured by the two techniques cannot
be simply equated to each other, but it is noteworthy that
they showed a very similar total variation (2 to 28 mmol
CO2 m
–2·h–1 for the field rates and 1 to 26 μmol CO2
g–1·h–1 for the laboratory measurements; [7], and that the
rankings of the habitats in the two data sets were, with a
few exceptions, quite similar. However, the habitats are
categorised into less precise groups by the field respira-
tion rates than by the laboratory rates. On laboratory
rates, they clearly fall into 5 groups [7]: 1) a very low
soil respiration rate group (Xeric and Mesic fellfields); 2)
a low rate group (Open fernbrake, Spring and flush, Coa-
stal fellfield and Dry mire); 3), a medium rate group
(Closed fernbrake, Mesic fernbrake, Dwarf-shrub fern-
brake, Slope drainage line and streambank, Mire drain-
age line, Mesic mire); 4) a high rate group (Wet mire,
Coastal herbfield, and Inland tussock grassland); 5) a
very high rate group (Biotic mire, Saline mire, Coastal
tussock grassland, Biotic mud, Cotu la herbfield and Bi-
otic lawn). On field rates, Anova and Tukey’s HSD test-
ing recognises only 3 groups (Table 1); a fellfield group
with very low rates (mean <3 mmol CO2 m
–2·h–1), a
group of habitats with high rates (mean >15 mmol CO2
m–2·h–1), all of which except the slope drainage line are
manured, and a group with moderately low, quite similar,
rates (mean, 5 to 8 mmol CO2 m–2·h–1; mostly these are
non-manured slopes and mires). The only anomaly in this
grouping on field rates is that Coastal tussock grassland,
which is influenced by seabirds and has a large and ac-
tive soil bacterial population and a high primary produc-
tion [2,3,15], falls into the medium rate group rather than
with the other manured habitats. As was suggested above,
low in vivo soil moisture content might be the reason for
this. The position of the Slope drainage line habitat in the
high respiration group is also somewhat surprising, but
drainage lines are minerotrophic [16], have a high pri-
mary production [15] and their soil moisture contents
that are around the optimum for soil respiration (3.5 in
Figure 1(b)).
Adding glucose to the fellfield and mire soils stimu-
lated soil respiration but adding N, P and NP did not.
However, adding glucose (on day 8) to fellfield soil that
had previously been fortified with P or NP (on day 5)
resulted in a 3.7-fold increase in respiration rate, com-
pared with 2.4- to 2.5-fold increases for the soil and that
received only water, glucose or N on day 5. This sug-
gests that, regarding inorganic nutrients, soil respiration
in the fellfield soil is limited primarily by P rather than N
and that there is no synergism between the two. Labora-
tory respiration measurements [8] also showed that add-
ing glucose to fellfield soil previously fortified with P

Field Soil Respiration Rate on a Sub-Antarctic Island: Its Relation to Site Characteristics and Response to
Added C, N and P
194
stimulated respiration rate 75% more than adding glucose
to soil fortified with N, but there was a strong synergistic
effect in the laboratory; glucose addition to soil fortified
with N and P stimulated rates 3 to 4 times more than
glucose addition to soil fortified with only N or only P.
N, P and NP addition did not affect field respiration
rate of the mire soil differently to adding water alone,
and adding glucose to mire soil previously fortified with
N, P or NP did not enhance rates more than adding glu-
cose to soil that had received no N or P fortification. This
is also different to the findings of the laboratory-incuba-
tions, where the glucose-mediated respiration response
for mire soil subsamples pretreated with N or P was
about 70% greater, and for subsamples pretreated with
both N and P nearly 300% greater, than the response to
glucose of subsamples pretreated with water [8].
For the Cotula herbfield soil, addition of N, P, NP or
glucose had no effect on respiration rate in the field. In
the laboratory incubations, glucose did stimulate respire-
tion of Cotula herbfield soil, but N and P did not [8]. In
both the laboratory and the field, adding glucose to soils
pretreated with N and/or P did not result in a greater
stimulation than adding glucose to untreated soil.
Overall, the results of adding inorganic nutrients and/
or glucose to soils in the field are less clear than what
was found previously in the laboratory, but both suggest
that the stimulatory influence of manuring by seabirds
and seals on soil respiration is primarily due to the addi-
tion of labile carbon substrate and secondarily through
the addition of inorganic nutrients such as N and P. How-
ever, manuring has a whole syndrome of consequences
that might enhance soil respiration, such as improved pri-
mary production (resulting in increased litter input and
root exudation), a higher quality of litter and soil organic
matter, larger, more active and more diverse soil micro-
bial populations better able to utilize a wider range of
organic substrates (including the more recalcitrant types),
and larger populations of microbivores that stimulate
microbial activity and turnover.
Botanical characteristics (the relative cover of five
plant guilds) proved the best predictors of soil respiration
rate measured in the field. In contrast, laboratory respira-
tion rates correlated best with soil chemistry and soil
microorganism characteristics [7]. Inorganic P alone ac-
counted for 81%, and together with (log) plate count of
bacteria for 86%, of the across-habitat variation in labo-
ratory-measured respiration rate. Adding (log) cover of
mat dicots increased the proportion of explained variance
to 93%, nearly the same as the best suite of predictors of
field rates (Figure 2). However, the overall predictive
capacity of the field-based model is considerably better
than the laboratory-based one, which seriously mispre-
dicts respiration rate for 10 of the habitats (figure 1(f) in
[7]), and performs especially poorly for habitats with low
to moderately low rates. The field respiration model se-
riously mispredicts rates for only 5 habitats and performs
well for the habitats with low to moderately low rates.
This bodes well for being able to successfully estimate
soil CO2 flux in a whole island model or carbon ex-
change, since such habitats comprise about 90% of the
island’s vegetated area.
5. Acknowledgements
The South African Department of Environmental Affairs
and Tourism provided logistical support for this study.
The Department of Botany and Zoology at Stellenbosch
University and the USAID Capacity Building Pro-
gramme of the Centre for Invasion Biology at Stellen-
bosch University financially supported A. Lubbe to carry
out this project. Elizma Yelverton of the South African
Weather Service assisted in measuring soil respiration
rates.
REFERENCES
[1] V. R. Smith and M. G. Steyn, “Soil Microbial Counts in
Relation to Site Characteristics at a Sub-Antarctic Island,”
Microbial Ecology, Vol. 8, No. 3, 1982, pp. 253-266.
doi:10.1007/BF02011429
[2] D. D. French and V. R. Smith, “Bacterial Populations in
Soils of a Sub-Antarctic Island,” Polar Biology, Vol. 6,
No. 2, 1986, pp. 75-82. doi:10.1007/BF00258256
[3] D. C. Grobler, D. F. Toerien and V. R. Smith, “Bacterial
Activity in Soils of a Sub-Antarctic Island,” Soil Biology
and Biochemistry, Vol. 19, No. 5, 1987, pp. 485-490.
doi:10.1016/0038-0717(87)90089-7
[4] V. R. Smith, M. Steenkamp and D. D. French, “Soil De-
composition Potential in Relation to Environmental Fac-
tors on Marion Island (Sub-Antarctic),” Soil Biology and
Biochemistry, Vol. 25, No. 11, 1993, pp. 1619-1633.
doi:10.1016/0038-0717(93)90018-7
[5] V. R. Smith and M. Steenkamp, “Macroinvertebrates and
Litter Nutrient Release on a Sub-Antarctic Island,” South
African Journal of Botany, Vol. 58, 1992, pp. 105-116.
[6] V. R. Smith, “Introduced Slugs and Indigenous Caterpil-
lars as Facilitators of Carbon and Nutrient Mineralisation
on a Sub-Antarctic Island,” Soil Biology and Biochemis-
try, Vol. 39, No. 2, 2007, pp. 709-713.
doi:10.1016/j.soilbio.2006.09.026
[7] V. R. Smith, “Soil Respiration and its Determinants on a
Sub-Antarctic Island,” Soil Biology and Biochemistry,
Vol. 35, No. 1, 2003, pp. 77-91.
doi:10.1016/S0038-0717(02)00240-7
[8] V. R. Smith, “Moisture, Carbon and Inorganic Nutrient
Controls of Soil Respiration at a Sub-Antarctic Island,”
Soil Biology and Biochemistry, Vol. 37, No. 1, 2005, pp.
81-91. doi:10.1016/j.soilbio.2004.07.026
[9] V. R. Smith and M. Steenkamp, “Classification of the

Field Soil Respiration Rate on a Sub-Antarctic Island: Its Relation to Site Characteristics and Response to
Added C, N and P
195
Terrestrial Habitats on Sub-Antarctic Marion Island
Based on Vegetation and Soil Chemistry,” Journal of
Vegetation Science, Vol. 12, No. 2, 2001, pp. 181-198.
doi:10.2307/3236603
[10] B. H. Svensson, A. K. Veum and S. Kjelvik, “Carbon
Losses from Tundra Soils,” In: F. E. Wielgolaski, Ed.,
Fennoscandian Tundra Ecosystems. Part 1. Plants and
Microorganisms. Ecological Studies, Vol. 16, Springer,
Berlin, Heidelberg, New York, 1975, pp. 279-286.
[11] L. Illeris, A. Michelsen and S. Jonasson, “Soil plus Root
Respiration and Microbial Biomass following Water, Ni-
trogen and Phosphorus Application at a High Arctic Semi
Desert,” Biogeochemistry, Vol. 65, No. 1, 2003, pp. 15-
29. doi:10.1023/A:1026034523499
[12] E. D. Vance and F. S. Chapin III, “Substrate Limitations
to Microbial Activity in Taiga Forest Floors,” Soil Biol-
ogy and Biochemistry, Vol. 33, No. 2, 2001, pp. 173-188.
doi:10.1016/S0038-0717(00)00127-9
[13] S. Sjögersten and P. A. Wookey, “Climatic and Resource
Quality Controls on Soil Respiration Across a Forest-
Tundra Ecotone in Swedish Lapland,” Soil Biology and
Biochemistry, Vol. 34, No. 11, 2002, pp. 1633-1646.
doi:10.1016/S0038-0717(02)00147-5
[14] C. Corradi, O. Kolle, K. Walter, S. A. Simov and E.-D.
Schulze, “Carbon Dioxide and Methane Exchange of a
North-East Siberian Tussock Tundra,” Global Change
Biology, Vol. 11, No. 11, 2005, pp. 1910-1925.
[15] V. R. Smith, “Terrestrial and Freshwater Primary Produc-
tion and Nutrient Cycling,” In: S. L. Chown and P. W.
Froneman, Eds., The Prince Edward Islands: Land-Sea
Interactions in a Changing Ecosystem, African SunMedia,
Stellenbosch, 2008, pp. 181-214.
[16] B. J. Huntley, “Vegetation,” In: E. M. Van Zinderen
Bakker, et al., Eds., Marion and Prince Edward Islands,
A.A. Balkema, Cape Town, 1971, pp. 98-160.
[17] V. R. Smith, M. Steenkamp and N. J. M. Gremmen,
“Terrestrial Habitats on Sub-Antarctic Marion Island:
Their Vegetation, Edaphic Attributes, Distribution and
Response to Climate Change,” South African Journal of
Botany, Vol. 67, No. 4, 2001, pp. 641-654.