Open Journal of Soil Science, 2012, 2, 111-115
http://dx.doi.org/10.4236/ojss.2012.22016 Published Online June 2012 (http://www.SciRP.org/journal/ojss)
111
Spatial Variation in Soil Chemistry on a Sub-Antarctic
Island
Everhard Christiaan Conradie, Valdon R. Smith
Department of Botany & Zoology, Stellenbosch University, Stellenbosch, South Africa.
Received February 10th, 2012; revised March 12th, 2012; accepted March 26th, 2012
ABSTRACT
On both west and east sides of sub-Antarctic Marion Island (47˚S, 38˚E), total Na and exchangeable Na, Mg and K
concentrations in the soil decline with increasing distance inland and altitude, related to a decrease in the intensity of
seaspray deposition. On the east side, the coastal plain is wide and slopes gently up to the mountainous interior and total
C, total N and soil moisture content all decrease significantly, whereas bulk density increases significantly, as one
moves away from the sea, reflecting a gradual change from organic, wet, low bulk density peats characteristic of low-
land coastal regions to mineral, dry, high bulk density volcanic soils characteristic of inland areas. On the west side, the
narrow coastal plain is bounded by an escarpment that rises up very steeply to the highland interior. There, sampling
was largely restricted to the coastal plain (soils are rare on the escarpment and interior) and did not cover the same tran-
sition from organic to mineral soils as on the east side. Hence, total C, total N and bulk density did not change signifi-
cantly with increasing distance inland on the west side. Most total Mg is in the mineral fraction of the soil, with a lesser
contribution by organic, exchangeable and soil solution forms of Mg. On the east side the gradual transition from highly
organic peats to very mineral soils results in an increase in total Mg going inland, but on the west, where there was not
this change in soil minerality, total Mg decreased with increasing distance inland, reflecting the decreasing intensity of
seaspray. Once the between-side differences in the influence of altitude and distance from the sea are accounted for,
there are significant differences in soil chemical composition between the two sides of the island. Overall, west side
soils are more influenced by both seaspray and the parent volcanic basalts than are east side soils.
Keywords: Sub-Antarctic; Soil Nutrients; Altitudinal Variation; Seaspray; Mineral-Organic Gradient; Soil Organic
Matter
1. Introduction
Marion Island (47˚S, 38˚E, 2100 km southeast of Cape
Town) has a hyperoceanic climate typical of the sub-
Antarctic region [1]. It is cool (annual mean air tempe-
rature is 6˚C, the mean temperatures of the coldest and
the warmest month differ by only 4˚C), wet (mean an-
nual rainfall is 2200 mm·y–1, mean relative humidity 80%)
and windy (mean wind velocity is 25 km·h–1, gales occur
on 110 days per year). The island is volcanic, 290 km2 in
area and consists of a central highland (highest peak
1230 m above sea level) that on the eastern and northern
sides slopes down gently to a coastal plain 3 to 5 km
wide. On the island’s western and southern sides the coa-
stal plain is narrow (400 to 900 m wide) and meets an
escarpment rising up very steeply to the mountainous in-
terior. The most prevalent (and strongest) winds are from
the west [2] so that haline plant communities resistant to
saltspray extend much further inland (hundreds of meters)
on the west side of the island than they do on its east side
(rarely more than a few score meters). The island’s
ecosystem is classified as sub- Antarctic tundra [3].
An early aim of the ecological research program on the
island was to draw up a nutrient budget for the whole ter-
restrial ecosystem [4], which involved quantifying nutri-
ent inputs to, and losses from, the island and assessing
the spatial patterns of nutrient concentrations in its soils
and plants. This reflected the influence of the Tundra
Biome Project of the International Biological Programme
(1967-1974), a major objective of which was to elucidate
patterns of distribution of nutrients within tundra ecosy-
stems [5]. Research at the island has subsequently pro-
vided much information on the magnitudes of nutrient
transfers from the ocean to the island through saltspray,
precipitation and dryfallout [6] and also by seabirds and
seals that feed in the ocean and deposit excreta, shells,
moulted feathers and fur on the island [7]. In contrast,
soil nutrient concentrations have been measured only at a
few localities, and detailed nutrient budgets have been
drawn up for only eight of the island’s 41 plant com-

Spatial Variation in Soil Chemistry on a Sub-Antarctic Island 112
munities [8,9], all within a 1.5 ha area about 35 m a.s.l.
and 600 m from the coast on the island’s eastern side.
Some soil nutrient information exists for other areas on
the east side but there is none for other parts of the island.
In fact, prior to this study there was no information on
the spatial variation in soil chemical composition at the
island.
One component of a nutrient budget is the soil nutrient
standing stock, which is the mass of a particular nutrient
per square meter of soil down to a specified depth. This
depends on the concentration of that nutrient in, and the
bulk density (mass per volume) of, the soil. Hence, to
construct a nutrient budget on a whole-island basis it is
important to know how soil nutrient concentration and
soil bulk density change with increasing distance inland
and elevation, and if the changes are different on the
various sides of the island. Here, we report the results of
an investigation aimed at providing such information for
the island’s eastern and western sides.
2. Material and Methods
2.1. Sampling Localities and Protocol
In April/May 2010, soils were sampled from 77 localities
on the east side, and 55 localities on the west side, of the
island (Figure 1). Each locality’s latitude and longitude
were recorded using a GPS and used with a digital ele-
vation model in ArcGIS 8 (ESRI, California) to estimate
its altitude and distance to the sea, and also to construct
Figure 1. A 5 cm diameter core, 25 cm deep where the
soil was deep enough to allow it, was taken from the soil
and the surface vegetation, litter and conspicuous roots
removed. If the core contained conspicuous horizons,
these were separated. The volume of the whole core or
horizon section was calculated from its diameter and
length.
2.2. Soil Moisture and Chemical Analyses
The fresh core or horizon section was weighed, air dried
and weighed again. A weighed subsample of the air-dried
soil was dried in an oven at 105˚C for 12 hours and re-
weighed. From the fresh, air-dried and oven-dried masses,
moisture content was calculated on an oven-dried basis.
From the oven-dried mass and volume of the soil core or
horizon section, bulk density was calculated.
All chemical analyses were performed on the air-dried
subsamples and the concentrations expressed on an oven-
dried mass basis using the air-dry:oven-dry mass ratios.
For total P, Ca, Mg, K and Na, a soil subsample was ex-
tracted with boiling HCl:HNO3 mixture (3:1) and the
concentrations of P and the cations in the extract deter-
mined by Inductively Coupled Plasma—Optical Emi-
ssion Spectrometry (Varian Vista-MPX; Varian, Inc.
California). Total N was measured with a Leco FP-528
Figure 1. Map of Marion island with 100 m contour lines
and showing localities where soils were sampled.
nitrogen analyser (LECO Corporation, Michigan). Total
carbon was determined by the Walkley-Black method
[10]. Exchangeable cations were determined by extract-
ing soil subsamples with 1 M ammonium acetate solution
and measuring the concentrations of Ca, Mg, Na and K
in the extract by ICP-OES.
2.3. Data Analyses
Linear regression analysis was used to test if the magni-
tudes of the soil chemistry variables were significantly (P
≤ 0.05) related to altitude or distance to the sea. For total
Na and exchangeable Na and Mg, the relationships were
approximately negative exponential ones, so the data
were log-transformed. Two regression analyses were
carried out, one on the east side data and the other on the
west side data. If one or both of the regressions were
significant, a homogeneity-of-slopes analysis was used to
test if the slopes differed significantly between east and
west sides. If they did not, the between-sides difference
in concentrations was tested by analysis of covariance; if
they did, a separate-slopes test was used (in both in-
stances altitude or distance to the sea was the covariate).
Where the values of a chemical variable were unrelated
to altitude or distance to sea, the east-west difference in
its mean values was tested by analysis of variance. The
statistical analyses were done using Statistica 9 software
[11].
3. Results and Discussion
Table 1 compares the relationships (slopes) of the soil
chemistry values with altitude and with increasing dis-
tance from the sea between the eastern and western sides
of the island, and also gives the mean values for each
side (or the means adjusted for the effect of altitude and
distance from the sea).
On both the eastern and the western sides of the island,
concentrations of exchangeable Na, Mg and K declined

Spatial Variation in Soil Chemistry on a Sub-Antarctic Island 113
Table 1. Relationships between soil chemistry variables and altitude or nearest distance to sea. Only slopes of significant re-
gressions are reported. Where slopes or means differed significantly (P ≤ 0.05) between the east and west side of the island,
one of the pair is marked with an asterisk*. Where a variable was significantly influenced by altitude or distance to sea, the
mean value is the mean (±standard error) adjusted for that influence and is given in italics. Otherwise, actual means
(±standard error) are reported. For total Na and exchangeable Na and Mg, the influence of altitude or distance to sea was
approximately a negative exponential one so the data were log-transformed for the analysis. In that case the logs of the ad-
justed means and standard errors are reported and the exponential value of the (log) mean given in brackets. The reported
slope values for total C, total N, CEC, exch. K and exch. Ca are the actual slope values × 104. N is the number of samples.
Altitude Distance to the Sea
N Slope Means ± S.E.
(Adj. Mean ± S.E.) Slope Means ± S.E. (Adj. Mean ± S.E.)
(East/West) East West East West East West East West
Range (m) 7 - 409 27 - 130 20 - 4958 200 - 1268
Moisture
content (%) 77/55 –1.69 –8.59 834 ± 73.7 640 ± 79.1 –0.101 829 ± 77.9 641 ± 80.1
Bulk density
(g/cm3) 77/55 12.18 0.17 ± 0.0250.25 ± 0.0260.771 0.18 ± 0.027 0.25 ± 0.028
Total C (g/kg) 76/55 –417.43 205 ± 11.4 155 ± 12.1*–29.89 207 ± 12.1 155 ± 12.3*
Total N (g/kg) 77/55 –28.54 13.6 ± 0.669.9 ± 0.71*–2.13 12.8 ± 0.63 9.0 ± 0.75*
Total P (mg/kg) 74/54 667 ± 35.2 654 ± 68.5 667 ± 35.2 654 ± 68.5
Total Na (mg/kg) 74/54 –1.05 6.5 ± 0.07
(644)
7.5 ± 0.09*
(1778) –0.14 –1.01*
6.46 ± 0.066
(636)
7.46 ± 0.0775*
(1745)
Total K (mg/kg) 74/54 262 ± 14.8 679 ± 87.4* –0.995 255 ± 51.8 682 ± 52.6*
Total Ca (mg/kg) 74/54 2036 ± 1445966 ± 634.7* –7.25 1942 ± 385.1 5999 ± 390.7*
Total Mg
(mg/kg) 74/54 14.07 2770 ± 596.36782 ± 630.2*0.87 –8.91* 2761 ± 584.9 6818 ± 593.4*
Exchangeable
Na (meq/100g) 77/55 –0.07 –1.97*
0.29 ± 0.086
(1.33)
1.15 ± 0.101*
(3.17) –0.54 –1.35*
0.21 ± 0.081
(1.23)
1.12 ± 0.096*
(3.08)
Exchangeable K
(meq/100g) 77/55 –21.13 –54.62 0.84 ± 0.0570.56 ± 0.062*–1.54 –4.49 0.86 ± 0.06 0.56 ± 0.06*
Exchangeable Ca
(meq/100g) 77/55 –549.93 3.5 ± 0.44 5.4 ± 0.469* –35.37 3.7 ± 0.46 5.4 ± 0.48
Exchangeable
Mg (meq/100g) 77/55 –0.58 –1.01
1.3 ± 0.09
(3.50)
2.1 ± 0.109*
(7.91) –0.433 –0.643
1.2 ± 0.09
(3.30)
2.1 ± 0.12*
(7.84)
significantly with increasing altitude or distance inland.
Saltspray and aerosols of seawater (together, these are
henceforth termed seaspray) is the sole source of Na, and
the main sources of Mg and K, at the island [12], so this
is unsurprising. It also explains the decrease in total Na
with increasing distance inland.
On the east side, total C, total N and soil moisture
content declined significantly, while bulk density in-
creased significantly, with increasing altitude and dis-
tance inland. The four variables varied in a similar pat-
tern on the west side but the effect of the both covariates
was significant only for moisture content. That total C,
total N and bulk density did not change significantly with
increasing distance inland on the west side can be ex-
plained as follows.
Soil moisture content and total N concentration are
positively, and bulk density negatively, strongly corre-
lated with total C concentration [13]. Organic soils (peats)
are wet, have low bulk density and high total C and N
concentrations. Mineral soils are dry, with a high bulk
density and low total C and N concentrations. Mineral
soils also tend to have high total Ca and Mg concentra-
tions, since much of the total complement of these two
elements is found in the parent volcanic rock and ash. On
the eastern side, total C, total N and soil moisture all de-
cline significantly as one moves away from the sea and
up in elevation, reflecting the change from organic, wet,
low bulk density peats characteristic of lowland coastal
regions to the mineral, dry, high bulk density volcanic
soils characteristic of inland areas. On the island’s west
side, a narrow (400 to 900 m wide) coastal plain meets
an escarpment rising up very steeply to the mountainous
interior. Soils are rare on the escarpment and absent in
the interior, the substrate being lava or volcanic ash.

Spatial Variation in Soil Chemistry on a Sub-Antarctic Island 114
Most samples on the western side were of the coastal
plain soils; sampling did not extend as far inland or as
high up as it did on the eastern side. There was thus not
the same strong gradation from organic to mineral soils
with increasing distance inland as there was on the east-
ern side. This accounts for the lack of a significant rela-
tionship of total C, total N and bulk density with altitude
or distance to the sea on the western side.
On the eastern side, total Mg concentration increased
significantly, whereas on the western side it decreased
significantly, with increasing distance from the sea. This
is also attributable to the fact that on the west the sam-
pling did not extend over as a wide range of organic ver-
sus mineral soil types as on the east coast. Most total Mg
is in the mineral fraction of the soil (the basaltic parent
lava and scoria). The organic, exchangeable and dis-
solved forms contribute little to total Mg, except perhaps
in peats close to the sea. On the east side, the gradual
transition from highly organic peats to very mineral soils
represented by the samples results in an increase in total
Mg with increasing distance inland. On the west side
there was not this range of soil minerality and the change
(decrease) in total Mg with increasing distance inland
reflects the decreasing intensity of seaspray. This is not
to say that west side soils are not more mineral, on the
whole, than east side soils—it is the change in minerality
that is important, and that is greater on the east side.
West side soils are actually intrinsically more mineral
than east side soils, as is shown by the significantly lower
total C concentrations of the west side soils when ad-
justed for the effect of altitude or distance from the sea.
Mineral soils have a higher bulk density and are drier
than organic soils, and the altitude- or distance to sea-
adjusted differences in bulk density and moisture content
between the west and east sides of the island, although
not significant at P ≤ 0.05, also point toward west side
soils being intrinsically more mineral and less organic
than the east side soils.
Because of their greater minerality, west side soils
have higher concentrations of total Ca, Mg and K than
east side soils. Because the wind (especially strong wind)
is predominantly westerly, west side soils also have
higher concentrations of total Na and exchangeable
forms of Na, Ca, and Mg than do east side soils. In con-
trast, soil exchangeable K concentrations are lower on
the west side than the on east side. The reason for this is
uncertain. K does possess cation exchange properties not
shown by the other exchangeable cations; an important
one is that it can convert from an easily exchangeable to
a fixed form in some types of clay [14]. The identities of
clay minerals in Marion Island soils are a matter of some
controversy [3]; indeed, most of the controversy con-
cerns whether there are any clays at all [15]. However,
recent Raman spectroscopic analysis reveals that crystal-
line minerals such as biotite and muscovite, both capable
of weathering into K-fixing clays, are common on the
island [16]. Possibly, K-fixation is more intense in the
mineral-rich west coast soils than in the organic east
coast ones. Another possible explanation for the lower
exchangeable K concentration in the west side than east
side soils is that the high amount of Na reaching the soil
surface through seaspray on the western side displaces
exchangeable K as it percolates through the soil. On the
east coast, replacement of K by Na on the cation ex-
change complex would be much less intense because of
the much less intense deposition of seaspray.
4. Conclusion
Altitude and nearest distance to the sea are strongly in-
terrelated, so it is not surprising that soil chemical com-
position is influenced by both. The influence is through a
decreasing influence of seaspray with increasing distance
inland and a change from organic peats characteristic of
the coastal plain to mineral soils more influenced by
parent basalts further inland and higher up. Furthermore,
the dependence of soil chemical composition on altitude
or distance from the sea is a function of aspect—here we
showed that it differs markedly between the west and
east sides of the island. There are intrinsic differences (i.e.
once this between-side difference in the influence of al-
titude and distance from the sea on soil chemical com-
position has been accounted for) in soil chemical compo-
sition between east and west sides. These findings are
crucial for modeling nutrient cycling budgets on a whole
island basis.
5. Acknowledgements
This study was funded by the South African National
Research Foundation (grant SNA2008050700003) and
the fieldwork was supported logistically by the Antarc-
tica and Islands Directorate of the South African Depart-
ment of Environmental Affairs and Tourism.
REFERENCES
[1] B. R. Schulze, “The Climate of Marion Island,” In: E. M.
Van Zinderen Bakker, J. M. Winterbottom and R. A.
Dyer, Eds., Marion and Prince Edward Islands, A.A.
Balkema, Cape Town, 1971, pp. 16-31.
[2] P. C. Le Roux, “Climate and Climate Change,” In:
Chown, S. L and Froneman, P. W. Eds., The Prince Ed-
ward Islands: Land-Sea Interactions in a Changing Eco-
system, African SunMedia, Stellenbosch, 2008, pp. 39-64.
[3] V. R. Smith and L. Mucina, “Vegetation of Subantarctic
Marion and Prince Edward Islands,” In: L. Mucina and M.
C. Rutherford, Eds., The Vegetation of South Africa, Le-
sotho and Swaziland, Strelitzia, Vol. 19, South African
National Biodiversity Institute, Pretoria, 2006, pp. 698-

Spatial Variation in Soil Chemistry on a Sub-Antarctic Island 115
723.
[4] E. M. Van Zinderen Bakker, “The Second South African
Biological Expedition to Marion Island 1971-72,” South
African Journal of Antarctic Research, Vol. 13, 1973, pp.
60-63.
[5] O. W. Heal, “Introduction,” In: L. C. Bliss, O. W. Heal
and J. J. Moore, Eds., Tundra Ecosystems: A Comparative
Analysis, Cambridge University Press, Cambridge, 1981,
pp. xxvii-xxxvii.
[6] V. R. Smith, “Chemical Composition of Precipitation at
Marion Island (Sub-Antarctic),” Atmospheric Environ-
ment, Vol. 21, No. 5, 1987, pp. 1159-1165.
doi:10.1016/0004-6981(87)90243-5
[7] V. R. Smith and P. W. Froneman, “Nutrient Dynamics in
the Vicinity of the Prince Edward Islands,” 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. 165-179.
[8] V. R. Smith, “Production and Nutrient Dynamics of Plant
Communities on a Sub-Antarctic Island. 5. Nutrient
Budgets and Turnover times for Mire-grasslands, Fjaeld-
mark and Fernbrakes,” Polar Biology, Vol. 8, 1988, pp.
255-269. doi:10.1007/BF00263174
[9] 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.
[10] L. E. Allison, “Organic Carbon,” In: C. A. Black, Ed.,
Methods of Soil Analysis, Part 2, Chemical and Microbi-
ological Properties, American Society of Agronomy,
Madison, 1965, pp. 1367-1378.
[11] StatSoft Inc., STATISTICA Version 9.0., 2009.
www.statsoft.com
[12] V. R. Smith, “Soil Chemistry of Marion Island (Sub-
Antarctic),” South African Journal of Science, Vol. 74,
1978, pp. 174-175.
[13] V. R. Smith and M. Steenkamp, “Classification of the
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
[14] E. W. Russell, “Soil Conditions and Plant Growth,” Ninth
Edition, Longman, London, 1960.
[15] K.-H. Gribnitz, L. E., Kent and R. R. Dixon, “Volcanic
Ash, Ash Soils and the Inferred Quaternary Climate of
Sub-Antarctic Marion Island,” South African Journal of
Science, Vol. 82, 1986, pp. 629-635.
[16] L. C. Prinsloo, P. Colomban, J. D. Brink and K. I. Meik-
lejohn, “A Raman Spectroscopic Study of the Igneous
Rocks on Marion Island: A Possible Terrestrial Analogue
for the Geology on Mars,” Journal of Raman Spectros-
copy, Vol. 42, No. 4, 2011, pp. 626-632.
doi:10.1002/jrs.2756