Content uploaded by Thomas J. Sauer
Author content
All content in this area was uploaded by Thomas J. Sauer
Content may be subject to copyright.
Commentary
Spatial variation of soil properties relating to vegetation changes
Thomas J. Sauer
1
, Cindy A. Cambardella & David W. Meek
USDA-ARS, National Soil Tilth Laboratory, 2150 Pammel Drive, Ames, IA, 50011-4420, USA.
1
Corresponding author*
Received 20 July 2005. Accepted in revised form 29 July 2005
Key words: forest, prairie, eastern red cedar, geostatistics
Abstract
Bekele and Hudnall provide an interesting perspective on the spatial variation of soil chemical properties in
a natural area undergoing transition from prairie to forest. Their focus is on the unique calcareous prairie
ecosystem of Louisiana where prairie remnants are being encroached upon by the forest, primarily eastern
red cedar (Juniperus virginiana L.). Bekele and Hudnall were especially interested in investigating any
differences in spatial variability among similar sites and in documenting the scale at which the variability
occurs. Geostatistical methods have been used to describe and model spatial patterns in soil data for more
than 20 years. The accessibility of user-friendly geostatistical software packages has increased the use
.
of
spatial analysis of soil’s data but carries the risk that these tools are used without due consideration of the
underlying theory, especially in the field of semivariogram modeling or recommended good practices. The
feedback between plant community composition and species distribution and soil properties in natural
systems has promise to provide enhanced insight into the short- and long-term relationships between plants
and soil properties. This is an intriguing area of research that couples plant ecology and soil science and
should provide valuable information on the interaction of soils with the processes of plant succession and
competition. Researchers in this area are urged to be cautious in verifying the assumptions behind popular
geostatistical methods and explicit in describing the important steps such as trend analysis, which can reveal
critical interpretive information.
Information on soil properties has often been
sought for agronomic applications and was ob-
tained from random sampling within defined
areas such as plots, fields, and farms. With
advancement in the ability to easily and accu-
rately locate sample sites came opportunities for
strategic sampling and to portray spatial represen-
tations of various soil properties (e.g.,
Cambardella et al., 1994; Sauer and Meek, 2003;
West et al., 1989). Such observations led to an
appreciation of the spatial variation of soil
properties as a complex product of anthropogenic
management practices superimposed on native
soil properties. As agronomic systems are typi-
cally annual monocultures, individual plant char-
acteristics have a limited effect on the spatial
distribution of soil properties. In natural ecosys-
tems, however, the composition, vigor, and matu-
rity of members of the plant community affect
competition among species and have potentially
profound implications for soil properties.
Bekele and Hudnall (pp. 7–21, this issue)
provide an interesting perspective on the spatial
* FAX No: +515-294-8125.
E-mail: sauer@nstl.gov
Plant and Soil (2006) 280:1–5 ÓSpringer 2006
DOI 10.1007/s11104-005-1545-8
variation of soil chemical properties in a natural
area undergoing transition from prairie to forest.
Their focus is on the unique calcareous prairie
ecosystem of Louisiana where prairie remnants
are being encroached upon by the forest, primar-
ily eastern red cedar (Juniperus virginiana L.).
Previous work by the same authors (Bekele and
Hudnall, 2003) reported on observations of soil
organic carbon (SOC) distribution in the same
prairie-forest transition zone. Although SOC dif-
ferences by vegetation type and landscape posi-
tion were not statistically significant, d
13
C data
suggest that the entire site may once have been
dominated by prairie vegetation. The current
paper focuses on unraveling the relationships be-
tween the spatial distribution of pH, EC, and sev-
eral macronutrients within and between the
forest, transition, and prairie vegetation zones.
Vegetation/soil interaction in prairie-forest
transition zones
There is a long and rich history of the study of
plant and landscape factors associated with prai-
rie-forest transition zones in the U.S. (Anderson,
1987; Curtis and McIntosh, 1951; Pettapiece,
1969; Transeau, 1935). Much of this attention
was directed at delineating the spatial extent of
native plant cover types at the onset of European
settlement and how the distribution of plant
communities affected ecosystem functioning and
pedological processes. The prairie-forest bound-
ary at the time of settlement has been perceived
as a somewhat static boundary; however, the
bulk of evidence now suggests that forest and
prairie plant communities have advanced and
receded across a diffuse transition zone. These
changes were largely in response to climatic vari-
ables, especially precipitation and temperature,
which in turn influenced grazing patterns and the
prevalence of fire. As a result, soils under prairie
or forest vegetation near transition zones at the
time of European settlement have been found to
exhibit characteristics associated with the alterna-
tive ecosystems (Bailey et al., 1964; Ruhe and
Scholtes, 1956; White et al., 1969).
Various factors have been investigated to
determine their influence on competition between
woody and herbaceous plants including aspect,
microtopography, disturbance, and soil patho-
gens (Aerts, 1999; Matlack, 1994; Mills and
Bever, 1998; Peltzer, 2001; Wilson and Tilman,
1993). Competition in natural plant communities
has also been differentiated based on nutrient
availability (Aerts, 1999). When nutrient and
water supplies are abundant, the primary compe-
tition is for light. In nutrient-poor environments,
however, there is poor consensus on the domi-
nant mechanism with two main opposing views.
The first is that successful species are more
competitive for nutrients and grow faster at a
limiting nutrient availability, and the second,
that successful species are those that reduce
nutrient losses. Ultimately, a combination of site
factors and soil properties produce either a posi-
tive or negative feedback on individual plant
species, which determines the composition of the
plant community.
Eastern red cedar is an aggressive invader
species in prairie ecosystems (Briggs and
Gibson, 1992; Gehring and Bragg, 1992; Norris
et al., 2001) owing to its fast growth, long
growing season, large cone crop, production of
cones at a young age, high seed-dispersal effi-
ciency, and tolerance of xeric conditions
(Axmann and Knapp, 1993; Holthuijzen and
Sharik, 1984, 1985; Ormsbee et al., 1976). Once
established in prairies, eastern red cedar trees
reduce surrounding herbaceous cover primarily
by shading (Gehring and Bragg, 1992). Even
isolated trees can have a profound effect on soil
properties due to fundamental differences be-
tween woody and herbaceous plants with regard
to the decomposition of organic inputs to the
soil. In forest systems, litter-fall on the soil sur-
face is the primary organic input, whereas in
prairie systems, the primary organic input is the
decomposition of fine roots (Anderson, 1987,
Pettapiece, 1969).
Decomposition of moist leaf litter on the soil
surface tends to be reasonably rapid and, in the
absence of mineral colloids, results in few clay-
humus complexes. For these reasons, forest soils
are often characterized by a thin, organic-rich O
horizon over an A horizon and deeper horizons
having relatively low concentrations of clay and
nutrients with significant loss of soluble organic
components (N, S, and P) and cations (Ca, Mg
and K) due to leaching (Anderson, 1987).
2
Geostatistical methods in soil science
Soil attributes frequently exhibit spatial structure
as the outcome of the combined interaction of
biological, chemical, and physical processes act-
ing at multiple scales (Parkin, 1993). Geo-
statistics is a powerful tool for characterization
and quantification of spatial variability. The
approach is based on the statistical theory of
autocorrelation and spatial variability character-
ization is accomplished using variography. Geo-
statistical methods have been used to describe
and model spatial patterns in soil data for more
than 20 years (Burgess and Webster, 1980a, b).
Geostatistical analysis has been used to estimate
spatial variability of soil physical properties
(Viera et al., 1981; Voltz and Goulard, 1994),
soil chemical properties (Sauer and Meek, 2003;
Yost et al., 1982), soil biochemical properties
(Bonmati et a1., 1991; Cambardella et al., 1994),
and soil microbiological processes (Aiken et al.,
1991; Rochette et al., 1991).
The accessibility of user-friendly geostatistical
software packages has increased the use of spatial
analysis of soil data, but carries the risk that
these tools are used without due consideration of
the underlying theory, especially in the field of
semivariogram modeling (Goovaerts, 1998) or
recommended good practices (see, e.g., Journel
and Huijbregts, 1978 or Meek and Sauer, 2004).
For instance, in variogram analysis, it is desirable
that the underlying data distribution meet Gauss-
ian (normal) assumptions, where departure from
normal is minimal, or normality is approximated
after appropriate mathematical transformation.
Another recommendation is that the intrinsic
hypothesis (stationarity of the mean) should be
met. If non-stationarity is evident, then data
de-trending is recommended prior to performing
variogram analysis. The variogram plots can be
constructed using all data pairs, including those
that represent the maximum extent of the data
domain. Variogram models, however, should be
estimated using only those data pairs from the
smallest lag distance up to a distance that repre-
sents half of the domain length, i.e. Journel’s
Practical Rule (Journel and Huijbregts, 1978).
This domain restriction is recommended to pro-
vide better parameter estimates for local (small-
scale) behavior, since there are more associated
data pairs at shorter lag distances. Estimation
procedures for variogram models should use a
generalized least squares, or similar, approach
that includes the pair count at each lag distance
in the model calculation (Cressie, 1993).
Bekele and Hudnall were especially interested
in investigating any
,
differences in spatial vari-
ability among similar sites, and in documenting
the scale at which the variability occurs. In order
to achieve these objectives, relative semivario-
grams were developed from soil data taken along
several transects that spanned remnant prairie,
prairie-forest transition, and forest vegetation.
Conventional statistical methods were employed
for Exploratory Data Analysis (EDA) prior to
the application of variography. EDA revealed
that all but one of the soil properties measured
was non-normally distributed, and that some or
all of the data exhibited a significant trend. The
trend was removed by trend surface analysis
according to methods described by White et al.
(1997). If information on which of the soil attri-
butes sampled from within which vegetation type
were de-trended, and the specific model or mod-
els used to remove the trend had been included,
then it would be possible for the reader to
discern just how well the model used to de-trend
the data fits the data pattern. This is particularly
important since the residual patterns that emerge
after de-trending depend to a great extent on the
model that is used to de-trend the data. For
instance, assume that a polynomial regression
model was used to de-trend the data when a seg-
mented-spline line model (Meek et al., 2001)
might provide a better fit. The semivariogram
developed from the residuals could show a cyclic
spatial trend, when in fact, the spatial trend is an
artifact introduced through the application of the
polynomial model. The possibility exists that
some of the most informative and interesting
findings of this type of study are missed because
of the choice of de-trending model. For instance,
the notch points from the segmented-line spline
model described above could be used to mathe-
matically delineate boundaries between the tran-
sition zone and prairie and forest areas. Such an
analysis would provide a spatially congruent pre-
dictive tool for identifying the boundary of the
transition zone. This is just one example of the
opportunities for enhanced data interpretation
presented by geostatistical analysis of soils data
sets.
3
Summary
Bekele and Hudnall demonstrate the utility of
spatial analysis of soil properties in a unique
ecosystem, a study that adds to the breadth of
literature regarding geostatistical analyses of soil
data. The feedback between plant community
composition and species distribution and soil
properties in natural systems has promise to
provide enhanced insight into the short- and
long-term relationships between plants and soil
properties. This is an intriguing area of research
that couples plant physiological ecology and soil
science, and should provide valuable informa-
tion on the interaction of soils with the pro-
cesses of plant succession and competition.
Researchers in this area are urged to be cau-
tious in verifying the assumptions behind popu-
lar geostatistical methods and to be explicit in
describing the important steps such as trend
analysis, which can reveal critical interpretive
information.
References
Aerts R 1999 Interspecific competition in natural plant
communities: Mechanisms, trade-offs and plant-soil feed-
backs. J. Exp. Bot. 50, 29–37.
Aiken R M, Jawson M D, Grahammer K and Polymenop-
oulos A D 1991 Positional, spatially correlated and random
components of carbon dioxide flux. J. Environ. Qual. 20,
301–308.
Anderson D W 1987 Pedogenesis in the grassland and adjacent
forests of the Great Plains. In Adv. Soil Sci. Ed. B A
Stewart. pp. 53–93. Vol. 7, Springer-Verlag, New York.
Axmann B D and Knapp A K 1993 Water relations of
Juniperus virginiana and Andropogon gerardi in an unburned
tallgrass prairie watershed. Southw. Nat. 38, 325–330.
Bailey L W, Odell R T and Boggess W R 1964 Properties of
selected soils developed near the forest-prairie border in east-
central Illinois. Soil Sci. Soc. Am. Proc. 28, 257–263.
Bekele A and Hudnall W H 2003 Stable carbon isotope study of
the prairie-forest transition soil in
.
Louisiana. Soil Sci. 168,
783–792.
Bonmati M, Cecanti B and Nanniperi P 1991 Spatial variability
of phosphatase, urease, protease, organic carbon and total
nitrogen in soil. Soil Biol. Biochem. 23, 391–396.
Briggs J M and Gibson D J 1992 Effect of fire on tree spatial
patterns in a tallgrass prairie landscape. Bull. Torrey Bot.
Club 119, 300–307.
Burgess T M and Webster R 1980a Optimal interpolation and
isarithmic mapping of soil properties. I. The variogram and
punctual kriging. J. Soil Sci. 31, 315–331.
Burgess T M and Webster R 1980b Optimal interpolation and
isarithmic mapping of soil properties. II. Block kriging.
J. Soil Sci. 31, 333–341.
Cambardella C A, Moorman T B, Novak J M, Parkin T B,
Karlen D L, Turco R F and Konopka A E 1994 Field-scale
variability of soil properties in central Iowa soils. Soil Sci.
Soc. Am. J. 58, 1501–1511.
Cressie N A C 1993 Statistics for Spatial Data. Academic Press,
New York.
Curtis J T and McIntosh R P 1951 An upland forest continuum
in the prairie-forest border region of Wisconsin. Ecology 32,
476–496.
Gehring J L and Bragg T B 1992 Changes in prairie vegetation
under eastern red cedar (Juniperus virginiana L.) in an
eastern Nebraska bluestem prairie. Am. Midl. Nat. 128, 209–
217.
Goovaerts P 1998 Geostatistical tools for characterizing the
spatial variability of microbiological and physico-chemical
soil properties. Biol. Fertil. Soils 27, 315–334.
Holthuijzen A M A and Sharik T L 1984 Seed longevity and
mechanisms of regeneration of eastern red cedar (Juniperus
virginiana L.). Bull. Torrey Bot. Club 111, 153–158.
Holthuijzen A M A and Sharik T L 1985 The avian seed
dispersal system of eastern red cedar (Juniperus virginiana
L.). Can. J. Bot. 63, 1508–1515.
Journel A G and Huijbregts C J 1978 Mining Geostatistics.
Academic Press, New York.
Matlack G R 1994 Vegetation dynamics of the forest
edge – trends in space and successional time. J. Ecol. 82,
113–123.
Meek D W and Sauer T J 2004 Suggestions for presenting
kriging results. In G Milliken Ed. Proc. 15th Appl. Stat.
Agric. Conf., Manhattan, KS, Apr. 29–May 1, 2003. Stat.
Dept., K S. St. Univ., Manhattan, KS, pp. 258–276.
Meek D W, Dinnes D, Jaynes D, Cambardella C, Colvin T,
Hatfield J and Karlen D 2001 An autoregression model
for a paired watershed comparison. In G. Milliken Ed.
Proc.12th Appl. Stat. Agric. Conf., Manhattan, K S, May
1–2, 2000. pp. 223–231. State. Dept., KS. St. Univ.,
Manhattan, KS.
Mills K E and Bever J D 1998 Maintenance of diversity within
plant communities: soil pathogens as agents of negative
feedback. Ecology 79, 1595–1601.
Norris M D, Blair J M and Johnson L C 2001 Land cover
change in eastern Kansas: litter dynamics of closed-canopy
eastern red cedar forests in tallgrass prairie. Can. J. Bot. 79,
214–222.
Ormsbee P, Bazzaa F A and Boggess W R 1976 Physiological
ecology of Juniperus virginiana in oldfields. Oecologia 23,
75–82.
Parkin T B 1993 Spatial variability of microbial processes in
soil – a review. J. Environ. Qual. 22, 409–417.
Peltzer D A 2001 Plant responses to competition and soil origin
across a prairie-forest boundary. J. Ecol. 89, 176–185.
Pettapiece W W 1969 The forest-grassland transition. In
Pedology and Quaternary Research. Ed. S Pawluk. pp.
103–113. Univ. of Alberta, Edmonton, AB.
Rochette P, Desjardins R L and Pattey E 1991 Spatial and
temporal variability of soil respiration in agricultural fields.
Can. J. Soil Sci. 71, 189–196.
Ruhe R V and Scholtes W H 1956 Ages and development of soil
landscapes in relation to climatic and vegetational changes in
Iowa. Soil Sci. Soc. Am. Proc. 21, 264–273.
Sauer T J and Meek D W 2003 Spatial variation of plant-
available phosphorus in pastures with contrasting manage-
ment. Soil Sci. Soc. Am. J. 67, 826–836.
Transeau E N 1935 The prairie peninsula. Ecology 16, 423–437.
4
Viera S R, Nielson D R and Biggar J W 1981 Spatial variability
of field-measured infiltration rate. Soil Sci. Soc. Am. J. 45,
1040–1048.
Voltz M and Goulard M 1994 Spatial interpolation of soil
moisture retention curves. Geoderma 62, 109–123.
West C P, Mallarino A P, Wedin W F and Marx D B 1989
Spatial variability of soil chemical properties in grazed
pastures. Soil Sci. Soc. Am. J. 53, 784–789.
White E M, Johnson J R and Nichols J T 1969 Prairie-forest
transition soils of the South Dakota Black Hills. Soil Sci.
Soc. Am. Proc. 33, 932–936.
White J G, Welch R M and Norvell W A 1997 Soil zinc map of
the USA using geostatistics and geographic information
systems. Soil Sci. Soc. Am. J. 61, 185–194.
Wilson S D and Tilman D 1993 Plant competition and resource
availability in response to disturbance and fertilization.
Ecology 74, 599–611.
Yost R S, Uehara G and Fox R L 1982 Geostatistical analysis
of soil chemical properties of large land areas. I. Semi-
variograms. Soil Sci. Soc. Am. J. 46, 1028–1032.
Section editor: H. Lambers
5
