Spatial Distribution of Soil Properties (1)
Spatial Distribution of Soil Properties and Soil Physical-Chemical Diversity in
the Greater Everglades Ecosystem
G.L. Bruland1*, S. Grunwald1, T.Z. Osborne1, K.R. Reddy1, and S. Newman2
of Florida, IFAS, Soil and Water Science Department, Gainesville, FL 32611. *Contact Info: Phone (352) 392-1951 ext. 210, Email: gbruland@ifas.ufl.edu
2South Florida Water Management District, Everglades Division, West Palm Beach, FL
Study Area and Maps
1
4 5
2
6
9
3
Hydrologic
Units
Locations
1 = WCA-1
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
7
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
2 = WCA-2A
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
3 = WCA-2B
5 = Holeyland
8
6 = WCA-3AN
11
7 = WCA-3AS
8 = WCA-3B
9 = BCN
12
10 = BCC
13
11 = BCS
12 = ENP
Methods
Bulk density values ranged from < 0.05 to >
1.25 g cm-3. The highest interpolated values
were observed in the marl and mixed marl
areas of BCNP, southeastern ENP, and Modellands. Western WCA-3AN and Rotenberger
also had relatively high BD possibly due to a
combination of factors including peat oxidation
and fire.
According to the RV map, the areas with the
highest pedodiversity were the WCAs,
especially WCA-1, and Shark River Slough. The
areas with the lowest pedodiversity included
BCNP, southeastern ENP, and Modellands.
This suggested that changes in hydrology and
nutrient loading have increased edaphic
heterogeneity in the GEE.
13 = Modellands
Laboratory Analysis
Soils were analyzed by the WBL for bulk density (BD), loss on ignition (LOI),
and total P, inorganic P, N, C, Ca, Mg, Al, and Fe by standard methods.
Geostatistics: Interpolation with Kriging and Splines
Where sample size > 90, ordinary kriging was used to interpolate soil properties
across the HUs. Semivariance values were fitted with spherical and exponential
semivariogram models. When sample size < 90, or when data exhibited no
spatial autocorrelation, a regularized spline function was used to interploate.
Pedodiversity Indices: Relativized Variance and Modified Shannon Index
Pedodiversity was first estimated with the relativized variance (RV). The RV
was calculated by dividing each soil property value at each site by the maximum
measured value for that property. This scaled values from 0-1, and the RV was
then estimated by calculating the variance of all the scaled values at each site.
Pedodiversity was also estimated using a modified version of the Shannon
diversity index (Shannon and Weaver, 1949). The raw soil data were converted to
10 evenly-spaced percentiles (data shifted from continuous to categorical).
140
120
100
80
60
40
20
0
Modified Shannon Index
Frequency distributions for the RV and SI values
calculated for each sampling point show the RV to
be bimodal and the SI to be skewed and truncated. Further examination indicates that the SI
overestimates diversity when concentrations are
high, while the RV is unbiased, more effectively
captures the expected data structure, and is a
better index of pedodiversity for this system.
Discussion
•Mapping the spatial distributions BD, LOI,
TCa, the RV, and the modified SI provide a
new perspective on pedodiversity in the
GEE.
Sampling Design
Cores were collected by the Wetland Biogeochemistry Laboratory (WBL) and
the South Florida Water Management District from >1,300 sites in the GEE
between 4/03-8/04. A stratified random sampling design was used to predict soil
properties at unsampled locations based on geostatistics.
Unit Boundary Definitions
A total of 13 hydrological units (HUs) were defined. For geostatistical purposes,
some of these HUs were divided into subsections based on physical boundaries
that affect hydrology such as highways and canals.
100
80
60
40
20
0
Relativized Variance
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
4 = Rotenberger
10
# of Cases
Frequency
Distributions
0.
00
0.
02
0.
04
0.
05
0.
07
0.
08
0.
10
0.
12
0.
13
0.
15
0.
16
0.
18
M
or
e
The spatial distributions of soil properties across wetland landscapes
represent the combined effects of various biotic and abiotic factors. Consequently, existing landscape patterns contain information about the processes
that generated these patterns. In the Florida Everglades, the spatial distribution
of soil nutrients can be used to assess long-term impacts to this system. To
this end, over 1,300 soil samples were collected by helicopter from the upper
0-10 cm of the Greater Everglades Ecosystem (GEE) in 2003-04. The objectives of this study were to: (i) characterize the spatial distributions of soil properties such as bulk density (BD), soil organic matter (by loss on ignition), and
total calcium (TCa) across the GEE, and (ii) to map the distribution of multivariate metrics of soil physical-chemical diversity across the GEE. Interpolated
maps of the individual soil properties revealed that BD was highest in the Big
Cypress National Preserve (BCNP), southeastern Everglades National Park
(ENP), Modellands, and northwestern Water Conservation Area (WCA) 3A.
LOI was highest in WCA-1 and southern WCA-3A. Total calcium displayed
high variability across the GEE with the highest values in Modellands, ENP,
and BCNP. Soil physical-chemical diversity, or pedodiversity, as expressed
with two metrics, a relativized variance (RV) and a modified Shannon Index
(SI), displayed an integrative spatial pattern that was a composite of the distributions of the individual soil properties. Pedodiversity appeared to be influenced by a combination of factors including underlying geology, hydrologic flow
patterns, vegetative communities, nutrient loading, and drainage. The indices
showed similar spatial patterns in some areas and divergent patterns in others.
Modified Shannon Index Continued
The percentiles were used to calculate Shannon’s
Index (SI) as follows: sp10
spi spi
SI = −∑
ln
S
spi S
where spi = percentiles for individual soil properties
at each site with i = 1,2,…10, and S = sum of all
percentile values for each site. Sites with higher
RVs or SIs have higher pedodiversity, and vice
versa.
# of Cases
Abstract
1.
73
1.
78
1.
83
1.
87
1.
92
1.
97
2.
01
2.
06
2.
11
2.
16
2.
20
2.
25
M
or
e
1University
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
LOI values ranged from < 10 to > 90 % and
generally exhibited an inverse trend to BD. Note
the high and homogeneous LOI values in WCA-1,
the area with the deepest peat. WCA-2A, WCA3AS, and the Shark River Slough area of the ENP
also exhibited high LOI values.
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Calcium exhibited high variability across the
GEE with values ranging from < 15,000 to >
300,000 mg/kg. The highest values were found
in Modelland, ENP, and BCNP, while the WCAs
generally had lower and more homogeneous
calcium concentrations. These patterns reflected differences in peat depth and substrate geology that occur across the GEE.
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Similar to RV, the map of modified SI revealed
that BCNP, southeastern ENP, and Modellands
had lower pedodiversity, while Rotenberger,
Holeyland, WCA-2, WCA-3, and Shark River
Slough had higher pedodiversity. Unlike the RV
map, the SI map indicated that WCA-1 had low
pedodiversity.
•Pedodiversity is influenced by both the
natural variability of geology, hydrologic
flow patterns, and vegetation as well as
anthropogenically-influenced variability
related to nutrient loading and altered
hydrology.
•Maps of RV and SI were similar in some
areas and diverged elsewhere. The RV
appears to be a more unbiased metric of
pedodiversity than SI. The categorization
of the continuous soil data required for the
SI also may result in a loss of information.
•More research is need to develop these
and other multivariate metrics of
pedodiversity.
Acknowledgements
Funding was provided by the South Florida Water
Management District. We would like to thank Y.
Wang of the WBL for her work with the laboratory
analysis, and R.G. Rivero and R. Corstanje for
their support of this research.
the Greater Everglades Ecosystem
G.L. Bruland1*, S. Grunwald1, T.Z. Osborne1, K.R. Reddy1, and S. Newman2
of Florida, IFAS, Soil and Water Science Department, Gainesville, FL 32611. *Contact Info: Phone (352) 392-1951 ext. 210, Email: gbruland@ifas.ufl.edu
2South Florida Water Management District, Everglades Division, West Palm Beach, FL
Study Area and Maps
1
4 5
2
6
9
3
Hydrologic
Units
Locations
1 = WCA-1
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
7
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
2 = WCA-2A
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
3 = WCA-2B
5 = Holeyland
8
6 = WCA-3AN
11
7 = WCA-3AS
8 = WCA-3B
9 = BCN
12
10 = BCC
13
11 = BCS
12 = ENP
Methods
Bulk density values ranged from < 0.05 to >
1.25 g cm-3. The highest interpolated values
were observed in the marl and mixed marl
areas of BCNP, southeastern ENP, and Modellands. Western WCA-3AN and Rotenberger
also had relatively high BD possibly due to a
combination of factors including peat oxidation
and fire.
According to the RV map, the areas with the
highest pedodiversity were the WCAs,
especially WCA-1, and Shark River Slough. The
areas with the lowest pedodiversity included
BCNP, southeastern ENP, and Modellands.
This suggested that changes in hydrology and
nutrient loading have increased edaphic
heterogeneity in the GEE.
13 = Modellands
Laboratory Analysis
Soils were analyzed by the WBL for bulk density (BD), loss on ignition (LOI),
and total P, inorganic P, N, C, Ca, Mg, Al, and Fe by standard methods.
Geostatistics: Interpolation with Kriging and Splines
Where sample size > 90, ordinary kriging was used to interpolate soil properties
across the HUs. Semivariance values were fitted with spherical and exponential
semivariogram models. When sample size < 90, or when data exhibited no
spatial autocorrelation, a regularized spline function was used to interploate.
Pedodiversity Indices: Relativized Variance and Modified Shannon Index
Pedodiversity was first estimated with the relativized variance (RV). The RV
was calculated by dividing each soil property value at each site by the maximum
measured value for that property. This scaled values from 0-1, and the RV was
then estimated by calculating the variance of all the scaled values at each site.
Pedodiversity was also estimated using a modified version of the Shannon
diversity index (Shannon and Weaver, 1949). The raw soil data were converted to
10 evenly-spaced percentiles (data shifted from continuous to categorical).
140
120
100
80
60
40
20
0
Modified Shannon Index
Frequency distributions for the RV and SI values
calculated for each sampling point show the RV to
be bimodal and the SI to be skewed and truncated. Further examination indicates that the SI
overestimates diversity when concentrations are
high, while the RV is unbiased, more effectively
captures the expected data structure, and is a
better index of pedodiversity for this system.
Discussion
•Mapping the spatial distributions BD, LOI,
TCa, the RV, and the modified SI provide a
new perspective on pedodiversity in the
GEE.
Sampling Design
Cores were collected by the Wetland Biogeochemistry Laboratory (WBL) and
the South Florida Water Management District from >1,300 sites in the GEE
between 4/03-8/04. A stratified random sampling design was used to predict soil
properties at unsampled locations based on geostatistics.
Unit Boundary Definitions
A total of 13 hydrological units (HUs) were defined. For geostatistical purposes,
some of these HUs were divided into subsections based on physical boundaries
that affect hydrology such as highways and canals.
100
80
60
40
20
0
Relativized Variance
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
4 = Rotenberger
10
# of Cases
Frequency
Distributions
0.
00
0.
02
0.
04
0.
05
0.
07
0.
08
0.
10
0.
12
0.
13
0.
15
0.
16
0.
18
M
or
e
The spatial distributions of soil properties across wetland landscapes
represent the combined effects of various biotic and abiotic factors. Consequently, existing landscape patterns contain information about the processes
that generated these patterns. In the Florida Everglades, the spatial distribution
of soil nutrients can be used to assess long-term impacts to this system. To
this end, over 1,300 soil samples were collected by helicopter from the upper
0-10 cm of the Greater Everglades Ecosystem (GEE) in 2003-04. The objectives of this study were to: (i) characterize the spatial distributions of soil properties such as bulk density (BD), soil organic matter (by loss on ignition), and
total calcium (TCa) across the GEE, and (ii) to map the distribution of multivariate metrics of soil physical-chemical diversity across the GEE. Interpolated
maps of the individual soil properties revealed that BD was highest in the Big
Cypress National Preserve (BCNP), southeastern Everglades National Park
(ENP), Modellands, and northwestern Water Conservation Area (WCA) 3A.
LOI was highest in WCA-1 and southern WCA-3A. Total calcium displayed
high variability across the GEE with the highest values in Modellands, ENP,
and BCNP. Soil physical-chemical diversity, or pedodiversity, as expressed
with two metrics, a relativized variance (RV) and a modified Shannon Index
(SI), displayed an integrative spatial pattern that was a composite of the distributions of the individual soil properties. Pedodiversity appeared to be influenced by a combination of factors including underlying geology, hydrologic flow
patterns, vegetative communities, nutrient loading, and drainage. The indices
showed similar spatial patterns in some areas and divergent patterns in others.
Modified Shannon Index Continued
The percentiles were used to calculate Shannon’s
Index (SI) as follows: sp10
spi spi
SI = −∑
ln
S
spi S
where spi = percentiles for individual soil properties
at each site with i = 1,2,…10, and S = sum of all
percentile values for each site. Sites with higher
RVs or SIs have higher pedodiversity, and vice
versa.
# of Cases
Abstract
1.
73
1.
78
1.
83
1.
87
1.
92
1.
97
2.
01
2.
06
2.
11
2.
16
2.
20
2.
25
M
or
e
1University
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
LOI values ranged from < 10 to > 90 % and
generally exhibited an inverse trend to BD. Note
the high and homogeneous LOI values in WCA-1,
the area with the deepest peat. WCA-2A, WCA3AS, and the Shark River Slough area of the ENP
also exhibited high LOI values.
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Calcium exhibited high variability across the
GEE with values ranging from < 15,000 to >
300,000 mg/kg. The highest values were found
in Modelland, ENP, and BCNP, while the WCAs
generally had lower and more homogeneous
calcium concentrations. These patterns reflected differences in peat depth and substrate geology that occur across the GEE.
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Everglades Soil Mapping Project
(PI: K.R. Reddy; co-PI: S. Grunwald)
Similar to RV, the map of modified SI revealed
that BCNP, southeastern ENP, and Modellands
had lower pedodiversity, while Rotenberger,
Holeyland, WCA-2, WCA-3, and Shark River
Slough had higher pedodiversity. Unlike the RV
map, the SI map indicated that WCA-1 had low
pedodiversity.
•Pedodiversity is influenced by both the
natural variability of geology, hydrologic
flow patterns, and vegetation as well as
anthropogenically-influenced variability
related to nutrient loading and altered
hydrology.
•Maps of RV and SI were similar in some
areas and diverged elsewhere. The RV
appears to be a more unbiased metric of
pedodiversity than SI. The categorization
of the continuous soil data required for the
SI also may result in a loss of information.
•More research is need to develop these
and other multivariate metrics of
pedodiversity.
Acknowledgements
Funding was provided by the South Florida Water
Management District. We would like to thank Y.
Wang of the WBL for her work with the laboratory
analysis, and R.G. Rivero and R. Corstanje for
their support of this research.