assessment of the spatial distribution of soil properties in a ......stormwater treatment areas...

Assessment of the Spatial Distribution of Soil Properties in a Northern Everglades Marsh

R. Corstanje, S. Grunwald,* K. R. Reddy, T. Z. Osborne, and S. Newman

ABSTRACTFlorida Everglades restoration plans are aimed at maintaining and

restoring characteristic landscape features such as soil, vegetation,and hydrologic patterns. This study presents the results from an ex-haustive spatial sampling of key soil properties in Water ConservationArea 1 (WCA 1), which is part of the northern Everglades. Three soilstrata were sampled: floc, upper 0- to 10-cm soil layer, and 10- to 20-cmsoil layer. A variety of properties were measured including bulk den-sity (BD), loss on ignition (LOI), total phosphorus (TP), total inor-ganic phosphorus (TIP), total nitrogen (TN), total carbon (TC), totaliron (TFe), total magnesium (TMg), total aluminum (TAl), and totalcalcium (TCa). Interpolated maps and model prediction uncertaintiesof properties were generated using geostatistical methods. We foundthat the uncertainty associated with spatial predictions of floc, par-ticularly floc BD, was highest, whereas spatial predictions of soil chemi-cal properties such as soil Ca were more accurate. The resultant spatialpatterns for these soil properties identified three predominant fea-tures in WCA 1: (i) a north to south gradient in soil properties as-sociated with the predominant hydrological gradient, (ii) areas ofconsiderable soil nutrient enrichment along the western canal of WCA1, and (iii) areas of considerable Fe enrichment along the easterncanal. By using geostatistical techniques we were able to describe thespatial dynamics of soil variables and express these predictions withan acceptable level of uncertainty.

ECOLOGICAL RESTORATION of degraded lands is anemerging challenge arising from the growing rec-

ognition of the value of services provided by healthyecosystems. Restoration efforts range in intensity andcomplexity aiming to recover previous functions (e.g., tore-establish exogenous flows of water and sediment)and/or landscape patterns (e.g., nutrient status, vegeta-tion structure). Restoration in theEverglades, as enactedunder the Everglades Forever Act (State of Florida,1994), is concerned with maintaining and restoring ahabitat mosaic that is comprised of, among others, saw-grass (Cladium jamaicense Crantz) prairies and patches,openwater sloughs, tree islands, andmarl-formingmarshes(Noe et al., 2001). Soil nutrient enrichment has been as-sociated with significant alterations of Everglades wet-land ecosystem structure and function (DeBusk et al.,1994, 2001; Davis et al., 2003; King et al., 2004). Nutrientgradients (Davis, 1991; DeBusk et al., 1994) produce apatterned response within the Everglades. For example,

cattail (Typha domingensis Pers.) dominates areas close tonutrient inflows, which were also found to contain highlevels of water column and soil phosphorus (P) content.This study focuses on Water Conservation Area 1 (WCA1), which is part of the Loxahatchee National WildlifeRefuge (LNWR) of the Greater Everglades. The areahas been designated as “Outstanding Florida Waters” bythe State of Florida and contains much of the distinctivehabitat favored under the Greater Everglades restorationplans. The conservation area is also a unique hydrologicunit of the Everglades that is a slightly acidic, rain-fedsystem that sits on deep peat (South Florida Water Man-agement District, 1992). Since it was last sampled in 1991there is little current knowledge on the movement anddistribution of soil nutrients and cations in this particularwater conservation area (Newman et al., 1997).

Soils in wetlands often exhibit characteristic, complexspatial patterns that indicate heterogeneity in soil re-sources and affect patterns of soil process rates (Ettemaet al., 1998). These patterns are often a combination ofcurrent and historical autochtonous and allochtonousfunctions that influencewetland systems (Stolt et al., 2001).Commonly, the spatial distribution of soil properties isnot uniform. This uneven distribution of soil character-istics, such as nutrient availability, organic content, andmineral content, implicitly reflects the processes thatoccur within the larger ecosystem. Spatial variabilityof soil characteristics can strongly affect the outcomesof logical, empirical, and physical models of soil andlandscape processes (Lin et al., 2005) including thoseavailable to managers and planners in the Evergladesrestoration efforts. As a result, an adequate understand-ing of soil variability as a function of space becomes es-sential. Geostatistics views soil properties as continuousvariables and models these as the most likely outcomesof random processes (Webster, 2000). This allows therandom variation in soil properties to be formulatedmathematically, whichminimizes the prediction error forthe observed variables and provides confidence in pre-dictions for the unsampled locations. In other words, bydesigning, executing, and analyzing a spatial study usinggeostatistics we can quantitatively assess the propertiesand associated uncertainty of the spatial distribution ofsoil characteristics. This paper will analyze underlyingcauses of soil variability, increase our understanding of asoft water system in the Everglades, and support futureresearch efforts aimed at protecting this portion of theFlorida landscape. The specific objectives of this studywere to identify spatial patterns of physical and chemicalsoil properties in the topsoil and floc (decaying plant

R. Corstanje and S. Grunwald, GIS Research Laboratory, and K.R.Reddy and T.Z. Osborne, Wetland Biogeochemistry Laboratory, Soiland Water Science Department, University of Florida-Institute ofFood and Agricultural Sciences, 2169 McCarty Hall, PO Box 110290,Gainesville, FL 32611-0510. S. Newman, Everglades Department,South Florida Water Management District, P.O. Box 24680, West PalmBeach, FL 33416-4680. Received 23 June 2005. *Corresponding author([email protected]).

Published in J. Environ. Qual. 35:938–949 (2006).Technical Reports: Wetlands and Aquatic Processesdoi:10.2134/jeq2005.0255ª ASA, CSSA, SSSA677 S. Segoe Rd., Madison, WI 53711 USA

Abbreviations: BD, bulk density; LNWR, Loxahatchee NationalWild-life Refuge; LOI, loss on ignition; TAl, total aluminum; TC, totalcarbon; TCa, total calcium; TFe, total iron; TIP, total inorganic phos-phorus; TMg, total magnesium; TN, total nitrogen; TP, total phos-phorus; WCA, Water Conservation Area.

Reproducedfrom

JournalofEnvironmentalQuality.PublishedbyASA,CSSA,andSSSA.Allcopyrights

reserved.

938

material), and to interpret the observed spatial distribu-tions in the context of predominant soil processes inWCA 1.

MATERIALS AND METHODS

Site Description

The WCA 1 is a teardrop-shaped 55900-ha area boundedby a 92-km levee and an interior borrow canal. Water Conser-vation Area 1 is delimited to the west by the Everglades Agri-cultural Area (EEA) and by heavily urbanized areas to theeast (Fig. 1). This conservation area forms part of the 60 250-haLNWR, which was established in 1951 in an agreement be-tween the South Florida Water Management District and theU.S. Fish and Wildlife Service, under the Migratory Bird Con-servation Act. This agreement was later amended to include

the 650-ha Strazzulla Marsh, which lies adjacent to WCA 1. In1978, the Refuge was recognized as “Outstanding FloridaWaters,” which affords it greater water quality protection. Cur-rently, WCA 1 contains characteristic Everglades habitat suchas open water (114 ha), wet prairies (22 856 ha), sloughs(110 ha), sawgrass (11348 ha), tree islands (8867 ha), cattail(Typha spp.) (2317 ha), and cypress (Taxodium spp.) swamp(162 ha) and some unclassified areas (United States Fish andWildlife Service, 2004). In the 1960s, cattail coverage was lim-ited to less than 1% of the vegetation community (UnitedStates Fish andWildlife Service, 2004). In response to elevatednutrient inputs from the perimeter canals, the cattail coveragehad increased to 4% by 1987 (South Florida Water Manage-mentDistrict, 1992) to over 10% in 1991 (Newman et al., 1997).

Restoration efforts have resulted in the construction ofstormwater treatment areas (STAs) to treat exogenous inputsof surface runoff from the adjacent areas (Fig. 1). Stormwater

STA 1W

STA 1E

L-40c anal

L-13 canalWest Palm Beach

Canal

L-8

cana

l

L-12 canal

L-7

cana

l

L-15canal

L-39canal

Hillsborocanal

L-6

cana

l

pillways

Culverts

S-6 pump

S-5A pump

Water Conservation Area 1 (WCA 1)

Major rivers

Coastline ofFlorida

WCA1

WCA2

WCA3

LakeOkeechobee

S

Fig. 1. Geographic location of Water Conservation Area 1 (WCA 1) within Florida. Insert shows a detailed schematic of the water control structuressurrounding WCA 1 and the location of the soil sampling sites within WCA 1. The term STA denotes storm water treatment area, L denotescanals, S denotes pumps.

Reproducedfrom


reserved.

939CORSTANJE ET AL.: SPATIAL DISTRIBUTION OF SOIL PROPERTIES IN THE EVERGLADES

Treatment Area 1W (STA-1W), a 2833-ha constructed wet-land adjacent to the northwestern boundary of WCA 1, treatssurface water diverted from S-5A canal before it enters the L-7perimeter canal. On the northeastern boundary, water fromthe ACME drainage basin is diverted into STA-1E (2699 ha)before entering the L-40 canal. At the time of the study, STA-1E did not yet discharge water into WCA 1. The S6 pumpwas halted in May 2001 and the water is now filtered throughSTA-2 and flows intoWCA 2a. A portion of STA-1W has beenoperating since 1994. As of 1991 WCA 1 received the largestportion of its water from rainfall (approximately 54%) inaddition to secondary inputs that enter WCA 1 from the perim-eter canal L-7 (Newman et al., 1997). The bedrock of WCA 1 islimestone and covered with a layer of peat that is about 2 to3.5 m thick. Soils are classified as euic, hyperthermic TypicMedisaprist of the Histosols soil order (McCollum et al., 1976).

Field Methods

Floc and soil samples were collected in March 2003 at120 sites and in August 2004 at 11 sites (Fig. 1). Not all sitesthat were sampled contained floc. A stratified random sam-pling design was used, based on a previous soil sampling of thearea (Newman et al., 1997) combined with historic ecologicaland hydrologic data layers (e.g., Normalized Difference Vege-tation Index) to determine the strata. The sampling design wasoptimized to account for short, medium, and long range vari-ability of attributes. The sampling sites were accessed usinga float helicopter. A global positioning system with real-timeWideArea Augmentation System (WAAS) was used to ensurea positional accuracy of ,3 m to locate the sites. We used aminimum of four satellites and average of six satellites toreceive accurate readings of x,y coordinates and minimize thepositional dilution of precision (PDOP) measure. Samplingwas restricted to herbaceous and open water plant communi-ties. Soils were sampled using a 10-cm (inner diameter) pushcore tube up to a depth of 20 cm beneath the soil surface. Flocwas measured by allowing settling of headspace water in thecore, pouring off excess water, then extruding floc to a point ofrecognizable soil. Floc depth was measured using a graded(cm) tube. Triplicates were obtained at about 10% of the sites.The cores were sectioned into three strata: floc, 0- to 10-cmlayer, and 10- to 20-cm layer. Floc was defined as material thatwould settle within approximately 60 s of the initial samplingdisturbance. For our sampling protocol we adopted the defi-nition of floc outlined by DeBusk et al. (2001), as consistingmostly of decaying macrophyte tissue, soil particles, algae,and microbes. Samples were placed and sealed in plasticbags, and stored on ice until they were received at the WetlandBiogeochemistry Laboratory where they were stored at 48Cuntil analysis.

Analytical Methods

Soil samples were analyzed at theWetland BiogeochemistryLaboratory, Soil andWater Science Department, University ofFlorida. After removal of any visible pieces of live and deadplant material, the samples were weighed and homogenizedthoroughly. Floc and soil bulk density (BD) was determinedon an oven dried (708C, 72 h), dry weight basis. The soils werethen fine ground to 0.01-mm sieve. Loss on ignition (LOI) wasdetermined as the sample mass lost after ashing at 5508C.

Total C (TC) and N (TN) concentrations were determinedwith a Carlo-Erba NA 1500 CNS Analyzer (Haak-BuchlerInstruments, Saddlebrook, NJ). Total phosphorus (TP) wasdetermined using the ashing method (Andersen, 1976). Thismethod consisted of ashing approximately 0.2 to 0.5 g oven-

dried soil in a muffle furnace for 3 to 4 h at 5508C. The ashresidue was then digested in 20 mL of 6 M HCl, in a digestionblock, until dry. The dry residue was consequently re-dissolvedin 2.25 mL of 6 M HCl and filtered through Whatman(Brentford, UK) 41 filter paper and brought to a final volumeof 50 mL. Soil total inorganic P (TIP) was determined byextracting 0.5 g air-dried, ground soil, with 25 mL of 1.0 MHCL for 1 h, and followed by vacuum filtration (0.45-mmmembrane filters; Reddy et al., 1998). Total phosphorus andTIP were determined using the absorbic acid automated col-orimetric procedure (Method 365.1; USEPA, 1993a) (Auto-analyzer II; Technicon, Terrytown, NY). Ashed and acidifiedsamples were also analyzed for total Ca (TCa), totalMg (TMg),total Fe (TFe), and total Al (TAl) by inductively coupled argonplasma spectrometry (ICAP) (Method 200.7; USEPA, 1993b).

Data Analysis and Geostatistics

Empirical semivariograms were constructed and modelvariograms fitted to soil attributes using the ISATIS soft-ware (Geovariances, 2005). The following approach was usedto analyze measured soil properties. First, we conducted anexploratory analysis including testing for normality and skew-ness, and the data were transformed when necessary. We thenconstructed isotropic experimental semivariograms using thefollowing equation:

g(h) 51

2n(h)On(h)i51

[z(xi 1 h) 2 z(xi)]2 [1]

where z(xi) and z(xi 1 h) represent pairs of observations sep-arated by a distance h (or lag size), n is the number of data pairs,and g(h) is the semivariance (Wackernagel, 2003). We testeddifferent lag sizes ranging from 950 to 2000 m. Interactive fit-ting was used to generate model semivariograms from the ex-perimental semivariogram. Cross-validation was performed andmeasures of model performance were computed. The modelsemivariograms were used in ordinary kriging, a weighted localinterpolation method, to construct the spatial patterns of eachsoil property. Empirical semivariance values were fitted withspherical and exponential semivariogram models.

The kriging variance or prediction variance (Webster andOliver, 2001) was calculated as:

var[Z(x0)] 5 E{[Z(x0) 2 Z(x0)]2}

5 2ONi51

lig(xi,x0) 2 ONi51ONj51

liljg(xi,xj)

[]

where Z(x0) is the estimate of the soil property Z at the krigingtarget point x0, g(xi, x0) is the semivariance of Z between ob-servations at xi and the target point x0, g(xi, xj) is the semi-variance of Z between observations at locations xi and xj, andli, lj are kriging weights.

The estimated prediction standard deviations (square rootof the variance term in Eq. [2]) were exported to ArcGIS(Environmental Systems Research Institute, 2005). Predictionswere made on a grid with 100- 3 100-m resolution. For all soilproperties, including carbon to nitrogen ratios, point values werefirst computed and then subjected to geostatistical analysis (i.e.,spatial interpolations were generated).

The effectiveness of subsequent interpolation efforts isreported using a set of three statistics: the mean predictionerror (MPE), the root mean square error (RMSE), and the G-statistics (Schloeder et al., 2001). The MPE and RMSE aremetrics that estimate the deviation of the kriged predictionsfrom the observed observations. The G-statistic is a test of theeffectiveness of the model fit compared to a mean model fit

[2]

Reproducedfrom


reserved.

940 J. ENVIRON. QUAL., VOL. 35, MAY–JUNE 2006

and is scaled from 0 to 100. Values closer to 100 indicate a per-fect model fit while values closer to 0 indicate some improve-ment over the mean model. Spearman’s rank correlationcoefficient was calculated between pairs of soil variables.

It is beyond the scope of this paper to present all soil mapsfor all measured properties and layers. The deeper soils (10–20 cm) exhibited very similar patterns to the 0- to 10-cm soillayer, and therefore only maps for floc and 0 to 10 cm areincluded in this paper.

RESULTSThe average of the physical and chemical properties

of the soils sampled in WCA 1 reflected the organic na-ture of the Histosols prevalent in this area (McCollumet al., 1976), documented by high moisture contents, lowBD, and high LOI and TC (Table 1). Generally, LOIand BD increased with depth, whereas moisture contentdid not change appreciably within the profile. Soil TPandTIP decreased by depth, while soil TC andTNvariedlittle by depth, as indicated by mean values. Floc waspresent in the majority of the sites (.95%). Average flocTP and TIP concentrations were generally larger thanthe corresponding soil averages; however, floc BD wasmuch smaller. Floc and soil layers exhibited similar meanphysical and chemical characteristics for all other mea-

sured properties. For the 0- to 10-cm depth, TP rangedfrom 113 to 1047mg kg21, whereas in the floc, TP rangedfrom 226 to 1462 mg kg21, indicating a wide range ofsoil nutrient conditions, from oligotrophic to nutrientenriched. Wide ranges in values were found for TCa,TMg, TFe, and TAl, whereas BD, LOI, TC, and TNweremore homogeneous (Table 1).

The spatial dependence of each soil property wasmodeled using analysis of semivariance (Table 2, Fig. 2).All properties exhibited spatial autocorrelative struc-tures, indicating that they responded to processes thatoccur throughout the conservation area. However, spa-tial dependencies did differ among soil properties, asillustrated by the nugget and sill values, as well as theranges (Table 2). The range establishes the outer limit atwhich points in space still interact spatially and the sillrepresents the total a priori variance, or s2 (Webster andOliver, 2001). Within the conservation area, soil and flocproperties displayed differences in their spatial struc-tures. Floc and soil TCa content exhibited the largestrange, indicating significant spatial structure extendingover 16.2 and 13.9 km, respectively, which is reflectedin the TCa semivariogram (Fig. 2d). With the exceptionof TCa, all soil ranges were larger than the ranges cal-culated for floc. Certain soil properties exhibited a nested

Table 1. Statistics of selected physicochemical characteristics of floc and soil samples collected in Water Conservation Area 1 (WCA 1).†

Property‡ Units Floc 0- to 10-cm soil depth 10- to 20-cm soil depth

n 125 131 131Moisture content % 92 6 1.51 (90–99) 92 6 0.25 (70–96) 92 6 0.2 (85–95)BD g cm23 0.022 6 0.0002 (0.001–0.14) 0.073 6 0.003 (0.03–0.40) 0.077 6 0.002 (0.05–0.15)LOI % 92 6 0.38 (73–99) 93 6 0.48 (41–99) 94 6 0.22 (0.05–0.15)TP mg kg21 632 6 24 (226–1462) 405 6 14 (113–1047) 241 6 6.0 (140–511)TIP mg kg21 158 6 8.2 (38–567) 80 6 4.3 (16–351) 34 6 1.2 (10–99)TN g kg21 33 6 0.49 (20–439) 33 6 0.42 (17–46) 34 6 0.42 (25–45)TC g kg21 447 6 2.5 (385–590) 473 6 2.8 (221–544) 492 6 2.0 (434–539)TCa mg kg21 14843 6 690 (3549–74258) 15971 6 538 (5973–33415) 17354 6 540 (6886–36609)TMg mg kg21 1969 6 93 (584–5990) 1928 6 91 (423–5507) 1933 6 91 (456–5062)TFe mg kg21 2046 6 99 (394–6332) 2017 6 103 (730–7886) 1978 6 105 (660–7927)TAl mg kg21 1469 6 82 (0–7030) 1774 6 80 (568–5461) 1395 6 35 (636–3192)

†Values are means 6 standard errors; minimum and maximum values are shown in parentheses.‡BD, bulk density; LOI, loss on ignition; TP, total phosphorus; TIP, total inorganic phosphorus; TN, total nitrogen; TC, total carbon; TCa, total calcium; TMg,total magnesium; TFe, total iron; TAl, total aluminum.

Table 2. Semivariogram models, interpolation parameters, and cross-validation statistics.

Property† Depth Lag distance Model Nugget Sill Range MPE§ RMSE§ G¶

m mDepth floc 1250 exponential 5.4 27.20 12513 20.021 4.3 25Bulk density floc‡ 1450 exponential 0.23 0.36 3722 0.003 0.02 28

0–10 cm‡ 1150 spherical 0.08 0.15 3105 0.003 0.02 60spherical 0.28 12729

TP floc‡ 1050 exponential 0.07 0.13 5693 33 227 280–10 cm‡ 1250 spherical 0.06 0.07 3290 17 134 25

spherical 0.11 9978TIP floc‡ 1050 exponential 0.16 0.26 7174 15 81 21

0–10 cm‡ 1750 exponential 0.07 0.16 7780 8 44 7TCa floc‡ 1500 spherical 0.01 0.15 16242 366 5034 74

0–10 cm‡ 1500 spherical 0.03 0.18 13904 389 3723 59C to N ratio floc 1500 exponential 3.4 5.61 5532 20.0013 2.6 11

0–10 cm 1050 spherical 2.0 2.80 5650 20.03 1.8 12TFe floc‡ 850 spherical 0.10 0.24 2796 131 1024 14

0–10 cm‡ 1050 spherical 0.09 0.10 3230 170 1093 11spherical 0.16 12090

†TP, total phosphorus; TIP, total inorganic phosphorus; TCa, total calcium; TFe, total iron.‡These properties were natural log transformed, semivariogram characteristics computed on the transformed observations.§Mean prediction error (MPE) and root mean square error (RMSE) were calculated on the original scales, data were back transformed where needed.¶Goodness of prediction (G), as defined by Schloeder et al. (2001).

Reproducedfrom


reserved.


spatial structure that was modeled with two sphericalsemivariogram models containing two ranges. Soil BD,TP, andTFe demonstrated nested spatialmodel structureswith ranges between 9.9 and 12.7 km for the long-scalemodel and 3.1 to 3.3 km for the short-scale model. Con-sidering that the conservation area is 22 kmwide fromeastto west and spans 36 km north to south, the spatial auto-correlation structure of most soil properties covered mostof the conservation area.

The nugget semivariance at distance h5 0 represents acomposite of the portion of fine-scale spatial variabilitythat has not been sampled, uncertainty introduced by thefield and experimental approaches, and random variabil-ity. The partial sill variance represents the portion of thetotal semivariance that comprises spatial autocorrelation.

Large nugget values, relative to the sill variance, werefound for BD, TIP, and TP of floc, and TP, TFe, and C toN ratio of soil, which indicate that other factors influ-

Distance (m)0 5000 10000 15000

D1M

Distance (m)0 2500 5000 7500 10000 12500

0.1

0.2

0.3

0.4

0

Bul

k D

ensi

ty l

n(g

cm-3

)

D1M

D1M

Dep

th (

cm)

0

10

20

30

Distance (m)0 5000 10000 15000

0.025

0.05

0.075

0.1

0

TP

ln(m

g kg

-1)

0.125D1M

0 5000 10000 15000Distance (m)

Ca

ln(m

g kg

-1)

0.05

0.10

0.15

0.20

0

c d

D1M

0 2500 5000 7500 10000 125000

1

2

3

4

5

6

C:N

rat

io

Distance (m)

D1

M

0 2500 5000 7500 10000 12500Distance (m)

0.1

0.2

0.3

0

Fe ln

(mg

kg-1

)

a b

e f

Fig. 2. Characteristic semivariograms of selected soil and floc properties: (a) floc depth, (b) floc bulk density (BD), (c) soil total phosphorus (TP),(d) soil total calcium (TCa), (e) floc C to N ratio, and (f) floc total iron (TFe). Some properties were natural log transformed. Thick line, indicatedwith label M, represents the fitted variogram model; thinner line represents the experimental variogram and is indicated with the label D1. Theunlabeled horizontal line represents the total sample variance. The distance (x axis) was measured in meters.

Reproducedfrom


reserved.


ence the variability of these measurements. In other soilproperties, spatial dependencies constituted a large con-tribution, particularly in TCa and in floc depth.The values of the statistics pertaining to the quality of

the fitted models indicated that the floc BD predictionswere by far the most uncertain. Other properties thatexhibited relatively high degrees of uncertainty wereTIP and TFe, for both strata. The uncertainty of predic-tions was lower for other properties. According to theG-statistics, the kriging model generally outperformedthe mean model for all properties except for floc BD.The kriging models performed best for soil BD (0- to10-cm soil) with a G-value of 60, floc and soil TCa withG-values of 74 and 59, respectively, and moderately wellfor the other properties.The interpolated floc depth, floc BD, and soil BD are

presented in Fig. 3. The inserts illustrate the uncertaintyassociated with the predictions of these properties. Pre-dictions of floc depth in areas (approximately 14% ofthe total area) in the southern and western part of theconservation area were highest, with floc depths rangingfrom 11 to 17 cm. A small number of localized floc con-centration areas, with maximum predicted floc depthsbetween 15 and 17 cm, were found along the south-western fringe of WCA 1, covering an estimated 1% ofthe total area. Floc and soil BD spatial patterns wereuniform over the conservation area. The northern por-tions of the marsh generally exhibited high floc BDpredictions, exceeding the median of 0.02 g cm23, andhigh soil BD predictions, exceeding the median of 0.07 gcm23. Similarly, spatial patterns of LOI (data not shown)showed the same north-to-south gradient. Predicted flocand soil LOI in the northern areas of WCA 1, whichranged from 81 to 85% and 85 to 88%, respectively, waslower than the predicted floc and soil LOI in the southernareas, which ranged from 92 to 99% and 94 to 98%, re-spectively. The inserted maps plot the standard devia-tions associated with the interpolated predictions toindicate the uncertainty associated with these predic-

tions. The results from the performance statistics for flocBD (Table 2) confirmed the relatively high uncertaintydepicted in the insert in Fig. 3b. These maps also highlightthat uncertainties close to sampling sites were small butincreased with distance from sampling sites. Along theedges of WCA 1 the uncertainties tended to be high dueto sparser sampling.

The spatial distribution of floc and soil forms of P(TP and TIP, Fig. 4a, 4b, 4c, and 4d) generally show agradient from west to east, with P enrichment on thewestern edge with predicted maxima of 1153 mg kg21 infloc TP and 633 mg kg21 soil TP. In comparison, theobserved maximum for the sampled soil TP was 1047 mgkg21 (Table 1). This discrepancy is the result of diver-gent observations (e.g., 1047 and 286 mg kg21) at closedistances (150–600 m) generating an overall predictedmaxima that is lower than the observed (sampled) max-ima. Localized P enriched areas that exceeded 500 mgkg21 were found adjacent to canal L-7, with a predictedaverage of floc TP (829mg kg21) and TIP (182 mg kg21),which were about twice as high as in the underlyingpredicted soil averages (554 mg kg21 for TP and 88 mgkg21 for TIP, respectively). The area in which the upper10 cm of soil TP exceeded the threshold of 500 mg kg21,identified by the State of Florida as “impacted areas” forthe Everglades (State of Florida, 2005), covered approx-imately 5% of the total area and was confined to thewestern edge, along canal L-7. FlocTIP patterns exhibiteda southward extension that was less pronounced than thefloc TP patterns and mirrored the patterns found for flocdepth (Fig. 2a). Soil TP and TIP enrichment was foundprimarily along the western edge of the conservation area(Fig. 4d and 4e), with a relatively randommottled patternin the interior. The inserted uncertainty maps indicatethat for both TP and TIP, the uncertainty associated withthe floc predictions were higher when compared to thecorresponding predictions for the soil layer.

Floc TCa (Fig. 4c) exhibited a more localized patternof enrichment, with little spatial variability in the

Floc Depth (cm)

2 - 4

4 - 6

6 - 9

9 - 11

11 - 13

13 - 15

15 - 17

Floc Depth Stdev (cm)

2.7 - 3.0

3.0 - 3.4

3.4 - 3.7

3.7 - 4.0

4.0 - 4.3

4.3 - 4.6

4.6 - 5.0

l Dept cm)c

Floc BD (g cm-3)

0.012 - 0.019

0.019 - 0.028

0.028 - 0.035

0.035 - 0.043

0.043 - 0.050

0.050 - 0.058

0.058 - 0.066

Floc BD Stdev(g cm-3)

0.0083 - 0.0086

0.0086 - 0.0090

0.0090 - 0.0093

0.0093 - 0.0096

0.0096 - 0.0100

0.0100 - 0.0103

0.0103 - 0.0106

0.02 -

Floc BD Stdev(g cm-3)

0-10 cm BD (g cm-3)

0.051 - 0.072

0.072 - 0.094

0.094 - 0.115

0.115 - 0.136

0.136 - 0.158

0.158 - 0.179

0.179 - 0.200

0-10 cm BD Stdev(g cm-3)

0.0105 - 0.0113

0.0113 - 0.0121

0.0121 - 0.0129

0.0129 - 0.0138

0.0138 - 0.0146

0.0146 - 0.0154

0.0154 - 0.0163

- 1

a b c

Fig. 3. Spatial distribution of floc depth (a) and floc and soil bulk density (BD) (b, c) in Water Conservation Area 1 (WCA 1), sampled in March2003 and August 2004. The scale bar shows the extent of the estimated properties.

Reproducedfrom


reserved.


interior of the marsh. Soil TCa (Fig. 4f) was also spatiallyaffected by the enrichment from the western edges origi-nating from canal L-7. This is superimposed on a north-to-south gradient, in which the lowest predicted valueswere found in the southeastern portion of WCA 1. Theuncertainties associated with the interpolations of TCawere low for both floc and soil (inserts in Fig. 4c and 4f).The spatial variability of floc and soil C to N ratios

across the conservation area, depicted in Fig. 5, repre-sent a combination of the spatial dynamics prevalent inthese two soil properties. High ratio values imply a rela-tively high TC respective to TN content. The spatial dis-tribution of soil TC exhibited a gradient north to southwith the highest predicted values in the southern areasof WCA 1 (average TC of 452 mg kg21 in the northernhalf to 504 mg kg21 in the southern half). These spatialpatterns were much the same in the floc, where theaverage predicted floc TC content was 410 g kg21 in thenorthern half and 500 g kg21 in the southern half. FlocTN exhibited more complex patterns in which the lowest

predicted TN values (ranging from 26 to 28 g kg21) werefound along the western edges of the marsh. Slightlyhigher predicted values (ranging from 28 to 33 g kg21)occurred along the northern-most and southern-mosttips of the area and a band of higher predicted TN valuesacross the center (ranging from 33 to 39 g kg21). Soil TNpatterns were similar to those of the floc TN in that thelowest predicted values were found on the western edgeof the area (ranging from 27 to 28 g kg21), superimposedon a north-to-south gradient, with the highest predictedvalues in the northern areas of the marsh (averaging 37 gkg21 in the north and 30 g kg21 in the south). The com-bined effect of the spatial patterns of these two soil prop-erties (TC and TN) can be seen in Fig. 5. The highestpredicted floc and soil C toN ratios were found in thewestas a result of the low N values and in the south as a resultof the C gradient. The uncertainty associated with thefloc C:N layer was found to be higher than that associatedwith the soil layer (inserts in Fig. 5a and 5b). The spatialdistribution of the predicted floc and soil TFe (Fig. 6)

Fig. 4. Spatial distribution of floc total phosphorus (TP) (a), total inorganic phosphorus (TIP) (b), total calcium (TCa) (c), soil total phosphorus(TP) (e), total inorganic phosphorus (TIP) (f), and calcium (Ca) (d) in Water Conservation Area 1 (WCA 1), sampled in March 2003 and August2004. The scale bar shows the extent of the estimated properties.

Reproducedfrom


reserved.


presents a gradient east to west with a focal point alongthe eastern edge of the area (from canal L-40). The un-certainty associated with the floc TFe predictions washigher than that associated with soil TFe predictions, as

seen in the upper bound standard deviations (inserts,Fig. 6a and 6b).

Most soil properties were shown to be weakly cor-related (Table 3). The strongest correlation coefficients

Fig. 5. Spatial distribution of floc (a) and soil (b) carbon to nitrogen ratios in Water Conservation Area 1 (WCA 1), sampled in March 2003 andAugust 2004. The scale bar shows the extent of the predicted properties.

Fig. 6. Spatial distribution of floc (a) and soil (b) iron (Fe) in Water Conservation Area 1 (WCA 1), sampled in March 2003 and again in August2004. The scale bar shows the extent of the predicted properties.

Reproducedfrom


reserved.


(Spearman’s r) were found between observations per-taining to measures such as BD and LOI (20.96) or TPand TIP (0.93) for properties that are, in any case, knownto be inherently correlated. Other properties that werestrongly, positively correlated were TCa and TMg (r 50.89). Pairs of properties that show a tight positive cor-relation can be expected to exhibit very similar spatialpatterns, which was the case when overlaying the TCaand TMg maps. Inversely, pairs of properties that havestrong, significant negative correlations are expected toexhibit spatial patterns that are mirror images as withTCa and LOI or TC and TP.

DISCUSSIONThe spatial modeling undertaken for this study re-

sulted in a detailed description of the spatial variabilityof the soil properties that we sampled and revealed thatthe predictions for the floc layer contained considerablyhigher uncertainty than those of the soil layer(s). Gen-erally, the floc component can be viewed as being farmore dynamic and responsive than the bulk soil (Reddyet al., 1999), which, as a result, also makes it more vari-able. However, floc variability is also higher due to thefact that it is more difficult to sample than soil, thus com-pounding the uncertainty. Within the soil propertiessampled, spatial predictions of the soil BD exhibited themost uncertainty, whereas the model derived for soil TCaseemed the most reliable.Spatial patterns observed in the Everglades landscape

mosaic and associated environmental driving forces havebeen the focus of a significant amount of research (Davisand Ogden, 1994; David, 1996; Busch et al., 1998). Anumber of broadly defined physical and biological fac-tors have been identified that influence the formation ofthe Everglades landscapemosaic (DeAngelis andWhite,1994; DeAngelis, 1994; Gunderson, 1994), including hy-dropattern, fire, storms, and water chemistry. In WCA 1,one of the least studied areas in the Everglades, some ofthe most prominent physical processes are affected byhydrology and water chemistry (Newman et al., 1997).The spatial soil patterns that we observed reflect thecurrent and historical processes that have predominatedin WCA 1.Of the threeWCAs in theEverglades,WCA1 is unique

in that it is mainly rainwater-fed (South Florida Water

Management District, 1992). Until recently, one of theother water inputs was canal water, which generally has ahigher pH (Table 4) than the water in the marsh interior.The canal water was fed by pump structures S-5A andinputs from S-6 (Fig. 1). A tertiary hydrological input wasthe ACME pump and its associated drainage basin to theeast ofWCA 1. The S-5A pumps drains a 59570-ha basin,while S-6 drains a 37810-ha basin in theEEA.TheACMEpump drains the ACME drainage basin, a 2800-ha areathat is composed of a mix of urban and recreationalland use areas. Currently, these inputs are either, in thecase of S-6, diverted, or in the case of S-5, filtered througha STA.

The soils in theWCA 1 are domed in shape and raisedrelative to the surrounding canals (Swift and Nicholas,1987), hence influx of canal water into the area is de-pendent on canal stage levels. At high canal stage levels,canal water enters the area along the periphery. At lowstage levels, the only water input to the area is rainfall.The water chemistry in the interior of WCA 1 tends tobe of a lower pH (Table 4). The acidity of surface waterin the interior of WCA 1 is the result of greater depth ofpeat than in other areas of the Everglades, and fromWCA 1 being a precipitation dominated, weakly buff-ered system. The variations in P water chemistry be-tween canal water and the interior of the marsh for theperiod 1993–2001 (Table 4) were less than the historic

Table 3. Correlation matrix of aggregated soil properties over the total sampled profile (floc, 0–10, and 10–20 cm).†

Floc depth Moist BD LOI TP TIP TN TC TCa TMg TFe

Floc depth 1Moist 20.24‡ 1BD 0.21‡ 20.96‡ 1LOI 0.11 0.13 20.11 1TP 20.11 0.49‡ 20.50‡ 20.32‡ 1TIP 20.11 0.52‡ 20.52‡ 20.34‡ 0.93‡ 1TN 0.00 0.16 20.10 0.23‡ 20.10 20.05 1TC 0.27‡ 20.19 0.22‡ 0.73‡ 20.60‡ 20.60‡ 0.24‡ 1TCa 0.16 20.51‡ 0.47‡ 20.61‡ 0.04 0.04 20.39‡ 20.34‡ 1TMg 0.08 20.33‡ 0.29‡ 20.68‡ 0.24‡ 0.22‡ 20.42‡ 20.53‡ 0.89‡ 1TFe 20.07 20.05 0.05 20.14 20.02 0.03 0.21‡ 20.05 0.03 20.14 1TAl 20.10 20.36‡ 0.37‡ 20.48‡ 20.11 20.08 0.12 20.30‡ 0.32‡ 0.32‡ 0.33‡

†Moist: moisture content; BD, bulk density; LOI, loss on ignition; TP, total phosphorus; TIP, total inorganic phosphorus; TN, total nitrogen; TC, total carbon;TCa, total calcium; TMg, total magnesium; TFe, total iron; TAl, total aluminum.

‡ Significant at the 0.0001 probability level.

Table 4. Summary of selected characteristics of water quality inWater Conservation Area 1, sampled in 2004 for the interiorand between 1993 and 2001 at the S6 canal. Source: DB Hydroand South Florida Water Management District.†

Parameter Canal water‡ Marsh interior

pH 7.35 6 0.03 (103) 6.68 6 0.05 (74)Conductivity, mS cm21 986 6 21 (118) 342 6 26 (91)

mg L21

TP§ 0.06 6 0.003 (97) 0.03 6 0.01 (74)Total soluble P 0.04 6 0.003 (79) 0.03 6 0.01 (70)Soluble reactive P 0.03 6 0.004 (87) 0.02 6 0.01 (74)Total N 2.2 6 0.08 (82) 1.41 6 0.06 (64)NH4–N 0.2 6 0.03 (78) 0.02 6 0.01 (63)NOx–N 0.3 6 0.05 (81) 0.02 6 0.01 (88)Ca 79 6 2.3 (88) 28 6 2.63 (89)Fe 71 6 5.3 (75) ND¶

†Values are means6 SE; numbers of samples (n) in shown in parentheses.‡Water collected in the S6 canal.§ Total phosphorus.¶None determined.

Reproducedfrom


reserved.


values reported by Newman et al. (1997) for the periodfrom 1978–1983. In the 1978 to 1983 period, canal waterTP, total soluble P, and soluble reactive P were reportedas 0.09 6 0.01, 0.06 6 0.01, and 0.05 6 0.01 mg L21,respectively, which are slightly higher than the valuesreported for the period 1993–2001 (Table 4).During the 1993 to 2001 period STA-1E filtered P,

resulting in lower P loads in the canal water enteringWCA 1. Other water chemistry properties were invari-ant over the time period. The P levels in the water sam-pled in the interior of the marsh have increased whencompared to the results obtained from Newman et al.(1997). This indicates that water chemistries in canalsand the marsh interior are now more similar (Table 4).A comparison of soil properties between the 1991

sampling (Newman et al., 1997) and the current sam-pling campaigns is confounded by the fact that theprevious sampling did not differentiate floc, which wasspecified in this study. Soil TC and TN did not varymuch between the sampling efforts (1991–2003), al-though in the current study soil N was 3 g kg21 higheracross both layers.The signatures in the spatial soil and floc property

distributions reflect both historic and current hydrologicconditions. Newman et al. (1997) found for their 1991sampling that areas adjacent to the western perimetercanal (L-7) were enriched with P, patterns that are simi-lar to those depicted in Fig. 4.Phosphorus has been recognized as the limiting nu-

trient in the Everglades ecosystem (Davis, 1991; Davisand Ogden, 1994) and its dynamics and occurrencethroughout many areas in the Everglades have been de-scribed extensively (DeBusk and Reddy, 1998; Newmanet al., 2001). The P in the canal water that is introducedinto this ecosystem is initially captured locally throughprocesses such as precipitation, sorption, and biologicaluptake (Qualls and Richardson, 2000). Subsequently, Pgradients are formed by continual influx of P, P satura-tion on the soils, and P remobilization and dispersion. In1991 in WCA 1 P showed a limited gradient from canalL-70 extending into the marsh interior (Newman et al.,1997). We found that the same areas on the westernportion of WCA 1 were still P enriched when comparedto the interior areas of the marsh. However, formalcomparison of the overall soil properties between thetwo studies is confounded by the fact that the previoussampling did not distinguish between floc and soil, whichwas specified in this study.DeBusk et al. (2001) identified a soil TP threshold

value for WCA-2a of 450 mg kg21 beyond which the Penrichment resulted in structural changes (e.g., cattailincursions) in the marsh vegetation communities. Only10% of the total area inWCA 1 exceeded that threshold,which is equivalent to an area of 55 ha. These areas vi-sually coincide with the areas containing cattail commu-nities as identified by Newman et al. (1997) in 1991.Further efforts should be directed at including a morecurrent vegetation community map. This would enablethe quantification of the effect of the prevailing vege-tation communities on the observed soil patterns beyondcattail incursion, which is particularly important in the

case of studying the relationships between existing soilcharacteristics and indigenous vegetation communities.

The largest portion of P inWCA 1 soils is organic withthe minor portion being inorganic P, which is consistentwith Everglades soils (Newman et al., 1997; DeBusket al., 1994; Qualls and Richardson, 1995). The largestinorganic P pool was found in the floc layer (approxi-mately 25% of TP), with a slightly smaller pool in thesoil layer (approximately 20% of TP). In other areasin the Everglades (e.g., Water Conservation Area 2a),inorganic P is co-precipitated with Ca and Mg, as a re-sult of the exposed limestone bedrock and relativelyhigh water column pH (DeBusk et al., 1994; Qualls andRichardson, 1995). In WCA 1, this binding behaviormight be more prevalent along the edges of the system,which are influenced by the canal water, than in themoreacidic interior. These dynamics were reflected in thespatial distributions of TIP and TCa (Fig. 4) as areasenriched in TIP also showed relatively high TCa levels.Further penetration of Ca into the conservation areamight be possible as Ca is not as readily taken up by biotaas P. The TCa maps showed higher Ca concentrationssouth of the western levee. In acidic systems, on the otherhand, Al and Fe chemistry controls P soil dynamics(Richardson, 1999). We found areas of Fe enrichmentalong the fringe of L-40 (Fig. 6), presumably from theACME drainage basin. Urban storm water runoff cancontain elevated levels of dissolved and colloidal Fe(Grout et al., 1999). In a smaller, localized area, coin-ciding with the highest TFe levels, we found an increasein soil BD (Fig. 3) and a decrease in LOI and soil TCcontent, presumably the result of the ACME basin in-puts and/or localized disturbances. However, TFe enrich-ment was not accompanied by P enrichment. Likewiseother soil properties including metals showed dissimilarspatial patterns.

Based on the hydroperiod and ponding depths pre-dicted by the South Florida Water Management Model(SFWMM; Stober et al., 2001), the southern portion ofWCA 1 has the longest flood periods, which steadilydecrease on the south-to-north axis. Hydropattern, in-cluding the duration, degree, and frequency of flooding,affects carbon decomposition (DeBusk andReddy, 2003).The hydropattern is, therefore, possibly the predominantfactor influencing TC, TN, TAl, LOI, and BD spatialpatterns as soil mineralization rates are higher in thenorthern areas of the marsh. This might also be related tothe higher soil C to N ratio in the southern areas of themarsh (Fig. 5). Nitrogen is limiting to microbial activitiesin high TP soils (Amador and Jones, 1995, 1997). How-ever, nitrogen content had not previously been found tovary considerably over nutrient gradients in the Ever-glades (Koch andReddy, 1992; Newman et al., 1997; Craftand Richardson, 1998; Vaithiyanathan and Richardson,1999). We found that the areas high in TP and TCa werealso relatively low in TN, which was expressed in termsof high localized floc and soil C to N ratios (Fig. 5), pos-sibly alluding to an increased N demand in these areas.In other areas of the Everglades, N dynamics are closelyassociated with the presence of nitrogen fixing cyano-bacterial periphyton mats (Inglett et al., 2004), but as

Reproducedfrom


reserved.


WCA 1 is a soft water system, other algal consortia prob-ably dominate the water column. The scale at which thesamples were taken in this study is probably too coarseto conclusively relate a particular N process to the predic-tion patterns. A detailed, finer spatial study focused onN cycling in these areas would be a more effective wayto overcome these divergences in scale (Bierkens et al.,2000). In a previous study in the same area, Newman et al.(2001) documented that P enrichment of soils in LNWRresult in increased NH4 concentrations in the porewaterand attributed this to mineralization of organic N in re-sponse to removal of P limitation. This porewater N poolwas subsequently depleted below background levels, pre-sumably in response to plant growth.

CONCLUSIONSThe spatially explicit analyses of floc and soil prop-

erties in WCA 1 enabled the identification of distinctpatterns likely to correspond to a variety of ecosystemprocesses such as nutrient cycling, resuspension, precip-itation, N fixation, and others. The prediction of certainproperties, such as floc BD, had a high degree of un-certainty. In contrast, soil properties generally exhibitedlower uncertainty than floc properties. The predominanthydrological north–south gradient in this area could bediscerned in the spatial distribution of a number of thesampled soil properties. Also evident in these patternswere distinct zones that have developed by the incursionof canal water resulting in east–west gradients of soil andfloc properties. We identified areas with enhanced levelsof TFe with the highest levels found adjacent to canalL-40. At approximately the same location, soil BD wasalso locally high. Areas exhibiting soil and floc nutrientenrichment were identified on the western side of themarsh, with the highest concentrations adjacent to canalL-7. Less than 5% of the area exceeded a 450 mg kg21

soil TP threshold. This study characterized the spatialvariability of physical and chemical soil properties inWCA 1 providing baseline values for future studies. Con-sidering the mapped geographic soil patterns obtainedin this study, and associations to environmental processes,this research supports ongoing restoration efforts by: (i)detecting vulnerable areas and (ii) identifying collocatedpatterns of soil characteristics, which can suggest hypoth-eses, or support underlying soil processes models. Thispaper was not able to address the effect of vegetationcommunities on the observed spatial patterns or the rela-tionship between the soil dynamics and plant/algal com-munities, but will effectively supply the foundation fromwhich these more detailed studies can be executed forWCA 1.

ACKNOWLEDGMENTS

This research was supported by the South Florida WaterManagement District and varied other federal sources. Theauthors would like to express their thanks to the analyticalassistance provided by Yu Wang, the insightful ideas, com-ments, and editorial help from Dr. Greg Bruland and the sup-port from Rosanna Rivero.

REFERENCESAmador, J.A., and R.D. Jones. 1995. Carbon mineralization in pristine

and phosphorus-enriched peat soils of the Florida Everglades. SoilSci. 159:129–141.

Amador, J.A., and R.D. Jones. 1997. Response of carbon mineraliza-tion to combined changes in soil moisture and carbon-phosphorusratio in a low phosphorus Histosol. Soil Sci. 162:275–282.

Andersen, J.M. 1976. An ignition method for determination of totalphosphorus in lake sediments. Water Res. 10:329–331.

Bierkens, M.F., P.A. Finke, and P. de Willigen. 2000. Upscaling anddownscaling methods for environmental research. Kluwer Aca-demic Publ., London.

Busch, D.E., W.F. Loftus, and O.L. Bass, Jr. 1998. Long-term hydro-logic effects on marsh plant community structure in the southernEverglades. Wetlands 18:230–241.

Craft, C.B., and C.J. Richardson. 1998. Recent and long-term organicsoil accretion and nutrient accumulation in the Everglades. Soil Sci.Soc. Am. J. 62:834–843.

David, P.G. 1996. Changes in plant communities relative to hydrologicconditions in the Florida Everglades. Wetlands 16:15–23.

Davis, S.E. III, C. Corronado-Molina, D.L. Childers, and J.W. Day. 2003.Temporally dependent C, N, and P dynamics associated with thedecay of Rhizophora mangle L., leaf litter in oligotrophic mangrovewetlands of the Southern Everglades. Aquat. Bot. 75:199–215.

Davis, S.M. 1991. Growth, decomposition and nutrient retention in ofCladium jamaicense Crantz and Typha domingensis Pers., in theFlorida Everglades. Aquat. Bot. 40:203–224.

Davis, S.M., and J.C. Ogden (ed.) 1994. Everglades: The ecosystemand its restoration. St. Lucie Press, Delray Beach, FL.

DeAngelis, D.L. 1994. Synthesis: Spatial and temporal characteristicsof the environment. p. 307–320. In S.M. Davis and J.C. Ogden (ed.)Everglades: The ecosystem and its restoration. St. Lucie Press,Delray Beach, FL.

DeAngelis, D.L., and P.S. White. 1994. Ecosystems as products of spa-tially and temporally varying driving forces, ecological processes,and landscapes: A theoretical perspective. p. 9–27. In S.M. Davisand J.C. Ogden (ed.) Everglades: The ecosystem and its restoration.St. Lucie Press, Delray Beach, FL.

DeBusk, W.F., S. Newman, and K.R. Reddy. 2001. Spatio-temporalpatterns of soil phosphorus enrichment in Everglades Water Con-servation Area 2A. J. Environ. Qual. 30:1438–1446.

DeBusk, W.F., and K.R. Reddy. 1998. Turnover of detrital organiccarbon in a nutrient-impacted Everglades Marsh. Soil Sci. Soc. Am.J. 62:1460–1468.

DeBusk, W.F., and K.R. Reddy. 2003. Nutrient and hydrology effectson soil respiration in a northern Everglades marsh. J. Environ.Qual. 32:702–710.

DeBusk, W.F., K.R. Reddy, M.S. Koch, and Y. Yang. 1994. Spatialdistribution of soil nutrients in a northern Everglades marsh: WaterConservation Area 2A. Soil Sci. Soc. Am. J. 58:543–552.

Environmental Systems Research Institute. 2005. ArcGIS. ArcMapGeographical Information System Version 9.0. ESRI, Redlands,CA.

Ettema, C.H., D.C. Coleman, G. Vellidis, R. Lowrance, and S.L.Rathbun. 1998. Spatiotemporal distributions of bacterivorous nema-todes and soil resources in a restored riparian wetland. Ecology79:2721–2734.

Geovariances. 2005. ISATIS. Geostatistical software package. Geo-variances, Houston, TX.

Grout, H., M.R. Wiesner, and J.Y. Bottero. 1999. Analysis of colloidalphases in urban stormwater runoff. Environ.Water Sci. Technol. 33:831–839.

Gunderson, L.H. 1994. Vegetation of the Everglades: Determinants ofcommunity composition. p. 323–340. In S.M. Davis and J.C. Ogden(ed.) Everglades: The ecosystem and its restoration. St. Lucie Press,Delray Beach, FL.

Inglett, P.W., K.R. Reddy, and P.V. McCormick. 2004. Periphytontissue chemistry and nitrogenase activity in a nutrient impacted Ever-glades ecosystem. Biogeochemistry 67:213–233.

King, R.S., C.J. Richardson, D.L. Urban, and E.A. Romanowicz. 2004.Spatial dependency of vegetation-environment linkages in an anthro-pogenically influenced wetland ecosystem. Ecosystems 7:75–97.

Koch, M.S., and K.R. Reddy. 1992. Distribution of soil and plant

Reproducedfrom


reserved.


nutrients along a trophic gradient in Florida Everglades. Soil Sci. Soc.Am. J. 56:1492–1499.

Lin, H., D. Wheeler, J. Bell, and L. Wilding. 2005. Assessment of soilspatial variability at multiple scales. Ecol. Modell. 182:271–290.

McCollum, S.H., V.W. Carlisle, and B.G. Volk. 1976. Historical andcurrent classification of organic soils in the Florida Everglades.Proc. Soil Crop Sci. Soc. Fla. 35:173–177.

Newman, S., H. Kumpf, J.A. Laing, and W.C. Kennedy. 2001. Decom-position responses to phosphorus enrichment in an Everglades(USA) slough. Biogeochemistry 54:229–250.

Newman, S., K.R. Reddy, W.F. DeBusk, and Y. Wang. 1997. Spatialdistribution of soil nutrients in a northern Everglades marsh: WaterConservation Area 1. Soil Sci. Soc. Am. J. 61:1275–1283.

Noe, G.B., D.L. Childers, and R.D. Jones. 2001. Phosphorus biogeo-chemistry and the impact of phosphorus enrichment: Why is theEverglades so unique. Ecosystems 4:603–624.

Qualls, R.G., and C.J. Richardson. 1995. Forms of soil phosphorus alonga nutrient enrichment gradient in the northern Everglades. Soil Sci.160:183–198.

Qualls, R.G., and C.J. Richardson. 2000. Phosphorus enrichment af-fects litter decomposition, immobilization and soil microbial phos-phorus in wetland mesocosms. Soil Sci. Soc. Am. J. 64:799–808.

Reddy, K.R., Y. Wang, W.F. DeBusk, M.M. Fisher, and S. Newman.1998. Forms of soil phosphorus in selected hydrologic units of theFlorida Everglades. Soil Sci. Soc. Am. J. 62:1134–1147.

Reddy, K.R., J.R. White, A. Wright, and T. Chua. 1999. Influence ofphosphorus loading on microbial processes in the soil and watercolumn of wetlands. p. 207–223. InK.R. Reddy and G.A. O’Connor(ed.) Phosphorus biogeochemistry of subtropical ecosystems. CRCPress, Boca Raton, FL.

Richardson, C.J. 1999. The role of wetlands in storage, release, andcycling of phosphorus on the landscape: A 25-year retrospective.p. 47–68. In K.R. Reddy et al. (ed.) Phosphorus biogeochemistry insubtropical ecosystems. Lewis Publ., Boca Raton, FL.

Schloeder, C.A., N.E. Zimmerman, andM.J. Jacobs. 2001. Comparisonof methods for interpolating soil properties using limited data. SoilSci. Soc. Am. J. 65:470–479.

South Florida Water Management District. 1992. Surface water im-

provement and management plan for the Everglades: Supportinginformation document. SFWMD, West Palm Beach.

State of Florida. 1994. Everglades Forever Act. Florida statutes.Available at http://www.sfwmd.gov/org/wrp/wrp_evg/projects/efa.html (verified 27 Feb. 2006). South FloridaWater Management Dis-trict, West Palm Beach.

State of Florida. 2005. Florida Administrative Code 62-302.540(3)(d).Chapter 62-302 SurfaceWater Quality Standards. Available at http://www.dep.state.fl.us/legal/Rules/shared/62-302.pdf (verified 20 Dec.2005).

Stober, S.Q.J., K. Thornton, R. Jones, J. Richards, C. Ivey, R.Welch, M.Madden, J. Trexler, and E. Gaiser. D., Scheidt, and S. Rathbun.2001. South Florida ecosystem assessment: Phase I/II. Evergladesstressor interactions: Hydropatterns, eutrophication, habitat alter-ation, and mercury contamination. Tech. Rep. EPA 904-R-01-003.USEPA, Tallahassee, FL.

Stolt, M.H.,M.H.Genthner,W.L. Daniels, and V.A.Groover. 2001. Spa-tial variability in palustrine wetlands. Soil Sci. Soc. Am. J. 65:527–535.

Swift, D.R., and R.B. Nicholas. 1987. Periphyton and water quality rela-tionships in theEvergladesWaterConservationAreas. Tech. Publ. 87-2. South Florida Water Management District, West Palm Beach, FL.

United States Fish and Wildlife Service. 2004. Arthur R. MarshallLoxahatchee National Wildlife Refuge. Refuge facts. Available athttp://loxahatchee.fws.gov/Refuge/quick-facts.asp (verified 27 Feb.2006). FWS, Washington, DC.

USEPA. 1993a.Methods for the determination of inorganic substancesin environmental samples. Environmental Monitoring Systems Lab,Cincinnati, OH.

USEPA. 1993b. Methods for the determination of metals in environ-mental samples. Environmental Monitoring Systems Lab, Cincin-nati, OH.

Vaithiyanathan, P., and C.J. Richardson. 1999. Macrophyte specieschanges in the Everglades: Examination along a eutrophicationgradient. J. Environ. Qual. 28:1347–1358.

Wackernagel, H. 2003. Multivariate geostatistics. Springer, Berlin.Webster, R. 2000. Is soil variation random? Geoderma 97:149–163.Webster, R., and M.A. Oliver. 2001. Geostatistics for environmental

scientists. John Wiley & Sons, New York.

Reproducedfrom


reserved.


assessment of the spatial distribution of soil properties in a ......stormwater treatment areas...

Documents