ww.sciencedirect.com

wat e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8

Available online at w

ScienceDirect

journal homepage: www.elsevier .com/locate /watres

Water quality of small seasonal wetlands in thePiedmont ecoregion, South Carolina, USA: Effects ofland use and hydrological connectivity

Xubiao Yu a,b, Joanna Hawley-Howard c, Amber L. Pitt d, Jun-Jian Wang b,Robert F. Balddin c, Alex T. Chow b,c,*

a School of Chemistry and Chemical Engineering, South China University of Technology, 510640, Chinab The Belle W. Baruch Institute of Coastal Ecology and Forest Science, Clemson University, Georgetown, SC 29440,

USAc Department of Forestry and Environmental Conservation, Clemson University, Clemson, SC 29634, USAd Department of Biological & Allied Health Sciences, Bloomsburg University of Pennsylvania, Bloomsburg, PA 17815,

USA

a r t i c l e i n f o

Article history:

Received 2 June 2014

Received in revised form

29 December 2014

Accepted 5 January 2015

Available online 13 January 2015

Keywords:

Dissolved organic matter

Fluorescence

Isolated wetlands

Nutrients

* Corresponding author. The Belle W. BaruchUSA. Tel.: þ1 843 546 1013x232; fax: þ1 843

E-mail address: achow@clemson.edu (A.Thttp://dx.doi.org/10.1016/j.watres.2015.01.0070043-1354/© 2015 Elsevier Ltd. All rights rese

a b s t r a c t

Small, shallow, seasonal wetlands with short hydroperiod (2e4 months) play an important

role in the entrapment of organic matter and nutrients and, due to their wide distribution,

in determining the water quality of watersheds. In order to explain the temporal, spatial

and compositional variation of water quality of seasonal wetlands, we collected water

quality data from forty seasonal wetlands in the lower Blue Ridge and upper Piedmont

ecoregions of South Carolina, USA during the wet season of February to April 2011. Results

indicated that the surficial hydrological connectivity and surrounding land-use were two

key factors controlling variation in dissolved organic carbon (DOC) and total dissolved

nitrogen (TDN) in these seasonal wetlands. In the sites without obvious land use changes

(average developed area <0.1%), the DOC (p < 0.001, t-test) and TDN (p < 0.05, t-test) of

isolated wetlands were significantly higher than that of connected wetlands. However, this

phenomenon can be reversed as a result of land use changes. The connected wetlands in

more urbanized areas (average developed area ¼ 12.3%) showed higher concentrations of

dissolved organic matter (DOM) (DOC: 11.76 ± 6.09 mg L�1, TDN: 0.74 ± 0.22 mg L�1,

mean ± standard error) compared to those in isolated wetlands (DOC: 7.20 ± 0.62 mg L�1,

TDN: 0.20 ± 0.08 mg L�1). The optical parameters derived from UV and fluorescence also

confirmed significant portions of protein-like fractions likely originating from land use

changes such as wastewater treatment and livestock pastures. The average of C/N molar

ratios of all the wetlands decreased from 77.82 ± 6.72 (mean ± standard error) in February

to 15.14 ± 1.58 in April, indicating that the decomposition of organic matter increased with

the temperature. Results of this study demonstrate that the water quality of small, sea-

sonal wetlands has a direct and close association with the surrounding environment.

© 2015 Elsevier Ltd. All rights reserved.

Institute of Coastal Ecology and Forest Science, 130 Heriot Road, Georgetown, SC 29440,546 6296.. Chow).

rved.

wa t e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8 99

1. Introduction

Wetlands performa suite of ecological functions such aswater

purification, nutrient retention, floodprotection, groundwater

recharge and habitat for wildlife. Based on different scientific

and policy objectives, various ecological functions of wetlands

have been studied and evaluated, including hydrology (Cook

and Hauer, 2007; Min et al., 2010), water quality (Trebitz

et al., 2007; Verhoeven et al., 2006; Whigham and Jordan,

2003), vegetation composition (Hebb et al., 2013) and animal

population dynamics (Fracz and Chow-Fraser, 2013; Seil-

heimer and Chow-Fraser, 2006). Among these factors, water

quality has generated substantial concern because it is not

only essential for all biological growth and reproduction, but

also directly interacts with other factors across multiple

spatial and temporal scales. For example, land use (Maassen

et al., 2012; Morrice et al., 2008), hydrological connections

(Cook and Hauer, 2007; Kazezyilmaz-Alhan et al., 2007), soil

and vegetation characteristics (Batzer et al., 2000;Montgomery

and Eames, 2008) have all been demonstrated to be related to

water quality. Thus, water quality serves as an important in-

dicator for characterizing the effects of land use on wetlands,

and furthermore, to direct policy and management of the

surrounding watershed. Particular water quality parameters

are selected as indicators of specific anthropogenic stressors,

such as dissolved organic carbon (DOC) with natural organic

matter (NOMs), nitrogen (N) and phosphorous (P) for agricul-

tural run-off, electronic conductivity (EC) for anthropogenic

discharge and chloride (Cl�) for point-source pollution (Tu,

2011). In most cases, dissolved carbon and nitrogen are sig-

nificant parameters due to their high content in all dissolved

organic matter (DOMs).

Small, seasonal wetlands are found throughout the Pied-

mont ecoregion of the southeastern United States. These

wetlands usually occur in areas with depressional topography

and are characterized by their small area and shallow depth

(Hayashi and van der Kamp, 2000). These wetlands typically

have a shorter hydroperiod than larger wetlands with greater

surfacewater area, and usually holdwater for only three to five

months per year (Hayashi and van der Kamp, 2000). Because of

their relatively small size and ephemeral nature, these wet-

lands are not typically protected as fully as coastal and riparian

wetlands, which satisfy the criteria for protection under the

Clean Water Act (McCauley et al., 2013; Stokstad, 2006). In

recent decades, wetlands have suffered significant habitat loss

and degradation across the US due to the rapid urbanization

processes (McCauley et al., 2013). Despite small surface areas

and short hydroperiods, these wetlands harbor a great deal of

biological diversity, especially for some amphibians (e.g., wood

frogs, spotted salamanders, spade foot toads) for which there

are selective pressures to breed in more ephemeral sites

(Stokstad, 2006). Moreover, the stagnant nature of seasonal

wetlands provide more time for microbes to convert excess

nutrients and prevent downstream algal blooms (Whitmire

and Hamilton, 2005). Based on the unique benefits for both

biodiversity and downstreamwater quality, many researchers

are calling for the amendment of wetland regulations to

includeprotectionof small, seasonalwetlands (Stokstad, 2006).

Research on water quality of small, seasonal wetlands has

been limited as a result of the small water surface area and

short hydroperiod that characterize these habitats. With re-

gard to hydrological characteristics, these seasonal wetlands

are generally considered to have little or no connection with

surface water or ground water, and they are primarily

dependent on precipitation and runoff (Whigham and Jordan,

2003). Because of this dependency on precipitation, these

wetlands may represent a simpler water pathway than that of

large wetlands (i.e., water quality of small wetlands may be

more closely linked to the characteristics of the surrounding

watershed such as land-use and vegetation cover). Moreover,

thesewetlands hold significantly lower volumes of water than

other wetlands and the water quality parameters may there-

fore respond to variation in the surrounding environment in a

more rapid and sensitive manner.

In order to better understand the water quality of these

seasonal wetlands, we collected water samples from forty

seasonal wetlands located in the lower Blue Ridge and upper

Piedmont ecoregions of northwestern South Carolina from

February through April of 2011. In addition to general water

parameters, DOM was quantified and characterized using UV

spectra and fluorescence to determine the compositional

variations among land-use and hydrological factors. We hy-

pothesized that small, seasonal wetlands have a strong rela-

tionship with the surrounding watershed, and can serve as a

sensitive indicator to evaluate the water quality at the land-

scape scale and also reflect the influences of human activities

on water quality rapidly.

2. Materials and methods

2.1. Study sites

Wetland study sites were distributed throughout north-

western South Carolina (34�480N, 82�560W), across Oconee,

Pickens, Anderson and Greenville Counties (Fig. 1). This region

is located at the base of the Appalachian Mountains and near

the headwaters of the Savannah River (boundary river of

South Carolina and Georgia). The water quality in this upper

portion of the watershed may influence the middle and lower

reaches where there are large urban populations. There is a

paucity of data regarding seasonal wetlands in this region,

and until recently few had been mapped (Pitt et al., 2012).

Increasing development pressure since 1990 associated with

the region's location between two major urban centers

(Atlanta, GA and Charlotte, NC) (Campbell et al., 2008), pro-

longed drought cycles, and increasing constraints on aquatic

resources necessitate timely assessment of small, seasonal

wetlands in this region. The southern Appalachians span the

northern portion of the study area and the general topography

transitions from steep mountainous terrain in the lower Blue

Ridge ecoregion to hilly foothills in the upper Piedmont ecor-

egion. Wetland elevation decreased approximately 700 m

from north to south. The climate in this portion of the

southeastern US is temperate forest and the annual precipi-

tation level is approximately 1700 mm.

Fig. 1 e Geographical position of studied seasonal wetlands in northwestern South Carolina, US. The circle and square

symbol represents the connected and isolated wetlands, respectively.

wat e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8100

The key criteria for selecting study wetlands were small

area (<0.5 ha) and relatively shallow depths (<200 cm) (Pitt

et al., 2012). The largest wetland had an area of 3542 m2

(S29) and the smallest wetland had an area of 3.0 m2 (S37),

with the median value of 95 m2. The depth of all the wetlands

was less than 40 cm (Table S1).

There were 40 wetland sites in this study, labeled S1 to S40

from South to North approximately. In Fig. 1, the region

highlighted with a circle represents a more developed area

adjacent to regional cities [e.g., Pendleton, SC

(population z 3000) and Anderson, SC (population z 27,000)]

and also in close proximity to pastureland, both of which

generate heavy anthropogenic inputs (e.g., urban stormwater

and agricultural run-off). Wetlands S1 to S6 are located in this

more developed region, with S3 to S6 located close to a

highway. Four wetlands (S7eS10) are located in the Clemson

Experimental Forest (7082 ha mixed use, managed forest

owned by Clemson University but open to public recreational

activities) and most of the studied wetlands (S7eS40) are

located on publicly owned state or federal lands across the

lower Blue Ridge and upper Piedmont (Fig. 1). Considering the

different extents of urbanization, the wetlands located in the

more-developed areas (i.e., S1, S2, S3, S4, S5 and S6), were

grouped separately from the otherwetlands. Furthermore, the

wetlands were sub-grouped as either connected or isolated

based on surficial hydrological connectivity.

The primary source of water input for the seasonal wet-

lands in the study area is precipitation and evaporation is the

main transportation pathway of water. These conditions

mean that most study wetlands only hold water during the

wa t e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8 101

rainy season (JanuaryeMay). From a hydrogeomorphic

perspective, wetlands were field characterized as having

ephemeral or permanent connections with surface waters

(referred to from here forward as “connected wetlands”), or

being relatively isolated from other bodies of water (referred

to from here forward as “isolatedwetlands”) using established

wetland delineation criteria (Department of the Army, 1987).

We collected data on land use/land cover (LULC) by

accessing the National Land Cover Database (http://www.

usgs.gov/climate_landuse/) and downloading the 2010 data

set for the region.We combined these data with our study site

maps and calculated the total area of 1) open water, 2)

disturbed/successional area, 3) cultivated area, 4) developed

area, 5) forested area and 6) pastureland at the 250 m, 500 m

and 1 km scale around eachwetland. All spatial analyses were

performed using Esri ArcMap (10.1, 2012). It should be noted

that the disturbed/successional area was characterized as an

area that was recently cleared or in the early stages of suc-

cession, as the result of some types of land use changes. The

open water area refers to major rivers and lakes located near

the wetlands.

2.2. Sample collection

Water samples were collected monthly in February (7the8th),

March (8the9th), and April (13the14th) in 2011. Water sam-

ples from all 40 wetlands were collected within a two

consecutive day sampling round each month. No obvious

precipitation occurred during the two day sampling events.

Temperature (�C) and dissolved oxygen (DO,mg L�1) of surface

water in wetlands were collected at each wetland using a

multi-parameter water quality meter (YSI 650MDS, USA). At

each site, 1 L of surface water was collected using acid-

washed, Nalgene bottles that were pre-rinsed with sample

water. Three bottles were collected at each wetland as repli-

cations. The water samples were kept on ice during trans-

portation from field sites to the laboratory.

2.3. Water quality

Electrical conductivity (EC) and pH of non-filtered samples

were measured upon arrival in the laboratory. The measured

EC was corrected to EC at 25 �C (EC25) using the temperature

correction equation: EC25 ¼ ECt/[1 þ 0.021(t-25)] (Hayashi,

2004). Water samples were filtered through a syringe filter

(Millipore Millex-HV Hydrophilic PVDC, 0.45 mm). All the

filtered samples were stored at 4 �C for the next chemical

analysis within a month.

Each filtered samplewas analyzed for DOC and TDNusing a

Shimadzu TOC/TN analyzer and the C/N molar ratio was

calculated accordingly. DOC was further characterized by Shi-

madzu UV-1800 visible and ultraviolet spectrophotometer

scanning from 200 to 700 nm. Specific ultraviolet absorbance

(SUVA254)wascalculatedbynormalizingultraviolet absorbance

at 254 nm (UVA254) to DOC concentration, recorded as

LmgC�1 cm�1. The E2:E3 ratio, absorbance at 254 nmdivided by

absorbance at 365 nm, was calculated. E2:E3 is related to the

molecular sizeofDOM(Helmsetal., 2008). The3-Dfluorescence

excitationeemission matrixes (EEM) were obtained from Shi-

madzu spectrofluorometer RF5301 and analyzed by

fluorescence regional integration basedonSimpson's Rule (FRI-SR) (Zhou et al., 2013) for selective samples. All samples were

diluted with Milli-Q water to an ultraviolet absorbance at

254 nm of 0.3 or less (Miller et al., 2010) to obtain the reliable

EEMs. The fluorescence scans (excitation wavelength range:

240e450 nm; emission wavelength range: 350e550 nm) for

DOM were measured with a 5-nm slits setting for both excita-

tion and emission. The raw EEM was corrected (Murphy et al.,

2010) for instrument-dependent effects, inner-filter effects,

and background effects, and standardized to the Raman's units(normalized to the integral of the Raman signal between 390

and 410 nm in emission at a fixed excitation of 350 nm).

The FRI program divides excitationeemissionmatrix (EEM)

spectroscopy into different regions based on the character of

different DOMs, and the normalized volumes of different re-

gions represent content of different DOMs (Chen et al., 2003).

Accordingly, a modified FRI program was developed which

characterized four typical organic matters as: aromatic pro-

tein region (RAP, lemission: 220e350 nm, lexcitation: 220e250 nm),

soluble microbial by-product-like materials region (RMB,

lemission: 220e350 nm, lexcitation > 250 nm), fulvic acid-like re-

gion (RFA, lemission >350 nm, lexcitation: 220e250 nm) and humic

acid-like materials region (RHA, lemission >350 nm,

lexcitation > 250 nm).

2.4. Data analysis

All water quality data were tested for normality using the

ShapiroeWilk goodness-of-fit test (p > 0.05). Differences in

water quality parameters among the sites were determined

using independent t-tests. Temporal variations in water

quality parameters were determined using paired-sample t

tests. To evaluate the influences of adjacent land use and

hydrological connectivity, we used the log-transformed water

quality parameters in amultiple-regression analysis. It should

be noted that proportion of forested area was excluded from

the regression analysis as we observed multicollinearity with

other land-use variables. DOC, TDN, C/N ratio and EC were

used as variables in the regression analysis. For comparing the

similarity of water quality for the wetlands, cluster analysis

was applied to the general water-quality data (EC, DOC, TDN

and C/N ratio) of 40 seasonal wetlands. Hierarchical cluster

analysis was performed on the normalized data set by the

Ward's linkage method, using Euclidean distances as a mea-

sure of similarity. All statistical analyses were performed

using SPSS 19.0 software (IBM Co. Ltd, USA).

3. Results

3.1. Surrounding land-use of wetlands

Fig. 2 shows the proportion of six land-use categories

(developed, cultivated, pastured, disturbed, open water and

forest) surrounding the wetlands at 250, 500 and 1000 m

scale. The developed, cultivated and pastureland areas

were the major land use changes surrounding our study

wetlands. Most wetlands that were located in close prox-

imity to the more developed city region had relatively high

proportions of developed area (S3: 16%, S4: 22%, S5: 5% and

Fig. 2 e The proportion of land-use surrounding the seasonal wetlands (highlighted wetlands are located inmore-developed

region). From inside to outside, three concentric rings are representative of 250, 500 and 1000 m surrounding area,

respectively.

wat e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8102

S6: 7% in 1000 m scale) as compared to wetlands located

further from this region (most wetlands having less than

0.1% developed area at the spatial scales studied). There

was no evidence of large-scale agricultural activity in the

study area but pastureland did constitute a higher propor-

tion of adjacent lands. There are several longitudinal,

anthropogenic lakes in this region such as Lake Jocasee,

Lake Keowee and Lake Hartwell, which connect to each

other and flow into the upper reaches of the Savannah

River. Forest is the dominant land cover type in this region

with the exception of the sites close to the city centers of

Clemson and Pendleton, South Carolina. Some wetlands

located in the upper Piedmont and central Clemson Exper-

imental Forest area (S7, S11, S35 and S39) had forest cover

levels greater than 90%.

Fig. 3 e DOC, TDN and C/N ratio of the seasonal w

3.2. Temporal variation of water quality

DOC, TDN and C/N ratio (molar ratio) of study wetlands be-

tween February and April are shown in Fig. 3.We found a clear

difference in DOC and TDN between isolated and connected

wetlands in both less-developed and more-developed areas

within the study region. For the isolated and connected wet-

lands in less-developed region (i.e., S7 through S40) t-test re-

sults indicated that these wetlands had statistically

significant higher DOC (p < 0.001) and TDN (p < 0.05) con-

centrations compared to those of connected wetlands. For the

wetlands in the more-developed area, we observed a nearly-

opposite trend. Although there was no statistically signifi-

cant difference found due to the obvious inter-individual

variation, the connected wetlands such as S5 and S6

etlands in February, March and April, 2011.

Fig. 4 e Dendrogram showing clustering of 40 seasonal

wetlands based on water qualities (DOC, TDN, C/N ratio)

and land-use proportions.

wa t e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8 103

expressed about 3 times higher DOC than isolated wetlands;

and S3, S5 and S6 expressed about 3 times higher TDN than

the isolated wetlands. For C/N ratio, there was no statistically

significant difference between isolated and connected wet-

lands in either less-developed or more-developed region.

Except for C/N ratio, data of DOC, TDN and EC had statis-

tically significant correlation with the land-use or water

connection. It can be seen from Table 1 that hydrological

connectivity exhibited one order of magnitude larger co-

efficients than land-use factors regarding DOC and TDN.

Among the land-use factors, the proportion of developed and

pastureland area had decisive effects on DOC and TDN. The

proportion of pastureland area has a stronger associationwith

the variation of DOC than that of more developed areas

although proportion of developed area had stronger overall

associations with TDN. For EC, proportion of developed area

was the only factor strongly related to its variation. Fig. 4

shows the cluster analysis-generated three groups of these

sites based on EC, DOC, TDN, and C/N ratio. It can be found

that S3, S4, S5 and S6were grouped in a cluster (Cluster III). For

the sites in less-developed region, isolated wetlands (S7eS22)

and connected wetlands (S23eS40) were grouped to Cluster I

and Cluster II approximately.

The temporal variations of water quality indicators for con-

nected versus isolated wetlands are illustrated in Fig. 5. Water

temperature showed a significant increase from February to

April (paired t-test, p < 0.001), whereas DO showed irregular

fluctuationwithno consistent trend through time.DOC showed

no clear trend for all the samples, and paired-samples t-test

indicated that therewas no statistically significant difference in

monthly comparisons. TDN analyses show that there was an

increasing trend of TDNconcentration fromFebruary toApril in

bothconnectedand isolatedwetlands (p<0.001), indicating that

therewas a statistically significant difference betweenFebruary

and April. As a result of the variation of DOC and TDN, the ratio

of carbon and nitrogen exhibited a clear decline with the time.

The paired t-test result (p < 0.001) indicated a significant dif-

ference of monthly C/N ratios.

3.3. Optical properties of DOM

3.3.1. UV absorbanceSpecific ultraviolet absorbance at 254 nm (SUVA254) is a useful

parameter to estimate the aromaticity of natural water and is

Table 1 e Best fit multiple-regressionmodels relating log-transvariables of seasonal wetlands located in the Piedmont and Bl

Water quality DOC (mg L�1)a

Factors 250 m 500 m 1000 m 250 m

Develop e e 0.028* e

Pasture 0.024* 0.029* 0.036** e

Open water �0.032* e

Cultivated e e e e

Disturbed e e e e

Water connection 0.529** 0.646** 0.613** 0.613**

r2 0.53** 0.71** 0.69** 0.29*

*: Sig. <0.05, **: Sig. <0.001.a Average values of all the sampling month were used in multiple-regres

determined by the ratio between UVA254 and DOC (Weishaar

et al., 2003). Fig. 6 shows the relationship between UVA254

and DOC for all the water samples collected over a three

month period, and the aromatic property can be approxi-

mated with the calculated slope. It can be found that the forty

wetlands exhibited a good relevance betweenUVA254 andDOC

(r2 ¼ 0.902, p < 0.001). Generally, the connected wetlands

appeared in lower UVA254 (<0.15 cm�1) and DOC (<3.0 mg L�1)

region whereas the isolated wetlands appeared in higher re-

gion (UVA254 > 0.15 cm�1, DOC>3.0 mg L�1). The connected

wetlands with higher value of UVA254 (>0.15 cm�1) and DOC

(>3.0mg L�1) were located in themore-developed region of the

study area (i.e., the area surrounding Clemson, SC).

Fig. 7a summarizes the data of SUVA254 and E2:E3 for all the

study wetlands. For E2:E3, it can be found that the isolated

wetlands in less-developed area exhibited more stable results

both temporally and spatially compared to other sites. On the

contrary, the E2:E3 results of S1 to S6 showed both temporally

and spatially-obvious patterns. The variation of E2:E3 from S1

to S6 had an inverse relationship with SUVA254; the relation-

ship between E2:E3 and SUVA254 for less-developed and more

more-developed region are shown in Fig. 7b.

formed water quality to the land-use and water connectionue Ridge ecoregions of South Carolina, USA.

TDN (mg L�1)a EC (mm cm�1)

500 m 1000 m 250 m 500 m 1000 m

e 0.044* 0.353* 0.414* 0.382*

e 0.027** e e e

e e e e e

e e e e e

e e e e e

0.463* 0.307* e e e

0.44* 0.53** 0.54** 0.66** 0.55**

sion.

Fig. 5 e Box plots of temperature and water qualities in different months. Description of box plots: top and bottom of

box ¼ 75th and 25th percentiles, respectively; top and bottom whisker end ¼ 90th and 10th percentiles, respectively; solid

line in box ¼ median value; dotted line ¼ mean value.

wat e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8104

There was a strong correlation (r2 ¼ 0.83, p < 0.05) between

E2:E3 and SUVA254 for wetlands in the more-developed region,

whereas the wetlands in less-developed region did not exhibit

a linear relationship. This further demonstrates that the var-

iations of E2:E3 and SUVA254 for wetlands of S1 to S6 shown in

Fig. 7b were correlated, but the variations of wetlands in the

less-developed region fluctuated irregularly.

3.3.2. FluorescenceExcitationeEmission matrix (EEM) fluorescence spectroscopy

of DOMs is an effective analytical approach for identifying the

character and source of natural organicmaterials (NOMs). The

EEM fluorescence spectroscopies for representative wetlands

in different regions are shown in Fig. S1. As demonstrated by

Fig. 6 e Relationship of UV254 and DOC for all the wetlands

in different months (P < 0.05).

the sample EEM fluorescence spectroscopies, wetland in the

more-developed region (e.g. S4) expressed expressed stronger

RAP and RMB peaks but weaker RFA and RHA peaks compared

with wetland in less-developed regions. RAP and RMB are

mostly contributed by the anthropogenic discharge such as

treated wastewater; RFA and RHA are mainly derived from

natural resources such as forest and soil (Chen et al., 2003).

Here, the sum of RAP and RMB (RAPþMB) and the sum of RFA and

RHA (RFAþHA) were calculated to evaluate the distribution and

associated percentages of anthropogenic and natural DOMs in

the selected wetlands, respectively (Table 2). Wetlands of S4

(49%), S5 (61%), and S6 (30%) exhibited higher RAPþMB results

than the other wetlands. Even though S1 was located in the

same geographical area as S4, S5, and S6, it exhibits com-

monalities (19%) with wetlands in less-developed region. The

difference of RAPþMB between isolated and connected wet-

lands was not found to be significant; with isolated wetlands

S8 (17%), S14 (14%) and S15 (20%) and connected wetlands: S24

(16%), S33 (18%) and S38 (15%) showing no significant

differences.

4. Discussions

4.1. Effects of connectivity on water quality

The seasonal wetlands investigated in this study were all

depressional wetlands that receive drainage from their sur-

rounding watersheds. The information in Table S1 indicates

that these seasonal wetlandswere small and shallow, and rely

on precipitation and runoff as their primary water sources.

Themajor differences among thesewetlandswere found to be

that part of the wetlands had some connection with the

Fig. 7 e (a): E2: E3 and SUVA254 of the forty seasonal wetlands. The vertical bars represent monthly variations, not the

analytical error (analytical error was usually <3%); (b): linear regression of E2:E3 and SUVA254 for the seasonal wetlands. Solid

triangle, open circle and solid circle represent the wetlands in more-developed region, connected wetland and isolated

wetlands in less-developed region, respectively.

wa t e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8 105

surface water, but the others were almost entirely geograph-

ically isolated. As a result of higher levels of water exchange,

the connected wetlands exhibited higher DO levels than iso-

latedwetlands (Fig. 5d).We did not find statistically significant

differences in pH, temperature and EC between connected

and isolated wetlands. However, the differences of DOC and

TDN did reveal significant differences between isolated and

connected wetlands (t-test, p < 0.001).

The isolated wetlands in less-developed areas exhibited

higher DOC and TDN as compared to the connected wetlands.

Multiple-regression results (Table 1) indicate that water

connection was the primary driving force for the

concentration of DOC and TDN. Result of cluster analysis also

demonstrates that the isolated and connected wetlands in

less-developed region showed different characteristics on the

basic water quality. It is reasonable to believe that lack of

water connections made the isolated wetlands more effective

for retaining higher amounts of DOMs. In other words, part of

the dissolved carbon and nitrogen could be carried to down-

stream areas from the connected wetlands. The utility of

isolated wetlands for trapping pollutants and nutrients have

been reported in other studies (Whitmire and Hamilton, 2005).

C/N ratios were not significantly different between isolated

and connected wetlands. Since C/N ratio is more dependent

Table 2 e Proportions of FRI regions for representativewetlands in less and more-developed regions within thePiedmont and Blue Ridge ecoregions of South Carolina,USA.

Wetlands Anthropogenicresource

Natural resource

RAP

(%)RMB

(%)RAPþMB

(%)RFA

(%)RHA

(%)RFAþHA

(%)

S1 9.0 9.8 18.8 37.9 43.3 81.2

S4 23.9 16.0 40.0 34.7 25.3 60.0

S5 2.6 58.2 60.8 16.8 22.4 39.2

S6 16.1 13.4 29.5 39.7 30.8 70.5

S8 7.8 8.9 16.8 38.2 45.0 83.2

S14 8.4 5.1 13.5 56.9 29.6 86.5

S15 10.6 9.0 19.6 36.3 44.1 80.4

S24 8.4 7.7 16.1 39.3 44.6 83.9

S33 8.7 8.9 17.6 39.0 43.5 82.4

S38 14.5 0.7 15.2 50.1 34.8 84.8

RAP: percentage of aromatic protein region; RMB: percentage of sol-

uble microbial by-product like materials region; RFA: percentage of

fulvic acid-like region; RHA: percentage of humic acid-like materials

region.

wat e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8106

on the microbial activity including ammonification and

immobilization (Bachand and Horne, 2000; Ehrenfeld and Yu,

2012), the influence of water connection was not a primary

driving factor in this particular instance. Moreover, the fitting

of DOC and SUVA254 for all the wetlands showed a strong

relationship (r2 ¼ 0.90, p < 0.001), which further illustrates that

there was no significant difference in the aromaticity of DOMs

between isolated and connected wetlands. The FRI result also

demonstrates a similar proportion of DOMs for both isolated

and connected wetlands without anthropogenic effects (Table

2).

The effects of hydrological connectivity in wetlands

located in more-developed areas were somewhat different

than what we observed in less-developed areas. Some con-

nectedwetlands (S3, S5 and S6) showed higher concentrations

of DOC and TDN than isolated wetlands. The multiple

regression results also indicated that land-use factors such as

developed and pastureland areas had an important effect on

water quality with the exception of hydrological connectivity.

Particularly, the developed proportion of land surrounding

study wetlands had a decisive effect on EC levels. The effects

of land-use on water quality will be further discussed below.

4.2. Nutrient dynamics

The temporal variation of the water quality can be discussed

on the context of both concentration and composition. Firstly,

the concentration variations of DOC and TDN for seasonal

wetlands were different. From February to April, most wet-

lands exhibited small variation on DOC, fluctuating between

5.71 ± 0.95 and 6.03 ± 1.30 mg L�1 (mean ± standard error,

n ¼ 40). In contrast, TDN showed a statistically significant

increase, from 0.14 ± 0.04 to 0.34 ± 0.05 mg L�1

(mean ± standard error, n ¼ 40). In natural water, the major

source of dissolved nitrogen comes from the decomposition of

plant detrital and soil matters to inorganic nitrogen with the

microbial catabolic activity (Fisk and Fahey, 2001). The

increased levels of TDN in seasonal wetlands were primarily

the result of microbial decomposition of leaf litter and soil

organic matter. During the rainy season, the temperature

increased from February to April, which could enhance mi-

crobial activity.

Secondly, variation of water composition in seasonal

wetlands can be analyzed using C/N ratio and optical prop-

erties. The C/N ratio decreased from 77.82 ± 1.38

(mean ± standard error, n ¼ 40) in February to 15.14 ± 0.26

(mean ± standard error, n ¼ 40). The significant variation of C/

N ratio further demonstrates the effect ofmicrobial activity on

N-transformation (Bedard-Haughn et al., 2006). However, re-

sults of SUVA254 and E2:E3 demonstrated irregular fluctuation

without an obvious trend throughout time. This result in-

dicates that the composition of DOC was relatively stable

compared with that of TDN, demonstrating that microbial

activity was not a driving force of DOC transformation.

4.3. Effect of land-use on water quality

Anthropogenic factors, which can be measured by remote

sensing as land-use, have an important influence on thewater

chemistry in different wetlands (Morrice et al., 2008; Reiss,

2006). Our results also indicated the substantial effect of

land-use factors on the water quality of seasonal wetlands.

Except for the hydrological connectivity, developed and

pastureland areas showed higher contributions of DOC and

TDN than other types of land uses. Previous researches have

demonstrated a positive relationship between developed/

urban areas and dissolved carbon and nitrogen in wetlands

(Haidary et al., 2013; Morrice et al., 2008). For pastureland area,

the positive relationship with dissolved carbon and nutrients

has also been found in streams (Buck et al., 2004) and large-

scale wetlands (Galbraith and Burns, 2007; Graves et al.,

2004). In the more-developed region, S3, S4, S5 and S6 were

adjacent to a highway (<500 m), which can be a primary

drainage pathway from areas with intense residential devel-

opment, whereas S1 and S2 were far away from the highway.

Results above indicate that water quality of small, seasonal

wetlands was sensitive to the land use of adjacent areas and

the land use could have a greater impact than the hydrological

connectivity. Furthermore, the proportion of developed area

was the only factor correlated with the variation of EC. As an

indicator of the total ionic level in water, EC is believed to

more strongly relate to anthropogenic discharge such as

sewage, runoff from lawns and other input sources (Haidary

et al., 2013). The exclusive correlation between the propor-

tion of developed area and EC further demonstrates the sub-

stantial influence of land-use on the water quality of adjacent

wetlands. In addition, there was no interaction between land-

use and C/N ratio in the multiple-regression analysis. As dis-

cussed in 4.2, C/N ratio is more dependent on the microbial

activity, thus the factors of land-use were not the driven force

to its variation. Cluster analysis based on DOC, TDN and EC

grouped wetland S3, S4, S5 and S6 as one cluster from other

wetlands, further demonstrating the clear effects of land use

on water quality of these seasonal wetlands.

Besides the effect on concentrations of DOC and TDN,

land-use factors may also affect the composition of DOM.

There are extensive studies which discuss the relationship

wa t e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8 107

between land use changes and water quality (Cabezas et al.,

2009; Panno et al., 1999; Tu, 2011), but few studies have

focused on the impacts of anthropogenic forces on DOM'scomposition in wetlands. The fitting of UVA254 and DOC with

all of the data showed good relevance, indicating that the

water in all studied wetlands had similar aromaticity. Since

the forty seasonal wetlands are located in a small region

(70 � 70 km), the features of climate, vegetation and soil for

these wetlands are similar. Result of E2:E3 for wetlands in the

more-developed region showedmore dramatic variation than

that of wetlands in the less-developed region. This result in-

dicates that anthropogenic stress had some substantial in-

fluence on the composition of DOMs. Wetlands in more-

developed areas showed a strong negative correlation be-

tween E2:E3 and SUVA254, whereas no linear correlation was

observed for wetlands in the less-developed region (Fig. 7b).

Normally, SUVA254 increases with the molecular size of DOM,

thus it is inversely correlated with E2:E3 for DOM derived from

different resources (Helms et al., 2008). Wetlands located in

more-developed areas drain higher levels of anthropogenic

runoff, which may contain more pollutants of smaller mo-

lecular size such as aromatic proteins, thus resulting in the

significant negative correlation between E2:E3 and SUVA254.

Alternatively, without the substantial input of these pollut-

ants, wetlands in less-developed areas showed irregular

variation of E2:E3 and SUVA254. The FRI results show that the

proportions of RAP and RMB of wetlands in the more-developed

region (43 ± 9%, mean ± standard error, n ¼ 3) were higher

than that of wetlands in the less-developed region (17 ± 1%,

mean ± standard error, n ¼ 7). The higher percetages of RAP

and RMB correspond with the larger contents of aromatic

protein and microbial by-product-like material which is a

common characteristic of natural water that is affected by

land use changes (Chen et al., 2003). By contrast, the larger

contents of RFA and RHA demonstrate that humic and fulvic

matters are the dominated DOM for wetlands in the less-

developed region. Furthermore, the low proportion of

anthropogenic source for wetlands far from the road (S1) in-

dicates that it was not affected significantly by the drainage

from adjacent developed and pasture regions. The distance

between S1 and S5 was only about 1.5 km, thus it demon-

strates that the small, seasonal wetlands responded to the

surrounding environment within a small spatial scale.

5. Conclusions

Forty small, seasonal wetlands in the upper Piedmont and

lower Blue Ridge ecoregions of South Carolina were investi-

gated to examine the concentrations and compositions of

dissolved organic carbon (DOC) and total dissolved nitrogen

(TDN) during the wet season (FebruaryeApril) in 2011. The

characteristics of DOC and TDN were dependent on both

wetland hydrological connection and the surrounding land-

use. The isolated wetlands exhibited higher concentrations

of DOC and TDN than the connected wetlands. However, the

local, more-urbanized land use in adjacent areas can offset

this phenomenon, with possible inputs of carbon, nitrogen

and other pollutants draining fromurbanized and pastureland

areas. The DOC of wetlands impacted by human activities

suggested land-use factors as primary drivers, whereas the

wetlands in the less-developed region showed obvious DOC

characteristics belonging to NOMs. The temporal variation of

water quality primarily exhibited on the nitrogen compounds,

which had a significant increase during the study period. Mi-

crobial activity was believed to be the major contribution for

this variation but this did not significantly change the DOC

composition. Results of this work demonstrate that the water

quality of small, seasonal wetlands is affected by multiple

variables, and the small water volume of these wetlands

makes them more sensitive to land use change.

Acknowledgment

This study was supported by US EPA Region 4 Wetland Pro-

gram Development Grant, Clemson University School of

Agriculture, Forest, and Environmental Sciences, and Chinese

Postdoctoral Science Foundation (No. 2012M521607). This

material is based upon work supported by NIFA/USDA under

project number SC-1700489. Technical Contribution No. 6327

of the Clemson University Experiment Station.

Appendix A. Supplementary data

Supplementary data related to this article can be found at

http://dx.doi.org/10.1016/j.watres.2015.01.007.

r e f e r e n c e s

Bachand, P.A.M., Horne, A.J., 2000. Denitrification in constructedfree-water surface wetlands: II. Effects of vegetation andtemperature. Ecol. Eng. 14 (1e2), 17e32.

Batzer, D.P., Jackson, C.R., Mosner, M., 2000. Influences of riparianlogging on plants and invertebrates in small, depressionalwetlands of Georgia, USA. Hydrobiologia 441 (1e3), 123e132.

Bedard-Haughn, A., Matson, A.L., Pennock, D.J., 2006. Land useeffects on gross nitrogen mineralization, nitrification, and N2Oemissions in ephemeral wetlands. Soil Biol. Biochem. 38 (12),3398e3406.

Buck, O., Niyogi, D.K., Townsend, C.R., 2004. Scale-dependence ofland use effects on water quality of streams in agriculturalcatchments. Environ. Pollut. 130 (2), 287e299.

Cabezas, A., Garcia, M., Gallardo, B., Gonzalez, E., Gonzalez-Sanchis, M., Comin, F.A., 2009. The effect of anthropogenicdisturbance on the hydrochemical characteristics of riparianwetlands at the Middle Ebro River (NE Spain). Hydrobiologia617, 101e116.

Campbell, C.E., Allen, J., Lu, K.S., 2008. Modeling growth andpredicting future developed land in the upstate of SouthCarolina. In: Proceedings of the 2008 South Carolina WateResources Conference. Clemson University RestorationInstitute, Clemson. http://tigerprints.clemson.edu/cgi/viewcontent.cgi?article¼1038&context¼scwrc.

Chen, W., Westerhoff, P., Leenheer, J.A., Booksh, K., 2003.Fluorescence excitationeemission matrix regional integrationto quantify spectra for dissolved organic matter. Environ. Sci.Technol. 37 (24), 5701e5710.

Cook, B.J., Hauer, F.R., 2007. Effects of hydrologic connectivity onwater chemistry, soils, and vegetation structure and function

wat e r r e s e a r c h 7 3 ( 2 0 1 5 ) 9 8e1 0 8108

in an intermontane depressional wetland landscape.Wetlands 27 (3), 719e738.

Department of the Army, 1987. Wetlands Research ProgramTechnical 1 Report Y-87-1. Corps of Engineers WetlandsDelineation Manual (Vicksburg, Mississippi).

Ehrenfeld, J.G., Yu, S., 2012. Patterns of nitrogen mineralization inwetlands of the New Jersey Pinelands along a shallow watertable gradient. Am. Midl. Nat. 167 (2), 322e335.

Fisk, M.C., Fahey, T.J., 2001. Microbial biomass and nitrogencycling responses to fertilization and litter removal in youngnorthern hardwood forests. Biogeochemistry 53 (2), 201e223.

Fracz, A., Chow-Fraser, P., 2013. Changes in water chemistryassociated with beaver-impounded coastal marshes ofeastern Georgian Bay. Can. J. Fish. Aquat. Sci. 70 (6), 834e840.

Galbraith, L.M., Burns, C.W., 2007. Linking land-use, water bodytype and water quality in southern New Zealand. Landsc. Ecol.22 (2), 231e241.

Graves, G.A., Wan, Y.S., Fike, D.L., 2004. Water qualitycharacteristics of storm water from major land uses in SouthFlorida. J. Am. Water Resour. Assoc. 40 (6), 1405e1419.

Haidary, A., Amiri, B.J., Adamowski, J., Fohrer, N., Nakane, K.,2013. Assessing the impacts of four land use types on thewater quality of wetlands in Japan. Water Resour. Manag. 27(7), 2217e2229.

Hayashi, M., 2004. Temperature-electrical conductivity relation ofwater for environmental monitoring and geophysical datainversion. Environ. Monit. Assess. 96 (1e3), 119e128.

Hayashi, M., van der Kamp, G., 2000. Simple equations torepresent the volume-area-depth relations of shallowwetlands in small topographic depressions. J. Hydrol. 237(1e2), 74e85.

Hebb, A.J., Mortsch, L.D., Deadman, P.J., Cabrera, A.R., 2013.Modeling wetland vegetation community response to water-level change at Long Point, Ontario. J. Gt. Lakes Res. 39 (2),191e200.

Helms, J.R., Stubbins, A., Ritchie, J.D., Minor, E.C., Kieber, D.J.,Mopper, K., 2008. Absorption spectral slopes and slope ratiosas indicators of molecular weight, source, and photobleachingof chromophoric dissolved organic matter. Limnol. Oceanogr.53 (3), 955e969.

Kazezyilmaz-Alhan, C.M., Medina, M.A., Richardson, C.J., 2007. Awetland hydrology and water quality model incorporatingsurface water/groundwater interactions. Water Resour. Res.43 (4).

Maassen, S., Balla, D., Kalettka, T., Gabriel, O., 2012. Screening ofprevailing processes that drive surface water quality ofrunning waters in a cultivated wetland region of Germany e amultivariate approach. Sci. Total Environ. 438, 154e165.

McCauley, L.A., Jenkins, D.G., Quintana-Ascencio, P.F., 2013.Isolated wetland loss and degradation over two decades in anincreasingly urbanized landscape. Wetlands 33 (1), 117e127.

Miller, M.P., Simone, B.E., McKnight, D.M., Cory, R.M.,Williams, M.W., Boyer, E.W., 2010. New light on a dark subject:comment. Aquat. Sci. 72 (3), 269e275.

Min, J.H., Perkins, D.B., Jawitz, J.W., 2010. Wetland-groundwaterinteractions in subtropical depressional wetlands. Wetlands30 (5), 997e1006.

Montgomery, J.A., Eames, J.M., 2008. Prairie wolf slough wetlandsdemonstration project: a case study illustrating the need forincorporating soil and water quality assessment in wetlandrestoration planning, design and monitoring. Restor. Ecol. 16(4), 618e628.

Morrice, J.A., Danz, N.P., Regal, R.R., Kelly, J.R., Niemi, G.J.,Reavie, E.D., Hollenhorst, T., Axler, R.P., Trebitz, A.S.,Cotter, A.M., Peterson, G.S., 2008. Human influences on waterquality in Great Lakes coastal wetlands. Environ. Manage 41(3), 347e357.

Murphy, K.R., Butler, K.D., Spencer, R.G.M., Stedmon, C.A.,Boehme, J.R., Aiken, G.R., 2010. Measurement of dissolvedorganic matter fluorescence in aquatic environments: anInterlaboratory comparison. Environ. Sci. Technol. 44 (24),9405e9412.

Panno, S.V., Nuzzo, V.A., Cartwright, K., Hensel, B.R., Krapac, I.G.,1999. Impact of urban development on the chemicalcomposition of ground water in a fen-wetland complex.Wetlands 19 (1), 236e245.

Pitt, A.L., Baldwin, R.F., Lipscomb, D.J., Brown, B.L., Hawley, J.E.,Allard-Keese, C.M., Leonard, P.B., 2012. The missing wetlands:using local ecological knowledge to find cryptic ecosystems.Biodivers. Conserv. 21 (1), 51e63.

Reiss, K.C., 2006. Florida wetland condition index for depressionalforested wetlands. Ecol. Indic. 6 (2), 337e352.

Seilheimer, T.S., Chow-Fraser, P., 2006. Development and use ofthe Wetland Fish Index to assess the quality of coastalwetlands in the Laurentian Great Lakes. Can. J. Fish. Aquat.Sci. 63 (2), 354e366.

Stokstad, E., 2006. Water quality e high court asks Army Corps tomeasure value of wetlands. Science 312 (5782), 1870e1870.

Trebitz, A.S., Brazner, J.C., Cotter, A.M., Knuth, M.L., Morrice, J.A.,Peterson, G.S., Sierszen, M.E., Thompson, J.A., Kelly, J.R., 2007.Water quality in great lakes coastal wetlands: basin-widepatterns and responses to an anthropogenic disturbancegradient. J. Gt. Lakes Res. 33, 67e85.

Tu, J., 2011. Spatial and temporal relationships between waterquality and land use in northern Georgia, USA. J. Integr.Environ. Sci. 8 (3), 151e170.

Verhoeven, J.T.A., Arheimer, B., Yin, C.Q., Hefting, M.M., 2006.Regional and global concerns over wetlands and water quality.Trends Ecol. Evol. 21 (2), 96e103.

Weishaar, J.L., Aiken, G.R., Bergamaschi, B.A., Fram, M.S., Fujii, R.,Mopper, K., 2003. Evaluation of specific ultraviolet absorbanceas an indicator of the chemical composition and reactivity ofdissolved organic carbon. Environ. Sci. Technol. 37 (20),4702e4708.

Whigham, D.F., Jordan, T.E., 2003. Isolated wetlands and waterquality. Wetlands 23 (3), 541e549.

Whitmire, S.L., Hamilton, S.K., 2005. Rapid removal of nitrate andsulfate in freshwater wetland sediments. J. Environ. Qual. 34(6), 2062e2071.

Zhou, J., Wang, J.J., Baudon, A., Chow, A.T., 2013. Improvedfluorescence excitation-emission matrix regional integrationto quantify spectra for Fluorescent dissolved organic matter. J.Environ. Qual. 42 (3), 925e930.