Surface Water Quality and Landscape Gradients in the North Carolina Cape Fear River Basin: the Key Role of Fecal Coliform

Surface Water Quality and Landscape Gradients in the North Carolina Cape Fear River Basin: The Key Role of Fecal Coliform

Jennifer B. Alford, Keith G. Debbage, Michael A. Mallin, Zhi-Jun Liu

Southeastern Geographer, Volume 56, Number 4, Winter 2016, pp. 428-453 (Article)

Published by The University of North Carolina Press DOI: https://doi.org/10.1353/sgo.2016.0045

For additional information about this article https://muse.jhu.edu/article/641170

Access provided by Elon University (13 Jan 2017 15:22 GMT) Surface Water Quality and Landscape Gradients in the North Carolina Cape Fear River Basin The Key Role of Fecal Coliform

Jennifer B. Alford Elon University Salem College Keith G. Debbage University of North Carolina at Greensboro Michael A. Mallin Center for Marine Science, University of North Carolina Wilmington Zhi-Jun Liu University of North Carolina at Greensboro

Although relationships between specific land Aunque las relaciones entre los tipos específicos types and water quality have been well estab- de tierra y la calidad del agua han sido bien es- lished, studies that observe these relationships tablecidos, los estudios que observan estas rela- across large heterogeneous landscapes are less ciones a través de grandes paisajes he terogéneos common. This paper seeks to add to the growing son menos comunes. En este trabajo se pretende body of literature related to how various land añadir al creciente cuerpo de literatura rela- use configurations might influence the geog- cionada con la forma en que varias configura- raphy of surface water quality across a large, ciones de uso de la tierra pueden influir en la heterogeneous river basin network like the Cape geografía de la calidad del agua de superficie a Fear River Basin (CFRB) in North Carolina. través de una red de cuencas hidrográficas am- Findings suggest that fecal coliform bacteria plia y heterogénea como la cuenca del río Cape concentrations that exceeded state standards Fear (CFRB) en Carolina del Norte. Los resulta- were positively associated with transitional dos sugieren que las concentraciones de bacte- land types (i.e. mixed forest and low density rias coliformes fecales que excedieron las nor- developed–open space (< 20 percent impervi- mas estatales se asociaron positiv amente con ous surfaces)), especially when these land types tipos de tierra de transición (es decir, bosques encircled urbanization in the upper reaches mixtos y baja densidad desarrollado–espacio of the river basin in both 2001 and 2006. As abierto (<20 por ciento de las superficies im- a result, it is argued that the transition from permeables)), sobre todo cuando estos tipos mixed forest and low density developed–open de tierra rodeaban urbanización en la parte space to a more urban environment provides alta de la cuenca del río. Como resultado, se a series of well recognized human activities argumenta que la transición de bosque mixto within the landscape gradient that can increase y baja densidad desarrollado–espacio abierto fecal coliform counts. a un entorno más urbano ofrece una serie de southeastern geographer, 56(4) 2016: pp. 428–453 NC Landscape Gradients and Fecal Coliform 429 actividades humanas bien reconocidos dentro Tu 2011, Crim et al. 2012, Huang et al. del gradiente de paisaje que se puede aumentar 2013). However, these studies typically el recuento de coliformes fecales. considered relationships at the local watershed scale or by observing a specific key words: Water quality, land use, fecal portion of a river basin. Although various coliform, watersheds, river basins federal and state agencies have conducted palabras clave: calidad del agua, usos extensive studies related to water quality de la tierra, coliformes fecales, las cuencas and land types, spatial analysis of larger hidrográficas, cuencas fluviales river basins is less common in the geography literature, because of the large spatial context and complexity of these heterogeneous landscapes. Analysis of the spatial introduction variation of surface water quality within There are over 3.6 million miles of a river basin context is critical if we are rivers and streams in the United States, to better understand how different land- each exhibiting unique characteristics scape features and activities in different that are physically, biologically, and portions of the basin impact the geospatial chemically influenced by the diverse characteristics of surface water resources landscapes they traverse (US EPA 2010). throughout the entire river basin network. Much of the lit erature has focused on the Bacterial pathogens (such as fecal coli- inter-relationships between various land form) in surface water systems historically use configurations and surface water qual- have been linked to agricultural and de- ity have been grounded in, but not limited velopment activities and the presence of to, geography, hydrology, aquatic biology, specific landscape eatures.f These features and environmental planning (Meybeck and activities may include increases in im- and Helmer 1989, Arnold and Gibbons pervious surfaces, high livestock density, 1996, Booth and Jackson 1997, Liu et al. and failing wastewater treatment plant in- 2000, Tong and Chen 2002, Rothenberger frastructure. Relationships between urban et al. 2009, Tu 2011). Although there is and suburban development and non-point a growing literature that highlights rela- sources of fecal concentrations in surface tionships between specific land types and waters have also been well documented. surface water quality, the U.S. Environ- Mallin et al. (2000) and Holland et al. mental Protection Agency (EPA) estimates (2004) observed that fecal coliform pol- that only 19 percent of streams and rivers lution in tidal creeks in North and South have been fully assessed by federal and Carolina was positively correlated with state agencies (US EPA 2013). watershed population and percent of de- Some studies have observed signifi- veloped land, and particularly correlated cant relationships between different land with percent impervious surface coverage. types and water quality across different In coastal watersheds of South Carolina, geog raphical scales (Aspinall and Pearson Kelsey et al. (2004) observed high fecal 2000, Tong and Chen 2002, Kelsey et al. concentrations in populated areas that 2004, Schoonover and Lockaby 2006, were likely contributed by the high con- Tu et al. 2007, Wilson and Weng 2010, centrations of domestic pets (i.e. dogs and 430 alford et al. cats) associated with human populations. of CAFOs and related activities can trig- Government agencies have also played ger a variety of landscape changes that a vital role in identifying the spatial pat- may result in habitat fragmentation and terns and sources of fecal coliform bacte- the reduction of continuous forestland. ria in surface water systems. The USGS Some of the adverse effects of these (2003) observed water quality trends practices include increases of pathogens across watersheds in Virginia primarily and nutrients entering groundwater sup- representative of urban, agricultural, and plies and surface waters both adjacent to mixed land types. They found that fecal and downstream of CAFO sites (Mallin sources were highly variable and key con- et al. 2015). This process can lead to the tributors included human, wildlife, and eutrophication of water bodies, often domesticated pet waste as well as various resulting in hypoxic conditions (i.e. low land use activities. In addition to non- dissolved oxygen) that alter both the point sources of fecal microbes, urban and quality and quantity of surface water suburban watersheds frequently contain resources and aquatic habitats across point sources of fecal inputs that are typ- multiple geographical scales (Mallin and ically associated with wastewater treat- Cahoon 2003, Burkholder et al. 2007). ment spills and leaks that contribute sig- Another consideration when observing nificantly to unpredictable fecal “spikes” the spatial context of surface water quality in surface water systems. across a large spatial scale, such as a river Across rural landscapes, sources of basin, are the variations and or changes fecal pollution may be highly variable on the landscape that result in different and are typically related to human activ- and often complex landscape patterns or ities. Potential sources of fecal microbes gradients. Landscape gradients may exist may result from the presence of livestock due to natural features in the landscape on the landscape, spraying livestock such as geological, climatic, and topo- waste onto fields, and failing septic sys- graphical patterns as well as anthropocen- tems or high density of septic systems. tric changes to the landscape that may be Changes in agricultural practices may driven by socioeconomic decisions. As a re- also increase fecal concentrations in sur- sult, gradients are far likelier to occur than face waters. Over the past few decades, sharp land use boundaries. These sorts of the concentrated animal feeding oper- land use gradients may include forest land ation (CAFO) industry has emerged as being converted to agricultural or urban a common practice in the United States areas or as more recent trends suggest, ag- and Europe as pasture livestock produc- ricultural land being converted to urban tion has moved from the field to large areas (Wear et al. 1998, Schoonover et al. building facilities (Mallin and Cahoon 2005, Grimm et al. 2008). Grimm et al. 2003, Rothenberger et al. 2009, Mallin (2008) note that when considering urban et al. 2015). CAFO activities on the land- form, various types of landscape gradi- scape include livestock holding facilities, ents radiate from urban centers from both related waste lagoons, and spray fields small and large cities, often resulting in that are often located adjacent to the landscape gradients at both the local and facilities. As a result, the development regional scales. NC Landscape Gradients and Fecal Coliform 431

The literature has also suggested that on a specific portion of a river basin. We transitional landscapes, including less will argue that fecal coliform is one of the developed areas surrounding the urban most volatile water quality indicators in core, may experience changes in stream the CFRB, and one that most frequently water quality as a direct result of an in- exceeds either state or federal water qual- crease in human disturbances to the ity standards and, therefore, merits addi- landscape. Transitional landscapes may tional attention. Second, it is hypothesized have different landscape configurations that the geography of fecal coliform across and transitional patterns such as urban to the CFRB is partially explained by certain suburban, rural to suburban, and forest unique land use gradients. In particular, we to rural and or suburban land-use types. will suggest that high fecal coliform counts During landscape transitions, especially are more likely in areas with both mixed from forestland or rural land to suburban, forest and low density developed–open the percentage of impervious surface and space (i.e. < 20 percent impervious sur- related stormwater infrastructure may face; includes large single-family lot-sizes, increase. This is typically in the form of parks, golf courses) land types especially residential and commercial development, in the more urbanized Upper Cape Fear an increase in spatially dispersed road net- physiographic region. It is argued that the works, or changes in agricultural practices. transition from mixed forest to low density Several studies have noted adverse impacts developed–open space to a more urban to stream quality in watersheds with im- environment provides a ser ies of well pervious surfaces as low as 5 to 20 percent recognized human activities within the (Schueler 1994, Booth and Jackson 1997, landscape gradient that can increase fecal Mallin et al. 2000, Schoonover et al. 2005, coliform counts. As a result, the primary Walsh et al. 2005 and others). Understand- objectives of this study are to understand ing the spatial distribution of land types, (1) relationships between specific land resulting patterns and changes throughout types and surface water q uality across the a river basin is important in the analysis of entire river basin network, (2) the extent to water quality data, because it helps us to which these relationships vary across the better understand how landscape activities three physiographic regions of the study in one reach of the basin may be impacting area and, (3) how changes in land types water quality throughout the entire river impact water quality during two distinct basin network. time periods. It is anticipated that observing relationships between land types and Study Objectives water quality across a large hydrological This paper adds to the growing body of network, and related heterogeneous land- literature related to how various land use scape, will assist in illus trating the extent configurations might influence the geogra- of these relationships across the entire phy of surface water quality across a large, river basin network. By implementing mul- heterogeneous river basin network like tiple assessment methods, this study may the North Carolina Cape Fear River Basin establish a framework for future research (CFRB). Most prior studies have largely endeavors that seek to address the ex- focused on the local watershed scale or tent to which specific land types may be 432 alford et al. impacting water resources over large spa- pollution inputs to its surface waters tial extents. This will assist resource agen- (NC DENR 2012). Water quality moni- cies and decision makers in developing toring stations are located throughout more comprehensive plans that address the river basin’s surface water network how socio-economic decisions and related (Figure 1). According to NC DENR (2012), human activities impact the landscape and there are over 480 km of impaired stream surface water resources across multiple ge- segments in the CFRB, meaning that ographical scale within a single river basin. stream water quality does not meet standards (e.g. recreation, drinking water sup- ply) set by the state regulatory agency. methodology From 2000 to 2010, the basin’s popula- Study Area tion increased 24.3 percent with approx- The Cape Fear River Basin (CFRB) imately 195 people per square mile and it is one of North Carolina’s largest river is expected to increase from 3 million to basins (23,734 km2), and it includes all about 5 million people over the next 20 or portions of 26 counties that contains years in largely urban or urbanizing areas approximately 25 percent of the state’s (NC DENR 2012). population and 17 percent of the state’s land area (NC DENR EE 2012). The Cape Land Use/Land Cover (LULC) Fear River itself is approximately 320 km Assessment long, originating in Chatham County, at In order to classify land use types and the confluence of the Haw and Deep Riv- patterns across the CFRB, National Land ers (NC DENR 2005, NC DENR 2012). The Cover Database (NLCD) imageries for basin displays a southeastern orientation 2001 and 2006 were downloaded from and is divided into three distinct physio- the Multi-Resolution Land Characteristics graphic regions: the Upper (i.e. Piedmont Consortium database website. At the time region), Middle (i.e. Sandhills region), of this study, this was the most recent and Lower CFRB (i.e. Coastal region) classified land cover imagery available each with distinct soil types and land- from the MRLC program. Observing the scape characteristics. As a result, the basin metadata related to the 2001 and 2006 encompasses different aquatic ecosys- NLCD imagery, one will note several tems, including wetlands, backwater sys- temporal differences across the physio tems, tidal creeks and other estuaries that graphic regions. The NLCD imagery repre- provide wildlife habitat for over 30 feder- senting “2001” for the Middle and Lower ally listed endangered species (NC DENR reaches of the basin was taken on May EE 2012). The CFRB basin also provides 11, 2000 and for “2006” the imagery was water resources for residential, commer- taken on April 21, 2007. For the Upper cial, and industrial uses (NC DENR 2005). reaches of the basin, the “2001” imagery In addition to being the most populated was taken on October 7, 1999 and for river basin in North Carolina, the CFRB is “2006”, the imagery was taken on October also the most industralized, with 203 per- 15, 2005. For the purposes of this study, mitted municipal and industrial wastewa- the imagery dates will be referred to 2001 ter dischargers contributing point source and 2006 to be consistent with other Figure 1. Cape Fear River Physiographic Regions and Water Quality Monitoring Stations. Source: Inset map: USGS 2016. 434 alford et al. studies that have applied this dataset (the Lower Cape Fear River Program, the (MRLC 2012). In an effort to represent the Cape Fear Middle Basin Coalition and the amount of each land type within the basin, Upper Cape Fear River Coalition) for var- percentages of each LULC were calculated ious biological, chemical, and physical using ArcGIS 10 and converted into per- water quality parameters. Field samples cent (km²) for each land type. are collected by either University labora- tories or consulting firms contracted by Watershed Delineation the coalitions. Each water sample taken Using Geographic Information Sys- adheres to laboratory sampling and anal- tem (GIS) methods similar to Rothen- ysis techniques certified by the NC DENR berger et al. (2009) and Merwade (2012), Division of Water Resources (DWR). Data watersheds within each hydrologic unit for stations throughout the three physio- that contain water quality monitoring sta- graphic regions of the study area were not tions were identified. These monitoring available until October 2000. As a result stations served as surface water drain- of this and the limitation previously noted age points and the landscape upstream for the MRLCs NLCD LULC data, this study of these stations were delineated using analyzed water quality data from October ArcGIS Hydro tools in addition to digital 2000 to October 2001 and October 2006 elevation models (DEMs) (30m) provided to October 2007. The monthly samples at by the US Department of Agriculture's each station were aggregated to construct Geospatial Data Gateway (USDA 2012). annual averages for each water quality pa- Once these drainage patterns were identi- rameter at all 72 stations observed in this fied, we imported the NLCD land cover im- study. Although monthly samples and an- agery into the ArcGIS system to calculate nual averages do not always capture abrupt the percent of a given LULC type within daily changes in water quality, this dataset each watershed for each of the 72 water is the only established monitoring program quality monitoring stations included in that provides state-certified water quality this study. This will assist in understanding data for the entire river basin. Table 1 illus- and illustrating surface water system flow trates the sources for and official govern- patterns, the percent of a given LULC type ment standards, where available, for the within a single watershed, and changes in water quality parameters analyzed in this land types from 2001 and 2006. study. Ambient standards are available for dissolved oxygen (DO) and fecal coliform Water Quality Analysis and bacteria, but no official ambient standards Ancillary Data Sources currently exist for nitrite-nitrate nitrogen

Water quality data were downloaded (NO2-NO3), ammonium nitrogen (NH3-N) from the Cape Fear River Basin Coalition’s and phosphorus (P)) although several sur water quality database that was initially rogate measures were utilized in this established by the NC DENR (Department paper. The NC DENR standards were used of Environment and Natural Resources). to determine if a water quality sample The water quality stations in the CFRB cho- was considered impaired for its specified sen for this analysis are monitored monthly or designated use (e.g. drinking water, by State-approved discharger coalitions recreational activities). Table 1. Sources and NC DENR Standards/EPA Criteria for Water Quality Parameters for the Cape Fear River Basin.

Water Quality NC DENR US EPA Potential Sources and Parameter Standard Criteria Seasonal Differences

Fecal coliform Maximum Currently working with state Human sources may include wastewater treatment plant 400 col/100ml to develop new criteria discharges, failing septic systems. Animal sources may for all streams based on more specific fecal include domestic pets, livestock, animal operations, and except SA HQW characteristics. wildlife. (43 col/100ml). Dissolved Oxygen (DO) Minimum 4 mg/L In most stream types, levels Increases with an increase in a streams contact with the except SA HQW below 5.0 mg/L cause atmosphere, high stream flow events, low temperature and SC (minimum stressful conditions for or produced by plants during photosynthetic processes. 5 mg/L) aquatic ecosystems. Decreases may result from an increase in nutrients, high temperature, urban and agricultural runoff, land clearing, untreated sewage, and salinity. Nitrite-Nitrate Nitrogen No ambient No ambient standard Anthropocentric sources include wastewater treatment plants,

(NO2-NO3) standard runoff from fertilized lawns and cropland, failing septic tanks, runoff from animal manure storage areas, and industrial discharge. Ammonium Nitrogen No ambient No ambient standard Natural sources may come from soil and rocks, the

(NH3-N) standard atmosphere, tissues of living and dead organisms. Phosphorus (P) No ambient No ambient standard Phosphorus is typically scarce in water under natural standard conditions. Seasonal differences may be related to precipitation events and/or crop rotation schedules, which may dictate when fertilizers are applied.

Sources: US EPA (2013), Cape Fear River Basin Coalitions Water Quality Monitoring Data (2012). 436 alford et al.

To account for the impact of precipita- exists throughout the river basin, a coded tion on water quality, we included a contr ol variable represented the Upper, Middle, variable total precipitation (centimeters and Lower CFRB physiographic regions. per year) derived from the weather station located closest to each water quality moni- Statistical Analysis toring station; de tailed surface water flow Based on prior studies and available data were not consistently available across data (Schueler 1994, Arnold and Gibbons the entire river basin. This may assis t in 1996, Mallin et al. 2000, Mallin and Ca- observing how surface runoff from specific hoon 2003, Potter et al. 2004, Schoonover land-use types may be impacting surface et al. 2005, Carle et al. 2005, Burkholder water quality. An additional control var- et al. 2007, Brabec 2009, Cookson and iable livestock head by permit was also Schorn 2009, Rothenberger et al. 2009, included to account for the potential im- Tu 2011), the dependent and independent pact of the presence of livestock located variables list in Table 2 will be analyzed in at CAFOs present during the study pe- an effort to understand and spatially illus- riod. Monthly total precipitation data was trate relationships between water quality downloaded from the National Oceanic and LULC types throughout the Cape Fear and Atmospheric Administration (NOAA) River Basin from 2001 to 2006. National Climatic Data Center while live- Using SPSS 22.0, a step-wise linear stock (i.e. cattle and swine) head counts for regression analysis was designed to iden- state and federally permitted operations, tify the most powerful predictors of sur face including CAFOs, were obtained from NC water quality across the CFRB from 2001 DENR DWR. The livestock data indicated to 2006. Since water quality data can be the maximum number of livestock head highly variable due to both human activ- allowed annually at a given CAFO site that ities and natural climatic conditions, a is permitted by the NC DENR. However, natural log transformation was applied to some facilities were not included in this the annual averages of each water quality analyses. First, the cattle and swine facili- parameter for each of the 72 water quality ties were not regulated until 1993. Second, monitoring stations under investigation. poultry CAFO locations and head counts Coefficients of the log transformation are not provided by the state of North were multiplied by 100 to determine the Carolina agencies to the public (Mallin percent change corresponding to a one and Cahoon 2003). In an effort to identify unit change in the independent variable potential impacts to surface water quality had on each water quality parameter’s from point sources of pollution, National concentration levels as described by the Pollution Elimination Discharge System USGS (2015) and Yuncong and Migliaccio (NPDES) permits and NC DENR Sanitation (2011). Since annual averages do not cap- and Sewer Overflow (SSO) reports were ture episodic events, each monthly sample also reviewed. The SSO reports indicated was also observed to understand monthly the specific location and amount of pol- and seasonal changes in each parameter. lution occurring during an infrastructure Based on several multicollinearity (MC) failure or spill. Finally, to determine if po- tests no serious MC problems were found tential regional difference in water quality in the selected models. We also tested the NC Landscape Gradients and Fecal Coliform 437

Table 2. Dependent and independent variables of interest.

Dependent Variables Independent Variables

Annual Average Dissolved Oxygen Percent Land-Use/Land-Cover Type (km2) Annual Average Mean fecal coliform Number of Permitted Livestock Head by Permit (CAFOs permitted by NC DENR)

Annual Average Ammonium Nitrogen (NH3-N) Total Precipitation Annual Average Nitrate-Nitrite Nitrogen Type of Physiographic Region

(NO2-NO3) Annual Average Phosphorus (P) potential for MC for all of the dependent basis. In regards to DO, this was expected variables (see Table 2) and did not observe because aggregating the monthly samples MC issues for any of the regression mod- into annual averages means that the typ- els. In addressing potential spatial auto- ically lower DO readings of the summer correlation in the water quality data due to months were masked by the seasonally water quality monitoring station locations higher DO levels in cooler months. Ob- within the river basin network, it should serving the monthly samples for each of be noted that a large majority of the sta- the 72 stations during the warmer months tions are mutually exclusive and in the few (i.e. June to September) revealed that 9 cases where the stations are nested (i.e. stations had monthly samples that failed located in the same watershed), they were the NC DENR standard in 2001 and 13 largely separated by substantive physical stations failed the standard in 2006, with distances (on average, roughly 30 miles) a majority of these monitoring stations so that the influence of an upstream sta- being located in the Lower CFRB for both tion on a downstream location was consid- years. Although, the annual average for ered negligible in most cases nitrite-nitrate nitrogen (NO2-NO3) was relatively unchanged over time, the standard deviation was slightly higher in 2006 results and discussion suggesting some variability. This may be Geospatial Characteristics of related, in part, to an increase in human Surface Water Quality activities on the landscape that resulted

The descriptive statistics for the water in elevated NO2-NO3 inputs to surface wa- quality parameters (Tables 3 and 4) suggest ters of the CFRB. Although the NC DENR that several water quality metrics were has not established nutrient standards, relatively stable with small variances by using data from 1,366 streams Dodds both location or over time. These included et al. (1998) concluded that total phos- dissolved oxygen (DO), ammonium nitro- phorus (TP) concentrations > 0.075 mg/L gen (NH3-N), and phosphorus (P). Each of were characteristic of eutrophic streams. these parameters generated relatively low It should be noted that the average TP standard deviations with annual averages during both study periods well exceeded that changed little from 2001 to 2006 and this threshold (see Tables 3 and 4). Thus, there was little variability on a monthly overall P concentrations in the CFRB can 438 alford et al.

Table 3. CFRB water quality descriptive statistics of 72 monitoring stations in 2001.

Number of Stations with Annual Averages Water Quality Standard Exceeding NC DENR Parameters Minimum Maximum Mean Deviation Standards

Fecal coliform 18 3,618 415 707 19 (26%) (col/100ml) DO (mg/L) 4.17 11.02 8.08 1.13 0

NO2-NO3-N 0.05 9.22 1.27 1.87 No ambient standard (mg/L)

NH3-N (mg/L) 0.03 0.82 0.12 0.13 No ambient standard P (mg/L) 0.03 2.14 0.27 0.30 No ambient standard

Table 4. CFRB water quality descriptive statistics of 72 monitoring stations for 2006.

Number of Stations with Annual Averages Water Quality Standard Exceeding NC DENR Parameters Minimum Maximum Mean Deviation Standards

Fecal coliform 24 1,472 318 327 23 (32%) (col/100ml) DO (mg/L) 5.03 10.65 8.06 1.15 0

NO2-NO3-N 0.06 12.54 1.32 2.29 No ambient standard (mg/L)

NH3-N (mg/L) 0.03 0.32 0.07 0.044 No ambient standard P (mg/L) 0.03 1.72 0.19 0.25 No ambient standard

be considered high, although there was lit- However, it is the fecal coliform met- tle variability over time. Regarding nitrate ric that demonstrated the largest absolute and ammonium, research in the lower change from 2001 to 2006. In 2001, the an- CFRB in blackwater streams determined nual average of fecal coliform counts was that concentrations of nitrate or ammo- 415 colony-forming units (CFU)/100ml nium > 0.50 mg/L significantly stimulated with a relatively high standard deviation chlorophyll a and BOD in these systems (i.e. 707) compared to an annual average of (Mallin et al. 2004). Ammonium nitrogen 318 CFU/100ml in 2006 and a much lower

(NH3-N) for stations in this study were rel- standard deviation (i.e. 327). Perhaps more atively stable, although monitoring station importantly, the fecal coliform annual aver- monthly samples and annual averages fre- age exceeded the NC DENR standard at one quently exceeded >0.50 mg/L. in four water quality monitoring stations in NC Landscape Gradients and Fecal Coliform 439

2001, and one in three stations in 2006 even that even for a single physiographic though the overall average declined over region, like the Upper CFRB, the explana- time. In addition, monthly samples were tion for high fecal coliform concentrations highly variable (e.g. 200 CFU/100ml to in surface water systems can be highly 12,000 CFU/100ml) with drastic changes variable and may derive from both point at stations located throughout the basin. source and non-point sources, highlight- The spatial distribution of stations with an- ing the spatial complexity of this water nual averages that exceeded state fecal col- quality metric. Furthermore, it is possible iform standards in the CFRB are illustrated that the types of land use arrangements in Figures 2 and 3 for 2001 and 2006. The in each watershed may be key predictors 19 stations (26 percent of the total) that of the variability in fecal coliform concen- exceeded the state standard for fecal coli- trations observed in this study. Due to the form in 2001 are located exclusively in the significant oler that fecal coliform plays Upper and Middle regions of the CFRB (see in shaping water quality throughout the Figure 2). A majority of the stations that CFRB network, at least in terms of exceed- exceeded the state standard for fecal are lo- ing NC DENR standards, we now turn to cated in urbanizing watersheds in the Upper a regression analysis to determine which CFRB which were characterized by low in- land types best explain the spatial varia- tensity developed–open space surrounded tion of this important water quality metric. by both agricultural (i.e. pasture/hay) and mixed forest landscapes. Some of these sta- Fecal Coliform Across the tions are located in watersheds that include Cape Fear River Basin 2001: portions of highly urbanized cities like The Key Predictors Greensboro, Durham, and Burlington. The Although this regression analysis f ocuses remaining monitoring stations were largely primarily on the land use characteristics of located in rural watersheds characterized the CFRB it should be noted that two control by a mix of land use types including mixed variables—total precipitation and livestock forest, low intensity de veloped–open space, headcount—were also included in this and agricultural land in Montgomery, Ran- analysis as well as a coded variable that rep- dolph, and Chatham counties. Although resented the three physiographic regions of these stations are located in watersheds the CFRB (i.e. Upper, Middle, and Lower). with different landscape characteristics, The step-wise regression analysis for fecal they all contained transitional land types coliform in 2001 revealed that four predic- (i.e. low intensity developed–open space tor variables were key factors in shaping the and mixed forest) where human activities spatial distribution of fecal coliform con- on the landscape appear to be increasingly centrations across the river basin (Table 5). altering the landscape that surrounds The R-squared value of the model was 0.48 urban centers. Broadly similar trends seem (p < 0.05) suggesting that 48 percent of the to occur in 2006 (see Figure 3) with a few variation in fecal coliform can be e xplained exceptions. by location of the monitoring station in the Given the wide variety of landscapes Upper CFRB, the percentage of the land use characteristics of watersheds with high in the watershed is open space or mixed for- fecal coliform concentrations, it appears est classification, and total precipitation. Figure 2. Cape Fear River Basin Stations with Annual Averages Exceeding Fecal Coliform Standards in 2001. Figure 3. Cape Fear River Basin Stations with Annual Averages Exceeding Fecal Coliform Standards in 2006. 442 alford et al.

Table 5. Regression model summary of fecal coliform across the Cape Fear River Basin for 2001. The final regression model for fecal coliform from 2001 can be formally expressed as follows: Log FC (2001) = 6.56 + 0.74 UCFRB + 0.05 DOS + 0.16 MF – 0.07 PT where UCFRB = Upper Cape Fear River Basin, DOS = Percent Developed, Open Space, MF = Percent Mixed Forest, PT = Total Precipitation.

Independent Unstandardized Variables b Coefficient Exponentalue V Percent Change

Constant 6.56 UCFRB Region 0.74 2.10 110% Percent Developed, 0.05 1.05 5% Open Space Percent Mixed Forest 0.16 1.18 18% Total Precipitation 2001 −0.07 0.93 −7% All p-values <0.05 R2 = 0.48

The regression analysis suggests that were located in watersheds primarily char- the relationship between fecal coliform acterized by both urban and rural land concentrations and the Upper CFRB use types. However, one common simi- region in 2001 was positive. Compared to larity among these watersheds was that the other regions of the study area, fecal they all contained transitional land types coliform concentrations at s tations locat ed primarily represented by mixed forest and in the Upper CFRB were 110 percent low intensity developed–open space. higher. The Upper CFRB region repre- When considering associations be- sented the most highly urbanized region tween specific land types and fecal concen- in the CFRB with development largely trations, the model predicted that as the concentrated in and around the cities percentage of low intensity developed– of Greensboro, High Point, James town open space (i.e. < 20 percent impervious (Guilford County), Burlington (Alamance surfaces) increased by one percentage County), Durham (Dur ham and Orange point, fecal coliform concentrations would Counties), and Chapel Hill (Orange increase by 5 percent, holding all other County). Furthermore, several of the predictor variables constant. In 2001, low water quality monitoring stations that intensity developed–open space accounted experienced “spikes” in fecal coliform for only 3 percent of all land classifica- concentrations were located in the Upper tions in the CFRB. Despite representing a CFRB region. These stations not only small percentage of the land types present experienced drastic changes in the annual across the entire CFRB, this type of land average of fecal coliform concentrations use appears to be significantly linked to but also experienced extremely elevated increases in fecal concentrations through- fecal coliform concentrations from month out the river basin. By contrast, denser to month. Stations experiencing these forms of development did not feature in extreme changes in fecal concentrations the final model, possibly because this type NC Landscape Gradients and Fecal Coliform 443

Figure 4. Suburban development on the fringe of mixed forest land in the Cape Fear River Basin, NC in 2015. Source: Mallin, 2015. of development has well-de veloped water In addition to the percentage of land and sewer systems that are frequently mon- allocated to developed–open space by itored as well as more compact wastewater watershed, the percent of land classified system infrastructure (i.e. pipes) that may as mixed forest played a significant role not traverse a large spatial extent. In addi- in shaping the geography of fecal coliform tion, once a dense urban center is devel- across the CFRB. The relationship between oped, large changes to the landscape may percent mixed forest and fecal concen- be limited. It is also possible that much of trations in 2001 was positive suggesting the developed–open space land lies along that as the percentage of mixed forest more rural landscape gradients where reg- land increased by one percentage point, ulatory oversight is laxer and or the waste- fecal coliform concentrations increased water infrastructure traverses a larger spa- by 18 percent, holding all other predic- tial extent that may include crossing more tor variables fixed.Alt hough the MRLC surface water systems. Rural areas are also classifies this land type as 20 percent more commonly associated with septic tree coverage in addition to other vegeta- systems in the absence of WWTPs. These tive cover, this study observed it in close systems have been linked to surface and proximity to land types associated with subsurface fecal inputs to surface waters human activities. Across the CFRB, mixed as observed by Cahoon et al. (2006). forest was typically identified adjacent to 444 alford et al.

Figure 5. CAFO facilities, spray fields, and waste lagoons within mixed forest land types located in the CFRB, Duplin County, NC in 2013. Source: Mallin, 2013.

developed–open space, especially in the activities have been linked to increases in upper and middle basin, and was primarily nutrients and pathogens, including fecal adjacent to agricultural land types and in coliform bacteria, in surface water sys- very close proximity to CAFOs in the lower tems across the nation including the CFRB basin. Figure 4 is one e xample of the land- (Mallin et al. 2000, Mallin and Cahoon scape transition from mixed forest to low in- 2003, Ahearn et al. 2005, Bur kholder et al. tensity developed–open space as observed 2007, Rothenberger et al. 2009). Although in the middle and upper basin. In contrast, mixed forest only accounted for 2 percent Figure 5 is an example of how CAFOs are of the CFRB’s total landscape in 2001, the nested within mixed forest landscapes in model suggested that even a small percent- the lower portion of the basin. Given these age increase of this land type can contrib- observations, it is anticipated that activ- ute to significant increases in fecal coliform ities related to mixed forest may include, concentrations in surface water systems but are not limited to, wastewater treat- throughout the basin. Furthermore, the ment plant (WWTP) spills, sep tic system regression analysis indicates that percent failures, and CAFO activities. All of these mixed forest may contribute to higher NC Landscape Gradients and Fecal Coliform 445 concentrations of fecal when compared or not monitoring stations were located to percent de veloped–open space (i.e. in the Upper CFRB. One important differ- 18 percent increase versus 5 percent in- ence in the two models is that total precip- crease), suggesting that a variety of human itation was replaced by the percent scrub/ activities may be contributing to higher shrub land classification as a key predictor concentrations of fecal coliform concen- variable in 2006. Overall, the R-squared trations in all three of the basins’ physio- value of the model was 0.31 (p-value graphic regions. < 0.05) suggesting that 31 percent of the The final variable to enter the 2001 spatial variation in fecal coliform concen- regression analysis for fecal coliform trations can be explained by percent mixed was total precipitation. The b coefficient forest, developed–open space, scrub/ suggested an inverse relationship existed shrub land, and being locat ed in the Upper indicating that as the total precipitation CFRB region (Table 6). The first predictor increased by one inch, fecal concen- variable to enter into the 2006 regres- trations will decrease by seven percent sion model was percent mixed forest. The across the river basin. The literature lacks model predicted that as the percent mixed consensus regarding the relationships forest land increased by one percentage that exist between precipitation patterns point, fecal coliform concentrations across and fecal concentrations in surface water the basin would increase by 14 percent. systems (Potter et al. 2004, Cahoon et al. This is a slightly lower percent increase 2006, Hill et al. 2006). However, sewage in fecal concentrations when compared leaks and spills, as well as septic system to the influence of percent mixed forest failures occur regardless of rain events in the 2001 regression model (18 percent (Tavares et al. 2008). The large month-to- increase). As previously noted, mixed month changes seen at some sites would forest is defined by the MRLC as areas support failures in infrastructure as prob- dominated by mixed vegetation. Simi- lematic. The duration of rainfall as well lar to 2001, in 2006 this study observed as the duration of time that passes prior this land type surrounding low intensity to a water quality sample being collected developed–open space especially in the are also important factors when consid- Upper and Middle basins indicating that ering the spatial relationships that might it may become more disturbed by human exist between rainfall totals and fecal activities over time. In contrast, this land concentrations in surface water systems. type was also located adjacent to CAFO activities including waste lagoons and spray Fecal Coliform Across the fields surrounded by forest. Given these Cape Fear River Basin in 2006: landscape configurations, the runoff from The Key Predictors these activities may be increasing fecal Many of the same variables in the concentrations on mixed forest land types, 2001 regression model featured in the re- which may explain the relationships ob- gression analysis for 2006 including the served in the both the 2001 and 2006 re- percentage of land in a watershed primar- gression models. Furthermore, this study ily classified as low intensity developed– observed that 66 percent of the water- open space or mixed forest, and whether sheds that had a reduction in mixed forest 446 alford et al.

Table 6. Regression model summary of fecal coliform across the Cape Fear River Basin in 2006. The final regression model for fecal coliform from 2006 can be formally expressed as follows: Log FC (2006) = 2.87 + 0.13 MF + 0.03 DOS + 0.07 SS + 0.67 UCFRB where FC = Fecal coliform col per 100ml., MF = Percent Mixed Forest, DOS = Percent Developed, Open Space, SS = Percent Shrub/Scrub Land, UCFRB = Upper Cape Fear River Basin Region.

Independent Unstandardized Variables b Coefficient Exponentalue V Percent Change

Constant 2.87 Percent Mixed Forest 0.13 1.14 14% Percent Developed, 0.03 1.03 3% Open Space Percent Scrub/Shrub 0.07 1.08 8% Land UCFRB Region 0.67 1.96 96% All p-values <0.05 R2 = 0.31

land from 2001 to 2006 had an increase occurring in the upper reaches of the in low intensity developed–open space, basin. Throughout the basin, low intensity further indicating that this land type is developed–open space typically occurred becoming more dis turbed by human activ- on the periphery of urban areas includ- ities across the river basins over time. ing the cities of Greensboro, High Point Percent low intensity developed–open (Guilford County), Burlington (Alamance space was the second variable to enter County), Durham (Durham and Orange the regression model where a one percent Counties), Chapel Hill (Orange County), increase in low intensity developed–open Fayetteville (Cumberland County), and space resulted in a three percent increase Wilmington (New Hanover County). in fecal concentrations in surface waters Percent shrub/scrub land was the third in the CFRB (see Table 6). Although this predictor variable to enter the regression is a relatively small percent increase in model for 2006. The b coefficient suggested fecal concentrations, it is fairly consist- that a positive relationship existed indicat ent with the 5 percent increase observed ing that a one percentage point increase in the 2001 regression model. Similar to in shrub/scrub land would increase 2001, low intensity developed–open space fecal concentrations across the basin by can typically be found encircling urban- seven percent, holding all other predictor ized areas and often serves as a transi- variables constant. In relation to North tional area between urban development Carolina’s landscape, the North Carolina and forested or agricultural land. From Wildlife Resources Commission (WRC) has 2001 to 2006, 71 percent of the water- noted that shrub/scrub land is typically sheds included in this study experienced located in the Coastal Plains region and is an increase in low intensity developed– characterized by low woody vegetation open space with the largest increases and herbaceous plants. Shrub/scrub land NC Landscape Gradients and Fecal Coliform 447 in this region is typically created by distur- The fourth and final variable to enter bances including clearcutting, disking, or the 2006 regression model for fecal burning and are often found at the transi- coliform was location of the water qual- tion between agricultural lands and forest- ity monitoring station in the Upper CFRB land and is frequently found in the under- region. The regression model suggested story of open pine stands (NC WRC 2013). that fecal concentrations for stations lo- Silviculture practices, which may cated in the Upper CFRB were 96 percent include the clear cutting and or burning higher when compared to other regions of vegetation, are prevalent throughout across the CFRB. As previously noted, the the CFRB, especially in the Middle CFRB Upper CFRB region is the most urbanized and Lower CFRB regions. These practices region in the CFRB and a majority of the often result in accelerated soil erosion monitoring stations’ annual averages and that may carry non-point sources of pol- monthly samples in this region exceeded lution, including fecal, to nearby surface the state standard for fecal concentra- waters (Kasprak et al. 2013). Ensign and tions (i.e. 400 CFU/100ml). Several of Mallin (2001) noted that increases in fecal the Upper CFRB stations’ monthly exceed- concentrations downstream from clear ances have been linked to WWTP spills cutting activities have occurred in the (i.e. overflows and leaks). Unlike the 2001 Lower CFRB even in the presence of a nar- regression model where the Upper CFRB row vegetative buffer. Along transitional region was the first variable to enter the landscapes, such as shrub/scrub land, equation, in the 2006 model the Upper Sebestyen and Verry (2011) observed CFRB region was the fourth and last varia- an increase in fecal pollution following ble to be included. Part of the logic for this the establishment of cattle grazing in a may be linked to the fact that although watershed that was largely converted from the number of stations whose annual forestland to agricultural pasture lands. averages exceeded the state guideline for They noted that although fecal concentra- fecal in this region increased from 2001, tions were present prior to this landscape a majority of these stations did not have transition, primarily from wildlife, there “extreme” fecal counts when compared were dramatic increases in fecal once to those that exceeded state standards in grazing was established. The results of 2001. As a result, the mean fecal coliform this regression model suggests that shrub/ concentration in 2001 as well as the vari- scrub land can contribute a higher per- ance was higher when compared to 2006. centage increase in fecal concentrations Given the large number of WWTPs located (+8 percent) when compared to low inten- in the Upper CFRB, it is likely that aging sity developed–open space (+3 percent), sewer infrastructure (Tavares et al. 2008), although mixed forest land can generate increases in development activities that even higher changes in fecal concentra- put additional strains on these systems, tions (+14 percent). Collectively, these and expansive stormwater systems often land types represent landscape transitions related to impervious surfaces have influ- taking place in the CFRB that may be im- enced the concentration of fecal coliform pairing water quality both locally and in surface water systems within this region throughout the basin network. (Mallin et al. 2000, Holland et al. 2004). 448 alford et al.

can lead to eutrophication-based prob- conclusions and future lems in water bodies (Dodds et al. 1998). research Step-wise linear regression models As the population in the Cape Fear River statistically demonstrated the extent to Basin (CFRB) continues to grow, it will be- which specific land types impacted surface come increasingly important to address waters in both 2001and 2006. Percent the geography of surface water quality mixed forest was the most influential land throughout the basin. Although past re- cover that shaped the geospatial charac- search has highlighted some relationships teristics of fecal concentrations in surface that exist between land types and surface water systems in both 2001 and 2006. water quality at the local watershed and This land type has been associated with subbasin levels, a comprehensive study human disturbances that may result in that specifically addresses the spatial dis- the development of transitional zones in tribution of these relationships across an which forest land is converted to agricul- entire river basin network has been less tural and urban land types over time. Low common. The overall goal of this research intensity developed–open space was also was to identify and spatially illustrate the identified as a land-use associated with in- spatio-temporal relationships between creases in fecal concentrations, although surface water quality (especially fecal its impact was not as significant as mixed coliform) and varying land types across a forest land. In addition, shrub/scrub land single heterogeneous river basin network. cover was also associated with an increase While there were only small changes in in fecal concentrations in 2006. This land land types between 2001 and 2006, spe- type has been identified with human cific land types were empirically linked to disturbances such as land clearing and surface water quality impairment across transitional areas between forestland and this river basin, with fecal coliform bac- agricultural and urban landscapes. teria emerging as a key pollution metric. The transition of forest to the urban When considering regional differences in environment provides a series of well- surface water impairment by fecal coli- recognized steps that increase stream form, this study illustrated that a majority water pollution (Arnold and Gibbons of stations that exceeded the state stand- 1996). As forests are removed, much tran- ard (i.e. 400 CFU/100ml) were located spiration capacity is lost and more runoff in the Upper CFRB. This is of particular occurs; also, with less ground cover ero- interest since the headwaters of the river sion increases. Additionally, the pollution basin are located in this region and ac- buffering ef fect of streamside vegetation is tivities that impair water in the upper reduced or removed. Overall, largely tran- reaches may be compounding surface sitional landscapes that are inclusive of water quality issues downstream. Regard- low intensity developed–open space and ing NO2-NO3, NH3-N and P, we note that a mixed forest appear to play a critical role primary problem with this analysis is that in shaping the geography of water quality North Carolina lacks ambient nutrient across the CFRB in both 2001 and 2006. standards. As a result, we recognize that This is an important finding because the even small increases in nutrient loading presence of these transitional land types NC Landscape Gradients and Fecal Coliform 449 in landscape gradients may be impacting to address the extent to which specific surface water quality on a local and re- activities (e.g. spraying animal waste on gional scale in other river basin systems fields, development, and fertilizer ap- across the country. It is our hope that this plications) are impacting surface water paper is a first step towards providing a resources throughout the river basin platform for research in other river basins network. To better identify these rela- that assesses and spatially illustrates how tionships, more public data sources are the geography of fecal coliform in surface needed that are inclusive of all CAFO ac- water systems is directly connected to spe- tivities (swine, poultry, and cattle) since cific land gradients and configurations. it has been well established that these It should be noted that the national land activities impact surface and groundwa- cover classification system utilized in this ter quality and exclusion of this data is a paper (i.e., MRLC 2012) does not have a limiting factor in analyzing these relation- category for CAFO-impacted land. We know ships throughout the basin. Finally, ap- from previous analyses that CAFOs are plying this research model to other river highly concentrated sources of nutrients and basins with different landscape charac- fecal bacteria (Mallin and Cahoon 2003) teristics and climatic conditions may help that can cause chronic pollution to local wa- elevate our understanding of how vari- terways and adjacent landscapes (i.e. mixed ous landscape configur ations and human forest) that surround CAFO facilities (Mallin activities might impact the quality surface et al. 2015). Thus, a more detailed land-use water resources over large heterogeneous classification system, as well as increased landscapes. rural sampling coverage at a higher spatial resolution could provide greater insight into acknowledgements the geography of surface water quality. We would like to thank the numerous indi- The results of this research also sug- viduals who supported this research endeavor gest several avenues for future research. including the North Carolina Department of More frequent monitoring of surface Environment and Natural Resources (NC DENR) water quality in watersheds characterized staff and our colleagues in the Department of by transitional land types would help im- Geography at the University of North Carolina prove our understanding of how specific at Greensboro. changes to the landscape influence surface water quality across various geographical references cited scales within a single river basin. Also, as Ahearn, D. S., Sheibley, R.W., Dahlgren, there are approximately five million heads R.A., Dahlgren, Anderson, M., Johnson, of swine in the Black and Northeast Cape H., and Tate, K.W. 2005. Land use and land Fear tributary basins of the CFRB (Mallin cover on water quality in the last flowing and Cahoon 2003) these rural areas are river draining the western Sierra Nevada, presently poorly represented by current California. Journal of Hydrology 313: sampling stations highlighting the needs 234–247. for more monitoring stations throughout Arnold, C.J. Jr., and Gibbons, C.J. 1996. the basin. In relation to human activities Impervious surface coverage: on the landscape, more data are needed The emergence of a key environmental 450 alford et al.

indicator. Journal of the American Planning City of High Point 2015, June 12. Consumer Association 62(2): 243–257. Water Report. Retrieved from https://www Aspinall, R. and Pearson, D. 2000. .highpointnc.gov/custsrv/docs/6_10.pdf Integrated geographical assessment Cookson, N. and Schorn, M.S. 2009. of environmental condition in water Correlations of Watershed Housing Density catchments: Linking landscape ecology, with Environmental Conditions and Fish environmental modeling and GIS. Assemblages in a Tennessee Ridge and Journal of Environmental Management Valley Stream. Journal of Freshwater 59: 299–219. Ecology 24(3): 555–561. Booth, D.B. and Jackson, C.R. 1997. Crim, J.E., Schoonover, J.E., and Lockaby, B.G. Urbanization of aquatic systems: 2012. Assessment of fecal coliform and Degradation thresholds, stormwater Escherichia coli across a land cover gradient detection, and the limits of mitigation. in west Georgia streams. Water Quality Journal of the American Water Resources Expo Health 4: 143–158. Association 33(5): 1077–1090. Dodds, W.K., Jones, J.R., and Welch, E.B. Brabec, E. 2009. Impervious surfaces and 1998. Suggested classification of stream water quality: A review of current trophic state: distributions of temperate literature and its implications for stream types by chlorophyll, total nitrogen, watershed planning. Journal of Planning and phosphorus. Water Research 32: Literature 16(4): 499–514. 1455–1462. Burkholder, J., Libra, B., Weyer, P., Heathcote, Ensign, S.H., and Mallin, M. 2001. Stream S., Kolpin, D., Thorne, P.S., and water quality changes following timber Wichman, M. 2007. Impacts of waste from harvest in a coastal plain Swamp Forest. concentrated animal feeding operations Water Resources 35(14): 3381–3390. on water quality. Environmental Health Grimm, N.B., Foster, D., Goffman, P., Grover, Perspective 115(2): 308–312. J.M., Hopkinson, C.S., Nadelhoffer, K.J., Cahoon, L. B., Hales, J. C., Carey, E. S., Pataki, D.E. and Peters, D.P. 2008. The Loucaides, S., Rowland, K. R., and changing landscape: Ecosystem responses Nearhoof, J. E. 2006. Shellfishing to urbanization and pollution across closures in southwest Brunswick climatic and societal gradients. Front County, North Carolina: septic tanks vs. Ecological Environments 6(5): 264–272. storm-water runoff as fecal coliform Hill, D. D., Owens, W. E., and Tchounwou, P. B. sources. Journal of Coastal Research 2006. The impact of rainfall on fecal 22(2): 319–327. coliform bacteria in Bayou Dorcheat Cape Fear River Basin Monitoring Coalitions’ (North Louisiana). International Journal of Water Quality Data 2012, May 10. Environmental Research and Public Health Retrieved from http://lcfrp.uncw.edu/ 3(1): 114–117. riverproject/ Holland, F.A., Sanger, D.M., Gawle, C.P., Carle, M.V., Halpin, P.N., and Stow, C.A. Lerberg, S.B., Santiago, M.S., Riekerk, 2005. Patterns of watershed urbanization G.H., Zimmerman, L.E., and Scott, G.I. and impacts on water quality. Journal of 2004. Linkages between tidal creek the American Water Resources Association ecosystems and the landscape and 41(3): 693–708. demographic attributes of their watersheds. NC Landscape Gradients and Fecal Coliform 451

Journal of Experimental Marine Biology and Mallin, M., Williams, K. E., Esham, E.C., Ecology 298: 151–178. and Lowe, R.P. 2000. Effects of human Huang, J., Li, Q., Pontius Jr., R.G., Klemas, V., development on bacteriological water and Hong, H. 2013. Detecting dynamic quality in coastal watersheds. Ecological linkage between landscape characteristics Applications 10(4): 1047–1056. and water quality in a subtropical coastal Mallin, M.A., McIver, M.R., Ensign, S.H., and watershed, southeast China. Environmental Cahoon, L.B. 2004. Photosynthetic and Management 51: 32–44. heterotrophic impacts of nutrient loading to Kasprak, A., Magilligan, F.J., Nislow, K.H., blackwater streams. Ecological Applications Renshaw, C.E., Snyder, N.P., and Dade, 14: 823–838. W. B. 2013. Differentiating the relative Mallin, M.A., McIver, M.R., Robuck, A.R. and importance of land cover change and Dickens, A.K. 2015. Industrial swine and geomorphic processes on find sediment poultry production causes chronic nutrient sequestration in a logged watershed. and fecal microbial stream pollution. Water, Geomorphology 185: 67–77. Air and Soil Pollution 226(12): 407. Kelsey, H., Porter, D.E., Scott, G., Neal, M., Merwade, V. 2012, May 10. Terrain Processing. and White, D. 2004. Using geographic Retrieved from https://web.ics.purdue information systems and regression .edu/~vmerwade/education/terrain_ analysis to evaluate relationships between processing.pdf land use and fecal coliform bacterial Meybeck, M. and Helmer, R. 1989. The quality of pollution. Journal of Experimental Biology rivers: From pristine stage to global pollution. and Ecology 298(2): 197–209. Palaeogeography, Palaeoclimatology, Line, D.E., White, N.M., Kirby-Smith, W.W., and Palaeoecology 75: 283–309. Potts, J.D. 2008. Fecal coliform export from Multi-Resolution Land Characteristics four coastal NC areas. Journal American Consortium (MRLC) 2012, April 3. National Water Resources Association 44(3): 606–617. Land Cover Program. Retrieved from http:// Liu, Z.-J., Weller, D.E., Correll, D.L., and www.mrlc.gov/finddata.php Jordan, T.E. 2000. Effects of Land Cover North Carolina Department of Environment and Geology on Stream Chemistry in and Natural Resources (NC DENR) 2005, Watersheds of Chesapeake Bay. Journal of May 12. Cape Fear River Basin Plan. the American Water Resources Association Retrieved from http://h2o.enr.state.nc.us/ 36(6): 1349 –1365. basinwide/draftCPFApril2005.htm Mallin, M. and Cahoon, L.B. 2003. North Carolina Department of Environment Industrialized animal production – A major and Natural Resources Office of source of nutrient and microbial pollution Environmental Education (NC DENR to aquatic ecosystems. Population and EE) 2012. North Carolina River Environment 24(5): 369–385. Basins. Retrieved from http://www Mallin, M. 2013. (Photograph). CAFO sites .eenorthcarolina.org/riverbasins.html. and mixed forest land, Duplin County, NC. North Carolina Department of Environment and Personal Collection. Natural Resources (NC DENR) 2012, May Mallin, M. 2015. (Photograph). Suburban 24. Cape Fear River Basin Integrated Report. development and mixed forest land, Cape Retrieved from http://portal.ncdenr.org/ Fear River Basin, NC. Personal Collection. web/wq/ps/bpu/basin/capefear 452 alford et al.

North Carolina Department of Environment brighteners and fecal bacteria to detect and Natural Resources (NC DENR) 2015, sewage leaks in tidal creeks. Journal of the June 10. North Carolina Academy of Science 124: Phase II Stormwater Regulations. Retrieved 91–97. from http://portal.ncdenr.org/web/lr/ Tong, S. T. Y., and Chen, W. 2002. Modeling phase-ii-designations the relationship between land use North Carolina Wildlife Resources and surface water quality. Journal of Commission (NC WRC) 2013, March 2. Environmental Management 66: 377–393. The Cape Fear River Basin. Retrieved from Tu, J. 2011. Spatially varying relationships http://www.ncwildlife.org/Conserving/ between land use and water quality Habitats/NorthCarolinaRiverBasins/ across an urbanization gradient explored NCRiverBasinMap/CapeFearRiverBasin by geographically weighted regression. .aspx Applied Geography 31: 376–392. Potter, K.M., Cubbage, F.W., Blank, G.B. and Tu, J., Xia, Z., Clarke, K.C. and Frei, A. 2007. Schaberg, R.H. 2004. A watershed-scale Impact of urban sprawl on water quality in model for predicting nonpoint pollution eastern Massachusetts, USA. Environmental risk in North Carolina. Environmental Management 40: 183–200. Management 34(1): 62–74. United State Department of Agriculture Rothenberger, M.B., Burkholder, J. M. and (USDA) 2012, May 10. Geospatial Data Brownie, C. 2009. Long-term effects of Gateway. Retrieved from https://gdg changing land use practices on surface water .sc.egov.usda.gov/ quality in a coastal river and lagoonal estuary. United States Environmental Protection Agency Environmental Management, 44: 505–523. (US EPA) 2010, May 14. Clean Water Act Schoonover, J.E. and Lockaby, B.G. 2006. Land Section 319 Program. Retrieved from cover impacts on a stream nutrients and http://www.epa.gov/owow_keep/NPS/ fecal coliform in the lower piedmont of west cwact.html Georgia. Journal of Hydrology 331: 371–382. United States Environmental Protection Schoonover, J.E., Lockaby, B.G. and Pan, S. Agency (US EPA) 2013, June 12. National 2005. Changes in chemical and physical Assessment Datebase. Retrieved from http:// properties of stream water across an urban- www.epa.gov/waters/305b/ rural gradient in western Georgia. Urban United State Geological Survey (USGS) 2000. Ecosystems 8: 107–124. Fecal-coliform bacteria concentrations in Schueler, T.R. 1994. The Importance of streams of the Chattahoochee River national Imperviousness. Watershed Protection recreation areas in metropolitan Atlanta, Techniques 1(3): 100–111. Georgia, May–October 1994 and 1995. U.S. Sebestyen, S. D., and Verry, E. S. 2011. Effects of Department of Interior Report. Retrieved watershed experiments on water chemistry from http://pubs.er.usgs.gov/publication/ at the Marcell Experimental Forest. Peatland wri20004139 biogeochemistry and watershed hydrology at United State Geological Survey (USGS) the Marcell Experimental Forest. CRC Press, 2015, July 25. Methods for Computing Boca Raton 433–458. Water Quality Using Regression Analysis. Tavares, M.E., Spivey, M.I.H., McIver, M.R. Retrieved from http://nrtwq.usgs.gov/ and Mallin, M.A. 2008. Testing for optical md/methods/ NC Landscape Gradients and Fecal Coliform 453

United State Geological Survey (USGS) Professor of Biology in the Environmental 2003. Patterns and sources of fecal coliform Studies program at Salem College Winston- bacteria in three streams in Virginia, Salem, NC 27101 (jennifer.alford@salem 1999–2000. U.S. Department of Interior .edu). Her research interests focus on human- Report. Retrieved from http://pubs.usgs environmental relationships across multiple .gov/wri/wri034115/ geographical scales including environmental United State Geological Survey 2016, June 13. sustainability, green infrastructure and Water Resources Data Report. Retrieved watershed management. from http://pubs.usgs.gov/wdr/2004/ dr. keith g. debbage (kdebb@hotmail wdr-nc-04/figures.html .com) is a Professor in the Department of Walsh, C.J., Roy, A.H., Feminella, J.W., Geography at the University of North Carolina Cottingham, P.D., Goffman, P.M., and at Greensboro in Greensboro, NC 27402. His Morgan, II, R.P. 2005. The Urban Stream research interests include urban development, Syndrome: Current Knowledge and the tourism and air transportation. Search for a Cure. Journal of Benthol Society dr. michael a. mallin (mallinm@uncw 24(3): 706–723. .edu) is Research Professor at the University of Wear, D.N., Turner, M.G. and Naiman, R.J. North Carolina Wilmington Center for Marine 1998. Land cover along an urban-rural Sciences Wilmington, NC 28403. He is an gradient: Implications for water quality. aquatic ecologist who specializes in pollution Ecological Applications 8(3):619–630. issues regarding fresh water, estuarine systems Wilson, C. and Weng, Q. 2010. Assessing and the coastal ocean. His publications have surface water quality and its relation included studies on the environmental impacts with urban land cover changes in the of industrial livestock production, urban and Lake Calumt Area, Greater Chicago. rural stormwater runoff, sewage and septic Environmental Management 45: 1096–1111. system failures and environmental effects of Yuncong, L. and Migliaccio, K. 2011. Water major storms, as well as studies on mitigative Quality Concepts, Sampling, & Analysis. measures to reduce pollution. Boca Raton, FL: Taylor and Francis Group. dr. zhi-jun liu ([email protected]) is an Associate Professor in the Department of Geography at the University of North Carolina dr. jennifer braswell alford is at Greensboro in Greensboro, NC 27402. His an Assistant Professor of Geography and research interests focus on the interaction of Environmental Science at Elon University Elon, human activities and natural systems and its NC 27244 ([email protected]) and an Adjunct impact on the environment.