9.1 Introduction
   In recent years the lower Cape Fear watershed in eastern North Carolina has been subject to a number of hurricanes. This region contains both rapidly urbanizing coastal areas and large numbers of concentrated animal operations (CAOs) in the middle and upper Coastal Plain. Anthropogenic development can magnify the effect of major storms on water quality damage, as was the case for Hurricane Andrew in Biscayne Bay (Tilmant et al. 1994) and Hurricane Hugo in Charleston, S.C. (Van Dolah and Anderson 1991). While Hurricane Bertha did little environmental damage in North Carolina in July 1996, Hurricane Fran in September 1996 led to severe water quality impacts, largely as a result of anthropogenic factors (Mallin et al. 1999).
    On August 26, 1998, Category 3 Hurricane Bonnie made landfall near Wilmington, North Carolina (approximately 33^o55’ N latitude and 77^o50’ W longitude). The hurricane remained nearly stationary for several hours, deluging the city with 20 inches of rain. It then moved slowly north-northwest, passing over the Northeast Cape Fear River watershed and eventually the New River, subsequently downgrading to a tropical storm. The hurricane caused power outages and flooding problems that were exacerbated twelve days later when the remnants of Tropical Storm Earl deposited another four inches of rain in the Wilmington area.
    Prior to the arrival of Hurricane Bonnie, the UNCW Lower Cape Fear River Program had been collecting monthly physical, chemical, and biological samples at 35 stations throughout the watershed. (Fig. 1.1). Benthic invertebrates and fish were also being sampled at a number of locations. We increased our sampling frequency at a number of locations following the passage of Bonnie to obtain more detailed information on its aquatic impacts and subsequent ecosystem recovery.

9.2 Methods
   Regular monthly physical water quality sampling included temperature, dissolved oxygen, pH, salinity/conductivity, and turbidity (measured by a YSI 6920 multiparameter water quality instrument). Light attenuation coefficients k were calculated from vertical PAR measurements taken with a LiCor-1000 Datalogger coupled with a spherical sensor. Nitrate+nitrite, ammonium, total Kjeldahl nitrogen (TKN), and total phosphorus were measured using Standard Methods 4500-NO3-F, 4500-NH3-B, 4500Norg-B, and 4500P-E, respectively (APHA 1995). Orthophosphate was measured using EPA Method 365.5 on a Technicon AutoAnalyzer. Five-day biochemical oxygen demand (BOD5) was assessed using Standard Methods 5210-B, and fecal coliform bacteria were enumerated by membrane filtration using Standard Methods 9222-D (APHA 1995). A complete set of samples was taken from all stations August 31-September 3. Physical parameters were collected on a weekly basis through the month of September at riverine and estuarine stations following the main collection. Physical parameters and fecal coliforms were also taken at Stations NC11, AC, DP, IC, NAV, and HB on September 8, and those parameters plus nitrate, ammonium, and orthophosphate were collected at Stations B210, SAR, NCF41, NCF53, NCF210, NCF117 on September 9. BOD5 was assessed at Stations BRN, COL, GCO on August 31, at NC11, AC, BBT, B53, B210, NCF53, NCF210, NCF117, and NCF6 on September 2, at NCF41, SAR, and ANC on September 3, and at NCF41 and NCF117 on September 9. From September 11 until September 28 we maintained a YSI Model 6920 multiparameter instrument in-situ approximately 1 m below the surface at NCF117, changing instruments approximately every six days. In addition to water quality samples, we obtained rainfall data from the State Climatologist Office in Raleigh, N.C., and river flow data from the U.S. Geological Survey in Raleigh. Data on human sewage bypasses were obtained from the North Carolina Division of Water Quality.

9.3 Results

Water Quality

Rainfall and River Flow: August and September 1998 rainfall at Clinton, a meteorological station representing the Black River was 7.3 and 3.2 inches, respectively. August and September rainfall at Castle Hayne, a station representing the lower Northeast Cape Fear River, was much higher at 18.0 and 6.4 inches, respectively. Rainfall in Fayetteville on the mainstem Cape Fear was near normal for September, as was river flow at Lock and Dam #1 on the mainstem. Peak flow in both August and September for the Black River was 1360 CFS, compared with the 1952-1997 mean of 713 CFS (August), and 620 CFS (September). Peak August and September flows for the Northeast Cape Fear River were 6900 CFS and 4580 CFS, respectively, compared with long-term averages of 918 CFS and 1072 CFS, respectively. The elevated rainfall and river flow conditions led to water-column freshwater conditions as far down the estuary as Channel Marker 54, with a strong salinity wedge forming downstream of M54. At M18, near the mouth of the estuary at Southport, surface salinity a week after the hurricane was 7.1 while bottom salinity was 30.1 ppt.

Light Attenuation: Inputs of highly colored swamp water began entering the mainstem from Livingston Creek (LVC), downstream of Station N11. This had the effect of increasing light attenuation k values from 2.60 at N11 to 5.48 at AC one week after Bonnie’s passage (Table 9.1). The rest of the river and upper to middle estuarine stations maintained September k values between 4.0 and 4.9, while the lower estuarine stations had much less light attenuation because of significant seawater dilution (Table 9.1). Surface turbidity was comparatively low (<20 NTU) in the river and estuary, so the high light attenuation was evidently caused by dissolved organic compounds.

Biochemical Oxygen Demand: BOD5 assessments were made on water collected at a number of stations between five and nine days following the hurricane. BOD5 in the Cape Fear mainstem at NC11 showed little difference from the long-term average (Table 9.2). Station AC lies 15 km downstream of NC11, and is affected by BOD inputs from a pulp and paper mill and a stream draining several other industries. BOD at this location considerably increased over normal (Table 9.2). BOD in the upper Black River at B53 was relatively low, and increased downstream at B210 to three times the long-term average, indicating some inputs from Bonnie. Station BBT is on the lower Black River but also receives inputs from the mainstem below AC via Thoroghfare and tidal movement, and BOD here was about four time normal (Table 9.2). The upper Northeast Cape Fear River displayed some of the heaviest BOD loads in the system following Bonnie, while BOD further downstream was lower but still well above normal (Table 9.2). Based on long-term data from NCF117, BOD in the Northeast Cape Fear River ranged from two to eight times normal after this hurricane.

Dissolved Oxygen: The week after Bonnie, dissolved oxygen levels dropped to very low levels in the lower river and estuary as far downstream as M35 (Fig. 9.1). The upper and middle estuary showed severe hypoxia until the middle of September, when DO levels rose above 2.0 mg/L at the surface. A number of stream stations in the Northeast Cape Fear and Black River watershed also experienced hypoxic to anoxic conditions. These were NCF117 (0.0 mg/L), ANC, SAR, and GS (all 0.3 mg/L), ROC (0.6 mg/L), GCO and SR (1.0 mg/L), COL (1.1 mg/L), B210 (1.2 mg/L), N403 (1.3 mg/L), PB (1.4 mg/L), and 6RC (2.2 mg/L). The other stream stations maintained DO levels of 4.0 mg/L or higher. The Northeast Cape Fear River experienced DO conditions <1.0 mg/L from SAR (100 km upstream of Wilmington) down to NCF6 (6 km upstream of Wilmington). Data from the in-situ meter at NCF117 indicated near-anoxic conditions persisting from late August until September 18, and not increasing to 2.0 mg/L until September 26. Recovery to DO levels exceeding 5.0 occurred only in early November, about 2.5 months after the passage of Bonnie (Fig. 9.2). Of the three main tributaries, the Northeast Cape Fear River experienced the lowest dissolved oxygen, followed in turn by the Black River and the mainstem Cape Fear (Fig. 9.3).

Nutrient Inputs: Nutrient concentrations were normal in the mainstem Cape Fear River following passage of Bonnie. However, distinct pulses of ammonium, TKN, TP, and orthophosphate occurred during the first two weeks of September at stations in the Black and Northeast Cape Fear Rivers (Table 9.3). Nutrient concentrations were highest in the uppermost reaches of the Northeast Cape Fear River (Stations NC41 and SAR) and decreased downstream. TKN was much higher than normal at SAR, but was similar to long-term concentrations at the remaining stations. At several stations ammonium was elevated, but nitrate was much lower than normal at all stations sampled (Table 9.3).  Nitrate was likely low due to the hypoxic conditions that followed the hurricane. It is also evident that organic nitrogen concentrations were considerably elevated at most of the stations. Total P and orthophosphate were both 2-3X greater than normal at most locations, particularly the upper river stations.

Fecal Coliform Bacteria: Following the hurricane there were sharp increases in fecal coliform counts throughout the watershed, with the exception of the upper mainstem Cape Fear River and the upper Black River (Table 9.4). Several loci of unusually high counts were the lower Cape Fear River and upper estuary, the upper and lower Northeast Cape Fear River, and several stream stations in the Black and Northeast Cape Fear River watersheds (Table 9.4).

Comparing an Anthropogenically Impacted Watershed with a Near-Pristine Watershed: Swamp water in summer typically maintains reduced DO, and can be a source of BOD to riparian systems. We wanted to compare the drainage of a wetland-rich, sparsely developed watershed with one under heavy agricultural usage. Colly Creek contains extensive riparian wetlands and only six concentrated animal operations (CAO’s). By contrast, Great Coharrie Creek contains 95 CAO’s. Samples collected at stations draining these two watersheds yielded considerably different constituent concentrations (Table 9.5). The very low pH of Colly Creek (3.8) indicated the blackwater swamp origination of the drainage, compared to the pH of 6.1 at Great Coharrie Creek. Colly Creek turbidity was lower than that of Great Coharrie as well (Table 9.5). Drainage from Great Coharrie Creek had about twice the BOD as Colly Creek drainage, but approximately 240X the fecal coliform bacterial concentrations (Table 9.5). Total phosphorus in Great Coharrie Creek was 10X that of Colly Creek. There was less difference in total nitrogen between the systems, with TN in both cases comprised almost totally of organic nitrogen and ammonium (Table 9.5).

Benthos

    The passage of Hurricane Bonnie in 1998 strongly affected the infaunal community at NCF6. An average of only 0.8 individuals per grab (equivalent to 4 total individuals) was found in samples taken after the hurricane compared to 3.2 individuals per grab immediately before the hurricane. The abundances observed at this site after Hurricane Bonnie represented the lowest observed at any site on any date since our monitoring began in 1996. However, it should be noted that abundances in August 1998, before the hurricane, were already low, possibly reflecting a period of natural decline in the community. Hurricane Bonnie had similar qualitative effects at NAV as at the other low salinity site. An average of 3.2 individuals per grab was found in samples taken after the passage of the hurricane compared to 10 individuals per grab immediately before the hurricane (Table 9.5). The abundances observed after Hurricane Bonnie were the lowest observed at this site for any date since monitoring began in 1996. However, it should be noted that abundances in August 1998, before the hurricane, were already low (representing the second lowest recorded value), suggesting the community was already in decline before the hurricane hit.
    Unlike the two upstream sites, there was little evidence for declines in faunal abundance at M54 associated with the passage of Hurricane Bonnie. Before the hurricane, total faunal abundance was 7.6 per grab, while after the hurricane abundances were 26.8 per grab (Table 9.5). The September 1998 abundances were higher than found in September 1997 or in October 1996 (after Hurricane Fran). It is possible that abundances at this site were elevated after Hurricane Bonnie because of wash down of fauna from upstream areas associated with increased water flow, but more than 1 non-hurricane year is needed to establish baseline trends. In contrast, at M31 there was evidence for significant declines in faunal abundance associated with the passage of Hurricane Bonnie. Before the hurricane, total faunal abundance was 67.8 per grab, while after the hurricane abundances were 4.4 per grab (Table 9.5). As at the NCF6 and NAV sites, the post-Bonnie faunal densities were the lowest recorded for the site since monitoring began. Unlike the other sites, there was no evidence that faunal abundances were in decline before the hurricane. Possible causes for faunal declines after Hurricane Bonnie include the exceptionally low salinity at the site (0.1 ppt) and low DO.

9.4 Discussion
   A variety of anthropogenic sources contributed BOD-containing materials to the LCFRP system following Hurricane Bonnie. Human wastewater treatment systems suffered power failures and had to reroute untreated or partially treated sewage into receiving waters. Reported major discharges included 94,000 gallons of primary treated wastewater from Faison (which had to flow down 20 miles of stream to reach the Northeast Cape River). Another 159,000 gallons of untreated human waste from Beulaville, 1,000,000 gallons of untreated wastewater from the Burgaw Treatment Plant, and 10,000,000 gallons of partially treated wastewater from the Wilmington Northside Treatment Plant (BOD of 36 mg/L – K. Vogt, City of Wilmington) entered the Northeast Cape Fear River. Other smaller discharges occurred in tributaries of the Cape Fear Estuary. Also, a swine waste lagoon on a floodplain near Chinquapin was inundated by high water, releasing unknown amounts of swine waste to the Northeast Cape Fear River. All of this material, in addition to swamp water, contributed BOD loading to the LCFR system. The human and animal waste also has the added human health danger of loading potentially pathogenic microbes to the river system.
    In recent years the Cape Fear system, particularly the Northeast Cape Fear, has frequently been subject to incidents of hypoxia and anoxia (Fig. 9.3). 1997 can be considered a baseline year for dissolved oxygen, since no major pollution spills or hurricanes occurred. In 1997 summer DO levels generally ranged between 4.0 and 6.0 mg/L, compared with the state standard of 5.0 mg/L. A major decline in 1995 was caused by a nine million gallon poultry lagoon rupture in the Northeast Cape Fear River (Mallin et al. 1997), and severe decreases in 1996 and 1998 followed Hurricanes Fran and Bonnie (Mallin et al. 1999). Severely reduced dissolved oxygen concentrations (hypoxia and anoxia) in rivers and estuaries result from inputs of BOD. In the Cape Fear system, swamp water forced into the streams and main channels is one such source of BOD. However, the areas with the largest recorded BOD values were also severely impacted by anthropogenic sources of BOD. In the two weeks following the passage of Hurricane Bonnie, many swine waste lagoons were in danger of overtopping, particularly with rainfall from Tropical Storm Earl further exacerbating the situation. In the upper Northeast Cape Fear River (SAR and NCF41) field biologists reported widespread spraying of swine lagoon wastes onto fields that were already saturated by rain from these storms. These two stations had very high BOD values (8X the river norm -Table 9.2), indicating large amounts of BOD laden runoff entering the river nearby. Farther downstream, the Northeast Cape Fear River was impacted by sewage bypasses, with river BOD about 4X normal (Table 9.2).
    High fecal coliform concentrations are indicative of untreated human or animal waste entering a water body. As another means of testing whether the high BOD loads were associated with anthropogenic loading, for the week following Hurricane Bonnie we ran a correlation analysis between BOD5 and fecal coliform counts for eleven stations where data for both parameters were available. There was a significant and positive correlation between these parameters (r = 0.605; p = 0.048) indicating the likelihood of similar anthropogenic sources for both parameters. We feel that this was a very conservative analysis since we excluded two stations from the potential matrix (SAR and ANC, with BOD5 levels of 8.7 mg/L and 8.2 mg/L, respectively) because the accompanying fecal coliform counts were classified as too numerous to count (TNTC). Since other stations exhibiting high BOD (such as AC and BBT) had fecal coliform counts between 2,300 and 3,800 CFU/100 mL, availability of usable fecal coliform data from SAR and ANC would have likely strengthened the correlation between these two parameters.
    Many LCFRP stations exhibited high fecal coliform counts following Hurricane Bonnie, demonstrating that there is an important human health concern in post-hurricane water bodies. For example, the Burgaw Wastewater Treatment Plant had to bypass one million gallons of untreated sewage, and the station downstream of the plant (BC117) registered counts of 16,700 CFU/100 mL. We recommend that civic health authorities post warnings at all public boat ramps and in the newspapers that water may not be safe for human contact for a period of at least three weeks following hurricane events.
    High concentrations of ammonium, orthophosphate and TP can also be considered indicative of human and animal waste inputs. Also, sediments will release more orthophosphate under severely hypoxic to anoxic conditions. These nutrient parameters were all elevated well above normal in most of the blackwater river stations (Table 9.3). High TKN can also be found naturally in swamp waters. Our comparison between Colly Creek and Great Coharrie Creek showed high TKN in Colly Creek, but less so than in the CAO-rich Great Coharrie Creek. Additionally, our watershed comparison found a tenfold difference in TP in waters exiting the two systems. We suspect that non-point source runoff from swine lagoon waste sprayfields, and bypasses of incompletely treated human sewage were important sources of nutrients to stream and river stations in the LCFR following Hurricane Bonnie.
    The effect of water quality changes on the benthic community varied by station. The community at NAV was severely reduced in September, but recovered by early October. This was likely due to the relatively rapid recovery of dissolved oxygen in the Cape Fear mainstem, compared with the Northeast Cape Fear River. The infaunal community at NCF6 was low in August before the hurricane, then declined to its lowest abundance since our sampling began following the hurricane. Abundances at NCF6 remained low in the November sampling as well. Infaunal abundances actually increased at M54 following Bonnie, and stayed at that level in October. This is in contrast to the declines suffered at this station after Hurricane Fran (Mallin et al. 1999). As mentioned, this may have resulted from a wash-down of organisms from higher up in the estuary. Abundances at M31 were high before Bonnie, than drastically declined, either as a result of low DO or salinity changes. We will continue to document recovery of the infaunal community as samples are processed. A longer-term data base would be helpful in separating effects of acute events from background seasonality. However, the fact that three of the four stations showed the lowest infaunal abundances since the project inception is cause for concern over the recovery ability of a system from successive major incidents.
    Both Hurricane Fran and Hurricane Bonnie caused severe environmental damage and increased risk of human exposure to potential microbial pathogens. The lower Cape Fear watershed contains many point-source sewage outfalls and the largest concentration of industrial-style swine farms in North Carolina. Hurricane impacts to these pollutant sources led to large inputs of nutrients, oxygen-consuming organic wastes (BOD) and fecal bacteria into the river system. Following both Fran and Bonnie the most severe damage, as measured by water quality indicators and losses in the benthic community, was in the Northeast Cape Fear River. This tributary hosts both point sources and numerous CAOs; additionally, lower flows than the Cape Fear mainstem and strong tidal action helps to retain pollutants for extended periods. While the river hydrology cannot be changed, better management of animal wastes from CAOs on the floodplain and reliable backup generating systems to treat human waste during storm events will help reduce post-storm pollutant impacts to the biotic communities.

9.5 Acknowledgments
   Funding was provided by the Lower Cape Fear River Program and the Water Resources Research Institute (Project # 70171). For field and laboratory help we thank Booty Baldridge of Cape Fear River Watch, Doug Parsons, Christian Preziosi, Ashley Skeen, and Tracey Wheeler. Helpful information was provided by Rick Shiver, Jimmie Overton, and Linda Forehand of the North Carolina Division of Water Quality, Ryan Boyles and Wendy Sellers of the State Climate Office at North Carolina State University, and Alex Cardinell of the U.S. Geological Survey.

9.6 References Cited

Mallin, M.A., J.M. Burkholder, M.R. McIver, G.C. Shank , H.B. Glasgow, Jr., B.W. Touchette and J. Springer. 1997. Comparative effects of poultry and swine waste lagoon spills on the quality of receiving streamwaters. Journal of Environmental Quality 26:1622-1631.

Mallin, M.A., L.B. Cahoon, D.C. Parsons and S.H. Ensign. 1998. Effect of organic and inorganic nutrient loading on photosynthetic and heterotrophic plankton communities in blackwater rivers. Report No. 315. Water Resources Research Institute of the University of North Carolina, Raleigh, N.C.

Mallin, M.A., M.H. Posey, G.C. Shank, M.R. McIver, S.H. Ensign and T.D. Alphin. 1999. Hurricane effects on water quality and benthos in the Cape Fear Watershed: Natural and anthropogenic impacts. Ecological Applications 9:350-362.

Tilmant, J.T., R.W. Curry, R. Jones, A. Szmant, J.C. Zieman, M. Flora, M.B. Roblee, D. Smith, R.W. Snow and H. Wanless. 1994. Hurricane Andrew’s effect on marine resources. BioScience 44:230-237.

Van Dolah, R.F. and G.S. Anderson. 1991. Effects of Hurricane Hugo on salinity and dissolved oxygen conditions in the Charleston Harbor estuary. Journal of Coastal Research 8:83-94.