NUTRIENT AND SEDIMENT LOADING IN SOUGAHATCHEE CREEK AND THE IMPACTS ON AQUATIC BIOTA. Report submitted to West Point Stevens and the

Transcription

1 NUTRIENT AND SEDIMENT LOADING IN SOUGAHATCHEE CREEK AND THE IMPACTS ON AQUATIC BIOTA Report submitted to West Point Stevens and the Cities of Auburn and Opelika, Alabama By David R. Bayne Eric M. Reutebuch Wendy C. Seesock E. Cliff Webber Contributors Gina Logiudice Michael Len Stephen Turner Karen Popp Justin Mitchell Todd Risk John Woodfin Andrew Rypel February 2004 Auburn University Department of Fisheries and Allied Aquacultures Alabama Agricultural Experiment Station

2 EXECUTIVE SUMMARY The Sougahatchee Creek headwaters arise just north of the city of Opelika, Alabama and flow westward to the Tallapoosa River (Yates Lake). The 217-mi² creek basin lies entirely within the Piedmont physiographic province. The Piedmont comprises a transitional area between the mostly mountainous Appalachians to the northeast and the relatively flat Coastal Plain to the southeast. The soils are finer-textured and lower in organic matter and nutrients than Coastal Plain soils. Rocky stream bottoms of boulder, cobble and gravel characterize streams in their natural, undisturbed condition. In its headwaters, Sougahatchee Creek received nonpoint urban runoff from the twin cities of Auburn and Opelika, Alabama, about 4.5 mgd of municipal wastewater and another 1.3 mgd of treated textile wastewater. The predominant (67%) land cover within the basin was managed forest. The stream has had a long history of pollution problems and, in fact, is in much better condition today than it was years ago. Nevertheless, Sougahatchee Creek continues to experience problems with excessive nutrient enrichment and sedimentation. It was one of the few Alabama streams to have two reaches (embayment of Yates Lake and an urban tributary) appearing on the State 303(d) List for impaired waters in This study was designed to determine the current condition of Sougahatchee Creek and its tributaries and, to the extent possible, identify those factors within the basin having an adverse effect on water quality and biological health of the system. The specific objectives were to: 1. Measure concentrations and estimate annual loading of plant nutrients, total suspended solids and metals; 2. Examine physical habitat conditions of the streams and conduct bioassessments utilizing benthic macroinvertebrate communities; 3. Using GIS technology, examine and quantify land cover for the entire Sougahatchee Basin from 1993 and 2001 satellite imagery and aerial photography; and, 4. Using regression analysis, examine relationships between basin land cover, nutrient and sediment loading and biological condition of the streams. Methods Water quality sampling began in January 2001 and continued through December Samples were collected at three wastewater discharges, at nine mainstream Sougahatchee Creek locations and at 12 tributary stream locations on 40 days during the 2- year study. Sampling was concentrated during those months (November through April) normally receiving more rainfall. All samples were analyzed for nutrients (N and P), suspended solids, metals and biochemical oxygen demand. A USGS gage was in operation at Lee Co. Road 188 (station 8), but stream discharge at all other stations was measured on the day of sampling. Annual loading estimates for nutrients, suspended solids and metals were calculated utilizing constituent concentration, instantaneous stream discharge and mean daily discharge for each sampling location. Nonpoint source loading was estimated by subtracting measured point source loading from total loading. ii

3 At selected stream sites in the Sougahatchee Basin the physical habitat was assessed utilizing visual observations and a standard form that ranks bottom substrate and cover, flow conditions, sedimentation, bank stability and riparian vegetation. At these same sites the biological condition was assessed utilizing rapid bioassessment procedures approved by the US Environmental Protection Agency. This approach utilizes benthic macroinvertebrates sampled from multiple habitats (e.g., cobble and boulder, logs, root masses along undercut banks and leaf packs) at each site. From the composite sample of macroinvertebrates collected in the reach, six metrics such as total number of taxa and the EPT Index are calculated for a multi-metric assessment of biological condition. The biological condition is then calculated by determining the percent comparability of the site to that measured at a reference site (relatively undisturbed) from the same ecoregion. Reference sites were sampled during the same time period as the streams in the Sougahatchee Basin. The physical habitat assessment and determination of biological condition was conducted on two dates in the fall each year of the study. Land cover for 1993 was obtained in grid format from the U. S. Geological Survey, National Land Cover Data and was developed using 1993 leaf-off and leaf-on Landsat thematic mapper satellite imagery. The 2001 land cover was classified from 2001 leaf-on and leaf-off Landsat 7 thematic mapper images. For both years, the numerous land cover classes were consolidated into eight primary land cover classes: water, urban/suburban, clearcut/barren, quarry/gravel pits, forest, pasture/grassland, tilled agriculture and wetlands. Annual loads of select variables (SRP, TP, TN, TSS, CBOD 5 ) were converted to metric tons/km 2 yr -1 (annual load per unit watershed area) and then regressed against the various land cover classes. Tributary stream watersheds were regressed separately from Sougahatchee Creek mainstream watersheds. Biological metrics derived from the health of the macroinvertebrate community were also regressed against watershed land cover. Water Quality Meteorological conditions affect water quantity and water quality of streams and lakes. Both years of this study, 2001 and 2002, were warmer and dryer than normal. Attempts to measure nutrient and sediment loading of Sougahatchee Creek under drought conditions likely underestimated the contribution of nonpoint sources of pollution to the total load. Mainstream Concentrations. The headwaters of Sougahatchee Creek were heavily influenced by activities and development in the urban areas of both Auburn and Opelika, Alabama. Treated effluent from two municipal wastewater treatment plants (WWTP), Opelika Westside (3.1 mgd) and Auburn Northside (1.4 mgd), and one industrial (textile) facility, West Point Stevens (1.3 mgd) entered Sougahatchee Creek in the upstream onethird of the basin. Waste from these facilities along with urban drainage and stormwater runoff from the cities of Auburn and Opelika caused an upstream peak in nutrient concentration. All nitrogen species peaked at mainstream station 6 (US Highway 280 bridge) and declined steadily downstream to station 12 near Reeltown, Alabama both years. Soluble reactive phosphorus and total phosphorus peaked at the first mainstream sampling iii

4 location (station 8) downstream from the Auburn Northside WWTP (station 3) where the mean phosphorus concentration in the effluent was usually considerably higher than that measured in other waste effluents. Elements in relatively high concentration in one or more of the point source effluents (e.g. Ca, Na, K, Mg and B) peaked at upstream station 6 and declined progressively downstream. Other minerals, like barium, aluminum, iron and manganese, showed no clear concentration pattern from headwaters to the mouth of Sougahatchee Creek. Annual mean concentrations of total suspended solids (TSS) in Sougahatchee Creek were variable both years with highest concentrations at station 10 in 2001 and station 8 in Variation in TSS concentration was pronounced both within a station and between stations among years. WWTP s contributed relatively little sediment to Sougahatchee Creek. Mainstream Loading. When total load was higher than point source load at a station, estimates of mean annual point and nonpoint source contributions to total load were calculated by subtracting the point source load (measured at the point source) from total load (measured at the stream sampling location). This approach assumes that all of a constituent entering a stream from a source travels all the way downstream to the mouth, an assumption that may not always hold true. Constituent loading was higher in 2001 than in 2002 because of the greater runoff and mean daily discharge that occurred that year. In 2001, total nitrogen (TN) load increased in a downstream direction from less than 7.0 metric tons (mt) at upstream station 4 to more than 233 mt at downstream station 12. The nonpoint source contribution of TN increased progressively downstream and the point source contribution decreased. In 2002, with lower hydraulic discharge, TN loading increased from less than 6 mt at station 4 to 130 mt at station 10 and then decreased downstream to 120 mt at station 12. Nonpoint source TN contributions were generally lower in 2002 than in Soluble reactive phosphorus (SRP), the form of phosphorus most readily available for plant use, is usually found in higher concentrations in point sources, particularly municipal WWTP effluent, than in nonpoint sources. This fact, along with the tendency of phosphorus to associate with settlable solids resulted in SRP loads at stations downstream from the point sources that were less than the sum of the point sources. This made it impossible to estimate the nonpoint source contribution of SRP load at stations There was an 80% reduction in the SRP load from West Point Stevens between 2001 and Total phosphorus (TP) loading of Sougahatchee Creek was similar to trends observed for TN loading for 2001 and In 2001, with higher runoff and discharge conditions, TP annual loads increased in a downstream direction from 0.3 mt at station 4 to 33.3 mt at station 12. The nonpoint source contribution of TP increased progressively downstream and the point source contribution decreased. In 2002, annual TP loads increased from station 4 (0.2 mt) to station 10 (17.5 mt) and then decreased at stations 11 and 12 (16.0 mt). Nonpoint source contributions at the two downstream stations (11 and 12) were higher in 2001 than in TP load in the West Point Stevens waste effluent was 36% lower in 2002 than in In 2001 and 2002 West Point Stevens contributed most of the sodium load measured in the creek. In addition, sodium load increased 3.5 times in the Opelika WWTP effluent iv

5 between 2001 and Potassium loading at mainstream sampling stations increased progressively in a downstream direction during both 2001 and Nonpoint source contributions of potassium ranged between 50 and 67% in 2001 and between 15 and 46% in Concentrations of total suspended solids (TSS) in point source effluents were relatively low and typically contributed <10% of the TSS load at any of the mainstream sampling stations in 2001 and TSS loads increased progressively in a downstream direction with a substantial increase between stations 9 and 10 during both years. Ropes Creek enters Sougahatchee Creek just upstream of station 10 and appeared to be the cause of this increased TSS load. The likely source of this sediment was extensive forest clear cuts on the watershed of Ropes Creek and its tributaries. Point source TSS concentrations were lower in 2002 than in 2001, led by a 55% reduction in West Point Stevens effluent. Tributary Concentrations. Throughout the Sougahatchee Basin, concentrations of variables measured in the tributary streams were probably of nonpoint source origins, except for Pepperell Branch. Pepperell Branch was sampled at two locations, station 15 upstream of the West Point Stevens wastewater outfall (station 1) and station 16 downstream of the outfall. This point source effluent was relatively high in some of the water quality variables measured and therefore, station 16 frequently had the highest concentrations of all of the tributary streams. Urban streams in Opelika and Auburn had the highest total nitrogen concentrations, while stations in the downstream tributaries generally had lower concentrations. Also, higher phosphorus concentrations were found in streams draining urban areas than in those draining forested and agricultural watersheds. Station 14, Rocky Brook Creek (Opelika City Park) had particularly high SRP concentrations. Even though annual mean stream discharge measured at the USGS gage was over twice as high in 2001 compared to 2002, mean annual total suspended solids (TSS) concentrations were higher in 2002 at almost every tributary station than levels measured in This difference probably reflected, in part, the fact that during 2002 we managed to arrive at the streams during the rising limb of the hydrograph more often than we did during Also, maximum TSS concentrations were much higher in 2002 than in Each year of the study highest TSS concentrations were found in urban tributaries in Opelika and Auburn probably as a consequence of development on the watersheds. Tributary Loading. Except for Pepperell Branch, loading to the tributary streams in this study was apparently related primarily to nonpoint sources on the watersheds. Also, for all constituents, mean annual loading was higher in 2001 than in 2002 because of the greater mean daily discharge that occurred that year. In 2001, Pepperell Branch (station 16) had the highest mean annual total nitrogen (TN) loading followed by Loblockee Creek. In 2002, Loblockee Creek had the highest TN loading followed by Pepperell Branch at station 16. The high loading measured at station 21 in Loblockee Creek was directly related to the fact that discharge in this stream was higher than that measured in any of the other tributary stations. Stations 14 and 20 also had major contributions to TN loading both years. Mean annual soluble reactive phosphorus loads were higher in Pepperell Branch (station 16) during both years than in all other tributaries. Total phosphorus load was higher in Pepperell Branch (station 16) than in Loblockee Creek (stations 20 and 21) in 2001 but v

6 Loblockee Creek (stations 20 and 21) had the highest load in Loblockee Creek (stations 20 plus station 21) led all tributaries in mean annual total suspended solids loads during both years. Cane Creek (station 22) was the second leading contributor of TSS to Sougahatchee Creek among the tributaries sampled. Forest clear-cuts on the Loblockee and Cane Creek watersheds apparently represented a major nonpoint source contribution of sediment to these streams. In addition, TSS loading from the Loblockee Creek watershed was the primary reason for the high total phosphorus loading. Much of this phosphorus enters streams bound to sediment particles. Among the smaller streams, station 17, draining the lower Fisheries Station, and station 14, Rocky Brook Creek (Opelika City Park), were also consistently high nonpoint source contributors of TSS to Sougahatchee Creek. Our data clearly showed that concentrations of TSS throughout Sougahatchee Creek, and TP and TN in those downstream areas dominated by nonpoint sources, increased in concentration as rainfall and stream discharge increased. In addition, we were unable to consistently sample streams on the rising limb of the hydrograph following a rainfall pulse. Our data suggested that this likely resulted in additional underestimation of nonpoint source contributions to Sougahatchee Creek. Thus, the unusually dry conditions that existed during 2001 and 2002 resulted in a conservative estimate of annual nonpoint source loading to Sougahatchee Creek. Physical Habitat The physical assessment of each station revealed three sites that were suboptimal when compared to the reference stream. One was the 2 nd -order site, station 18; one was the 3 rd -order site, station 15; and one was the 4 th -order site, station 6. All of these sites exhibited heavy sediment deposition primarily by runoff from the urban subwatersheds. The sedimentation reduced habitat diversity in the streams and provided less cover for macroinvertebrate colonization. All other stations were ranked as optimal, meaning that healthy aquatic communities would be expected at these locations. Biological Assessment During 2001 a total of 193 taxa were collected from all stations while during 2002 the number of taxa totaled 190. The term taxa refers to different species of macroinvertebrates collected in our samples, although not all are identified to the species level. A taxon is the lowest level to which an organism could be identified. Each year the fauna was remarkably similar with aquatic insects comprising 88 to 90% of the macroinvertebrates collected. The remainder of the fauna consisted mostly of oligochaetes, crayfish, snails, mussels, amphipods and isopods. The dipteran Family Chironomidae (midges) was the most diverse group of macroinvertebrates each year comprising 34 to 36 % of the total number of taxa. Of the six metrics calculated from macroinvertebrate samples we found three to be the most useful in explaining changes in these communities in the Sougahatchee Basin. The three metrics were the total number of species (taxa richness), the EPT Index (the number of insect taxa in the Ephemeroptera, Plecoptera and Trichoptera that are more sensitive to pollution), and the Hilsenhoff Biotic Index (HBI). Typically in unpolluted streams taxa richness and the EPT Index are relatively high compared to impacted streams. However, the method of calculating the HBI is such that the higher this value, the more polluted the stream. vi

7 Second-Order Streams. The biological condition of second-order sites did not compare favorably with the reference stream on any date except for stations 20 and 24. On 3 of the 4 sample dates macroinvertebrate communities at station 24 were ranked as nonimpaired, but for station 20 only one date was nonimpaired. Higher nitrate concentrations and sedimentation probably contributed to the impairment measured at these 2 nd -order sites. Third-Order Streams. The biological condition of third-order sites in Pepperell Branch at stations 15 and 16 was moderately impaired on all dates. Concentrations of nitrogen and phosphorus at station 16, below the West Point Stevens outfall, greatly exceeded those measured upstream from the outfall and apparently represented a major point source impact to macroinvertebrate communities at station 16. However, the EPT and HBI metrics reflected poor stream quality at both stations 15 and 16. In addition to the high nutrient concentrations at station 16, heavy sediment deposition was evident at station 15 and greatly reduced habitat available for colonization by macroinvertebrates. Fourth-Order Streams. The biological condition for station 3A, just downstream from the Northside WWTP, was non-impaired in October 2001, but on each of the other sample dates this station ranked as slight to moderately impaired. On all dates the physical habitat at station 3A exhibited a high degree of sedimentation. Rocky substrates were covered with sediment at this site although woody debris, root masses, and gravel provided some stable habitat for macroinvertebrate colonization. Nutrient concentrations (both nitrogen and phosphorus) reaching station 3A would have been high because of the point source discharges upstream and partially explain why macroinvertebrate communities were impaired on 3 of 4 dates. The biological condition at station 6 was slight to moderately impaired on each date. Nutrient concentrations at this station were high from upstream point sources and helped explain the poor biological condition observed. At station 7 the biological condition was non-impaired during October 2001, but for each of the remaining dates this site was slightly impaired. For stations 6 and 7 the high nutrient concentrations plus heavy sediment deposition appeared to represent the primary factors impacting macroinvertebrate communities. The metrics calculated for macroinvertebrate communities at stations 8, 9, 11 and 12 compared favorably with those from the reference stream except for the November date in In fact, occasionally metrics from these stations exceeded those measured in the reference stream (Hatchet Creek). Except for the November 2001 date the biological condition measured at stations 8, 9, 11, and 12 was assessed as non-impaired. During November 2001, each of these stations was found to be slightly impaired. In assessing the biological condition of sites in the Sougahatchee Basin, nutrient concentrations existing at the stations apparently had a greater impact on macroinvertebrate communities than the loading. Loading rates are closely related to discharge, which increased as catchment area (or impervious land cover) increased. Mainstream stations in vii

8 Sougahatchee Creek showed definite improvement in biological condition downstream beginning with station 8 near Loachapoka. Relatively high concentrations of nutrients, CBOD 5 and minerals from point source inputs were critical in the Sougahatchee Basin because stations closest to the effluent discharge from the waste treatment facilities of Opelika, Auburn and West Point Stevens consistently exhibited the most biological impairment. Additionally, nonpoint source runoff of sediment to the stream contributed to habitat degradation that was part of the reason for the impaired macroinvertebrate communities found at stations 6, 7 and 3A. Landscape Analysis Changes in land cover occurred in the Sougahatchee Creek Basin during an 8-year period from 1993 to In 2001, the basin as a whole had about 67% forest, 12% clearcut, 7% urban, 7% tilled agriculture, 5% grassland and 1% each water and wetland. Forest and grassland coverage declined 17.3% and 1.1%, respectively, from 1993 to Increases in coverage during this time span were clearcut/barren (10.4%), urban/ suburban (4.2%) and tilled agriculture (3.4%). These trends document the effects of increased human activities in the basin as a result of population growth. Lee County is one of the fastestgrowing counties in Alabama. Landscape Effects on Stream Loading and Biological Condition Based on regression analyses, loading of CBOD 5, SRP, TN and TP in both 2001 and 2002 increased significantly (positive regression slope; p<0.05) as urban coverage of the basin increased. Loading of TSS was not significantly (p>0.05) related to urban coverage. On the other hand, increases in forest cover in the basin resulted in decreased loading of CBOD 5, SRP, TN and TP in Sougahatchee Creek. TSS loading was not significantly (p>0.05) related to forest coverage of the basin. The only land cover that was significantly (p<0.05) related to TSS loading was grassland. Increases in grassland coverage would be expected to decrease TSS loading of the mainstream Sougahatchee Creek. However, grassland cover was less than 6% of the Sougahatchee Creek Basin in For the twelve tributary streams, variations in urban, forest and grassland coverage of the watersheds did not significantly (p>0.05) influence loading of CBOD 5, SRP, TN, TP or TSS during either year. One possible explanation for this disconnect between land cover and tributary stream loading was the lack of rainfall during 2001 and 2002 needed to mobilize sediment particles and nutrients from the watersheds to the streams. Based on the regression equations for the mainstream Sougahatchee Creek Basin, increases in forestland cover and decreases in urban land cover would improve water quality and reduce loading of nutrients and CBOD 5. However, trends in the basin are for increasing urban growth (4.2% increase from 1993 to 2001) and decreasing forest cover (17.3% decrease from 1993 to 2001). The biological assessment of streams in the Sougahatchee Basin established that several metrics exhibited significant correlations with landscape cover. For mainstream stations in 2001, a significant regression (p<0.05) between the Hilsenhoff Biotic Index (HBI) and land cover was found for all cover types. The HBI was positively correlated viii

9 (positive regression slope) with urban and grassland cover, but negatively correlated (negative regression slope) with forest cover. The HBI is a metric that relates nutrient enrichment to impacts on macroinvertebrate communities. Higher HBI values are associated with nutrient enrichment. The significant positive regression (p<0.05) between the HBI and urban cover in 2001 indicated high nutrient concentrations associated with these stations, thus impairment of the biota. The trend between the HBI and urban cover was positive in 2002 although not quite significant (p = 0.056). The significant negative correlation between the HBI and forest cover indicated watersheds with fewer nutrients, thus less impairment to the biological communities. In 2002 the HBI regression was significant (p<0.05) only for forest and grassland cover. However, the significant positive regression between forest cover and the EPT Index during 2002 was important. This finding meant that as forest cover declined on the watershed, the EPT taxa (aquatic insects that are more sensitive to pollutants than most other macroinvertebrates) also declined. Or conversely, the watersheds in the basin with greater forest cover had streams with higher numbers of EPT taxa. Additionally, analysis for 2001 showed a positive relationship between the number of EPT taxa and forest cover although the probability values, while not significant, were close to the p=0.05 level. Also, correlation analysis showed that the EPT taxa and urban cover in both years exhibited a negative relationship as expected, although the probability values were not significant. The regression analyses for the tributary streams provided more consistent relationships between metrics and land cover than those for mainstream sites. For example, we found significant regressions (p<0.05) between the EPT Index, HBI and Biological Condition score for urban and forest cover during both years of the study. This was also true for the taxa richness metric in Regressions between metrics and grassland cover were not significant (p>0.05) on any date. All of the tributary watersheds were smaller than those of the mainstream stations and for most land cover fell into one or two categories that dominated the watershed. Two land cover types, forest and urban, showed the strongest relationships with macroinvertebrate communities and stream health. Watersheds with the most urban land cover were found to have the poorest biological condition, while those watersheds with the most forest cover had the healthiest macroinvertebrate communities. In this study of nutrient and sediment loading in the Sougahatchee Basin, and impacts to the macroinvertebrate communities, results at both mainstream and tributary stations reflected similar relationships to land cover type. Urban and forest cover showed the strongest relationship with these variables, whether considering nutrients, sediments or macroinvertebrates. Also, the relationships were strongest among the tributary watersheds where one particular land cover dominated. Mainstream stations are influenced by a greater diversity of land cover types upstream of a given site, all of which are integrated into the quality of the stream environment as the water flows through Sougahatchee Creek into Yates Lake on the Tallapoosa River. ix

10 TABLE OF CONTENTS EXECUTIVE SUMMARY..ii INTRODUCTION... 2 RESEARCH METHODS... 4 WATER QUALITY AND STREAM LOADING... 4 BIOASSESSMENTS OF STREAM QUALITY... 9 LANDSCAPE ANALYSIS RESULTS AND DISCUSSION METEOROLOGICAL AND HYDROLOGICAL CONDITIONS ESTIMATED DISCHARGE OF UNGAGED STREAMS MAINSTREAM SOUGAHATCHEE CREEK SOUGAHATCHEE CREEK TRIBUTARY STREAMS PASSIVE SAMPLING OF STORM EVENTS LANDSCAPE ANALYSIS LANDSCAPE EFFECTS ON STREAM LOADING PHYSICAL HABITAT FEATURES OF STREAMS BIOLOGICAL ASSESSMENT LANDSCAPE EFFECTS ON MACROINVERTEBRATE COMMUNITIES LITERATURE CITED

11 INTRODUCTION Nutrient and sediment loading of surface waters from both point and nonpoint sources have been identified as major problems in the United States. Agricultural activities were reported to be responsible for more than 60% of surface water pollution in the United States (EPA 1990). The federal government has mandated the restoration of surface waters and assessment of nonpoint source pollution through the Clean Water Act. Surface water pollutants of special concern in Alabama were listed as nonpoint source pollution in the form of organic enrichment and siltation from poultry operations, forest clearcuts and livestock operations (ADEM 1994). In recent years, the importance of nonpoint sources of pollution has caused a shift in pollution abatement toward a watershed management approach (Browne 1981). Alabama is developing a watershed management strategy (McIndoe 1996). Tools that enhance water quality management at the watershed level include Geographic Information System (GIS) technology and remote sensing (RS). These tools are popular among private businesses, government organizations and researchers. GIS and RS incorporate diverse spatial attribute data into watershed analyses. The value of GIS/RS tools is in their ability to determine, map and quantify land use/land cover over large areas, and to relate these changes to data such as nutrient/sediment loading and biotic changes in streams. Recent studies of the Mobile River Basin revealed highest nutrient and suspended sediment concentrations were from watersheds with predominantly urban or agricultural land uses (McPherson et al. 2003). Total phosphorus concentrations exceeded the U.S. Environmental Protection Agency s (US EPA) recommended maximum concentration of 0.1 mg/l in 41% of all of the samples analyzed in However, compared to other 2

12 sites across the United States where similar studies were conducted, sites in the Mobile River Basin were generally in the lower to middle percentile for most nutrient and sediment variables. Unfortunately, no sampling stations were established in the Tallapoosa River portion of the Mobile Basin. Over the last decade, citizens in East Central Alabama have become concerned about the Tallapoosa River Basin as a consequence of excess nutrients and sediments entering the river from the watershed. For example, the upper portion of Lake Martin has undergone a rapid and dramatic increase in trophic status, from mesotrophic to eutrophic (Deutsch et al. 2000). Concerns also exist in the lower Tallapoosa River Basin with the Sougahatchee Creek embayment in Yates Lake. The Alabama Department of Environmental Management (ADEM) added this embayment to its 303(d) List of impaired state waters in 1998 (ADEM 1998). Pepperell Branch, a tributary to Sougahatchee Creek also was added to the 303(d) List by the US EPA (1999). The listings are a consequence of high nitrogen and phosphorus levels in the two systems (ADEM 1997). The high nutrient levels entering the Sougahatchee embayment have resulted in elevated chlorophyll a values caused by the large standing crops of algae that develop in the lake. High levels of sedimentation have also been recorded for Pepperell Branch and Sougahatchee Creek. Total maximum daily loads (TMDLs) are required for Sougahatchee Creek in order to correct these problems. In this study we conducted intensive instream measurements of nutrient and sediment loading at selected sites on Sougahatchee Creek. In addition, we sampled most of the major tributary streams flowing into Sougahatchee Creek. Similar data were collected for the three major point source discharges, including the cities of Opelika and Auburn and West Point Stevens. At selected stations on the tributary streams and the mainstream of 3

13 Sougahatchee Creek, bioassessments were conducted utilizing benthic macroinvertebrate communities as a measure of the biological condition of the streams. Bioassessments were also conducted in reference streams from the same ecoregion. Satellite imagery was examined to quantify land cover of the entire Sougahatchee Basin using GIS technology. We estimated the amounts of nitrogen, phosphorus and sediment entering Sougahatchee Creek; the proportion that originated from point sources versus the proportion that originated from nonpoint sources; where the loads originated (i.e. watersheds), and how they were impacting the upper, mid and lower reaches of the creek. Regressions between land cover, nutrient and sediment concentrations and loading and the biological health of the streams were examined. Data from this study can be used to develop watershed management plans that establish nutrient and sediment limits required to eventually remove Pepperell Branch and the Sougahatchee Creek embayment from the 303(d) List and protect the system during future development. RESEARCH METHODS Water Quality and Stream Loading The Sougahatchee Creek Basin and all reference streams utilized in this study lie within the Piedmont ecoregion. The Piedmont comprises a transitional area between the mostly mountainous Appalachians to the northeast and the relatively flat coastal plain to the southeast (Omernik 1987; Griffith et al. 2001). It is a complex mosaic of Precambrian and older Paleozoic metamorphic and igneous rocks with moderately dissected irregular plains with some hills. The soils are finer-textured and lower in organic matter and nutrients than Coastal Plain ecoregions. Rocky bottoms of boulder, cobble and gravel characterize streams 4

14 in this ecoregion although we have observed Piedmont streams that contain large quantities of sand and silt as a consequence of erosion from human activities on the watersheds. All stations within the Sougahatchee Creek Basin appear in Table 1 and Figure 1. Reference sites included 2 nd, 3 rd and 4 th -order streams used to compare with the study stations. Water quality sampling began in January 2001 and continued through December During months normally receiving more rainfall (November through April) all sites were sampled twice each month except for the 2 nd -order and 4 th -order reference streams. These two reference streams were sampled only once each year to characterize the water quality. During drier months (May through October) samples were collected once monthly. One sample following heavy rains was collected during both the wet and dry seasons for a total of 40 sampling occasions during the study. Variables measured along with the analytical method used appear in Tables 1 and 2. Stream and effluent samples were taken at mid-depth near mid-stream. Suspended sediment samples were collected with a US DH-59 depth-integrated suspended sediment sampler when stream velocity exceeded 2 ft/s (Glysson and Edwards 1988). At each site all variables were measured on each sampling occasion. The Auburn University Soils Laboratory analyzed metals in water samples. At ungaged locations, stream discharge was measured using the six-tenths method (Gordon et al. 1993). Instantaneous discharge was calculated by measuring flow velocity along a stream transect which was divided into equal intervals (1 or 2 m) at each site. Stream discharge (m 3 /sec) was estimated as the product of interval width (m) x depth (m) x velocity (m/sec), summed across the entire transect. Stream velocity was measured 5

15 Table 1. Sampling stations and variables measured within the Sougahatchee Creek Basin, N = nutrients; M = metals; S = sediments; D = discharge; B = benthos. Stations Description Variables Point Sources 1 West Point Stevens waste effluent N, M 2 Westside WWTP Opelika N, M 3 Northside WWTP Auburn N, M Nonpoint Sources Sougahatchee Creek 3A Sougahatchee Creek 300 m downstream of B Northside WWTP-Auburn 4 Sougahatchee Creek upstream of N, S, M, D, B Sougahatchee Lake at Lee Co. Rd Sougahatchee Lake spillway N S, M, D 6 S. Creek at Hwy 280 N, S, M, D, B 7 S. Creek at N. Donahue N, S, M, D, B 8 S. Creek at Lee Co. Rd. 188 N, S, M, D, B 9 S. Creek at Lee Co. Rd. 65 N, S, M, D, B 10 S. Creek at Lee Co. Rd. 217 N, S, M, D 11 S. Creek at Hays Mill Rd. N, S, M, D, B 12 S. Creek at Lovelady Rd. N, S, M, D, B Tributary Streams 13 Unnamed tributary upstream of N, S, M, D Sougahatchee Lake at Lee Co. Rd Opelika City Park Stream N, S, M, D 15 Pepperell Branch upstream of West N, S, M, D, B Point Stevens outfall 16 Pepperell Branch downstream of N, S, M, D, B West Point Stevens outfall 17 AU Fisheries Station Stream N, S, M, D 18 N. Auburn Stream N, S, M, D, B 19 Auburn University Club pond/stream N, S, M, D 20 Unnamed tributary to Loblockee at N, S, M, D, B Lee Co. Rd Loblockee Creek at Lee Co. Rd. 188 N, S, M, D, B 22 Cane Creek at Lee Co. Rd. 217 N, S, M, D, B 23 Sycamore Creek N, S, M, D, B 24 Buck Creek N, S, M, D, B Ref Unnamed tributary to Ropes (2 nd Order site) B Ref Hatchet Creek (4 th Order site) B 6

16 7

17 Table 2. Analytical methods used for measuring water quality variables and discharge in streams of the Sougahatchee Basin, Variable Method Reference Total suspended solids (TSS) Vacuum filtration APHA 1998 Total ammonia (NH 3 -N) Phenate method APHA 1998 Nitrite (NO 2 -N) Diazotizing method APHA 1998 Nitrate (NO 3 -N) Cadmium reduction method APHA 1998 Total nitrogen (TN) Persulfate digestion, then APHA 1998 UV spectrophotometry Total phosphorus (TP) Persulfate digestion, APHA 1998 Ascorbic acid method Soluble reactive Ascorbic acid method APHA 1998 Phosphorus (SRP) Metals ICAP APHA 1998 CBOD 5 5-day CBOD test APHA 1998 Stream discharge Six-tenths method Gordon et al using a Marsh-McBirney Model 2000 Flo-Mate portable flow meter. Daily discharge data for Sougahatchee Creek were obtained from the USGS gage at station 8 (Lee Co. Rd. 188). Using the gage data and site-specific instantaneous discharge data, mean daily discharge at ungaged sites was estimated using regression analyses (Thomas 1967). Annual total loading estimates (metric tons/yr) of nutrients, total suspended solids and metals for each sampling location were calculated using FLUX (Walker 1996). FLUX is a computer program that utilizes water sample constituent concentration values, instantaneous stream discharge and mean daily discharge values in estimating nutrient and sediment loading. The calculation of nonpoint source loading was then determined by subtracting measured point source loading from total loading calculated using FLUX. 8

18 Passive water samplers were attached to bridge piers at two sampling locations to collect rising water following rainfall/runoff events. One sampling station (station 6) was located on Sougahatchee Creek downstream from the city of Opelika and the other station (station 7) was downstream from both Auburn and Opelika (Figure 1). Samples were retrieved as soon as water levels receded enough to reach the samplers. The water samples were always analyzed for total suspended solids; total nitrogen and total phosphorus were measured in those samples that were retrieved within 24 hours. Each quarter, a separate sample was collected in this manner and submitted to the Alabama Department of Agriculture and Industries, Pesticide Residue Laboratory in Auburn, Alabama for analysis of the insecticide diazinon. Bioassessments of Stream Quality Physical Habitat Assessment. A visual assessment of physical habitat features in the stream was conducted at stations where macroinvertebrate communities were sampled. Habitat variables were evaluated according to methods described in the US EPA rapid bioassessment protocol (Plafkin et al. 1989; Barbour et at. 1999). A copy of the data sheet used in the field assessment of physical habitat is shown in Appendix I. Habitat in each reach was scored on the basis of instream variables such as bottom substrate and available cover, channel morphology, bank features and streamside (riparian) vegetation (Table 3). Scores for each variable were summed and then compared to the reference stream using the scale shown in Table 4. If physical variables are dramatically altered by contaminants then the aquatic communities colonizing this habitat often change, usually in detrimental ways. Physical habitat variables were evaluated on only one date each year because little change was expected between the sample dates. 9

19 Table 3. Scoring criteria 1 used to visually estimate physical habitat quality at stations in the Sougahatchee Creek Basin during the fall A total score is tabulated and compared with the reference stream. Range of Condition Criterion Optimal Suboptimal Marginal Poor Primary Substrate and Instream Cover 1. Bottom substrate/ Available cover Embeddedness of substrate material Velocity/Depth Regime Secondary Channel Morphology 4. Sediment deposition Channel flow status Channel alteration Frequency of riffles ( or bends) Tertiary Riparian and Bank Structure 7. Bank Stability (each bank) Bank Vegetation (each bank) Streamside cover (each bank) The field data sheet used for this assessment is shown in Appendix 1. Table 4. Assessment used to evaluate physical habitat quality at all stations in the Sougahatchee Creek Basin during the fall Assessment Category Percent Comparability Comparable to Reference 90 % Supporting % Partially Supporting 60-73% Non-Supporting 58 % 1 Potential to support an expected assemblage of aquatic invertebrates. 10

20 Bioassessment of Macroinvertebrate Communities. At each station benthic macroinvertebrates (benthos) were sampled as a measure of stream quality, or the biological health of the stream. Benthic macroinvertebrates are bottom-dwelling invertebrates that are ubiquitous in streams. Many are larval forms of aquatic insects and most macroinvertebrates are food items for stream fishes. Benthic macroinvertebrates live on the stream bottom among rocks, logs, sediment, leaf packs and aquatic vegetation. Life cycles of benthic animals range from a few days to several years, but many are relatively long, which allows an examination of periodic changes to these communities caused by perturbations. While some movement is typical among macroinvertebrates, the relatively sedentary nature of these animals allows effective spatial analyses of pollutant or disturbance effects. Consequently benthic macroinvertebrates act as continuous monitors of the water they inhabit. This enables long-term analysis of both regular and intermittent discharges, variable concentrations of pollutants, single or multiple pollutants, and even synergistic or antagonistic effects (Rosenberg and Resh 1984). Methods followed rapid bioassessment protocols established by the US EPA (Plafkin et al. 1989; Barbour et al. 1999). Rapid bioassessment protocols involve comparison of metrics from streams suspected of impacts with those having minimal disturbance (reference streams). Reference streams of about the same size were selected from the same Piedmont ecoregion as Sougahatchee Creek and its tributaries. The 2 nd -order reference stream was an unnamed tributary to Ropes Creek in the Sougahatchee Basin. The 3 rd -order reference stream was Cane Creek (Station 22) and the 4 th -order reference stream was Hatchet Creek in Coosa County. 11

21 Benthic macroinvertebrates were sampled twice at low flows between September and November during both 2001 and Reference streams were sampled at the same time. Sampling utilized D-frame aquatic dipnets and a 1-m kicknet. At each site we sampled all available microhabitats that macroinvertebrates might colonize (e.g., rocks, logs, gravel, sand, leaf packs, root masses along undercut banks, and aquatic vegetation). Usually the net was placed just downstream of the microhabitat and the substrate disturbed so that the current washed organisms into the net. Smaller macroinvertebrates that reside in sand and those living on large rocks and logs were sampled with a 240-µm mesh net. The D- frame net and kicknet had a 1,000-µm-mesh size. Macroinvertebrates were sorted in the field and preserved in 80% ethanol. Sample size usually ranged between two and four hundred organisms. The purpose of the sorting was to obtain the most diverse sample possible from that site. If a particular organism could be reliably identified during the field sorting, we collected only about 20 to 30 specimens of that animal. In the laboratory macroinvertebrates were identified and counted. From each sample a total of six biocriteria (metrics) was calculated. Examples of the metrics calculated were taxa richness, the Ephemeroptera-Plecoptera-Trichoptera (EPT) Index, and the Hilsenhoff Biotic Index (HBI). Taxa richness and the EPT Index are good measures of the biodiversity of the sample. EPT taxa include macroinvertebrates that are usually the most sensitive to pollution of all the varied benthic organisms that inhabit the stream. The HBI is a biotic index that is calculated based on the pollution tolerance value for each organism in the sample. Data analysis was performed according to modified procedures following the US EPA Rapid Bioassessment Protocol III (Plafkin et al. 1989). 12

22 Biological condition, or stream health, at each station was determined using a multimetric approach. All of the metrics and scoring criteria used in this study are shown in Table 5. Once the metrics were calculated they were scored based on their comparability to the reference metric. For each station a total score was then calculated as the sum of all individual metrics and compared to the reference score to determine biological condition as shown in Table 6. Webber and Blevins (2000) provided a complete description of each metric. Table 5. Biocriteria (metrics) and biological scores for evaluating stream quality at sites in the Sougahatchee Creek Basin during the fall, Score Metric Taxa Richness 1 >80% 60-80% 40-60% <40% 2. EPT Index 1 >90% 80-90% 70-80% <70% 3. HBI 2 >85% 70-85% 50-70% <50% 4. Scrapers/Filterers Ratio 1 >50% 35-50% 20-35% <20% 5. EPT/Chironomidae Ratio 1 >75% 50-75% 25-50% <25% 6. Shredders/ Total Abund. Ratio 1 >50% 35-50% 20-35% <20% 1 Score is based on the ratio of the study site to reference site x Score is based on the ratio of the reference site to study site x

23 Table 6. Percent comparability values required for assigning biological health of sites in the Sougahatchee Creek Basin, fall Percent Comparability to Reference Score 1 Biological Condition Attributes > 83% Nonimpaired Comparable to best situation expected within an ecoregion. Balanced trophic structure. Optimum community structure (composition and dominance) for stream size and habitat quality % Slightly Impaired Community structure less than expected. Composition (taxa richness) lower than expected due to loss of some intolerant forms. Percent contribution of some tolerant forms increases % Moderately Impaired Fewer species due to loss of most intolerant forms. Reduction in EPT Index. < 17% Severely Impaired Few species present. If high densities of organisms, then dominated by one or two taxa. 1 Percentage values intermediate to these ranges requires judgment of researcher for final placement using physical habitat and chemical data. 14

24 Landscape Analysis Land cover for 1993 was obtained in grid format from the US Geological Survey (USGS) National Land Cover Data (NLCD). The Multi-resolution Land Characterization (MRLC) Consortium developed this land cover data set using 1993 leaf-off and leaf-on Landsat thematic mapper (TM) satellite imagery. Pixel size for the Landsat TM imagery is 30 meters x 30 meters. The 1993 NLCD was classified into 23 land cover classes using a modified Anderson land cover classification system (see National Land Cover Characterization Home Page for details) (Anderson et al. 1976). For this study, the 23 land cover classes were consolidated into eight primary land cover classes: water, urban/suburban, clearcut/barren, quarry/gravel pits, forest, pasture/grassland, tilled agriculture and wetlands. The 2001 land cover was classified from 2001 leaf-off and leaf-on Landsat 7 TM images, path 19, and row 37. The Landsat images used in the classification were scenes acquired on 3/7/2001, 5/18/2001 and 10/1/2001. Each scene was classified into a 60-class grid using the ISODATA algorithm in Erdas Imagine (version 8.6) to conduct an unsupervised classification. The 60 classes were then consolidated into eight classes to match the 1993 consolidated classes. Identification and consolidation of the 60 classes was achieved through the use of color infrared Landsat imagery, 2001 aerial photography (obtained from the USDA Farm Service Agency, 600 South 7 th Street, Opelika, AL 36801), and site visits. The pasture/grassland class is usually composed of typical grassed livestock pastures and meadows, but also includes other grasslands such as lawns, golf courses and grassed areas on roadsides. The clearcut/barren class is usually composed of silviculture clearcuts 15

25 and cleared construction sites, though some pixels can be confused with tilled agricultural land. Stream watersheds were delineated using Arc Hydro Tools (version 1.0) to process digital elevation model (DEM) data obtained from the USGS National Elevation Database (see USGS National Elevation Dataset Home Page ) and USGS National Hydrography Dataset (NHD) vector files of the stream network (see USGS National Hydrography Dataset Home Page for details). Land cover within the Sougahatchee watershed and within watersheds was computed by subsetting the eight-class land cover grid using the whole watershed as well as watershed polygons (boundaries) derived in Arc Hydro. Percent of the entire Sougahatchee watershed was computed for each of the eight classes from the 1993 and the 2001 land cover grids. Percent change from 1993 to 2001 was then computed by subtraction for each land cover class. Note that in comparing 1993 to 2001, the 1993 NLCD land cover classification was done for the entire United States. This classification, under the direction of the MRLC, was not ground-truthed as thoroughly as the 2001 classification (which covered only the Sougahatchee and Chewacla Creek watersheds, thus ground-truthing was more thorough). Percent of watersheds was also computed for each of the eight classes from the 2001 land cover grid. Annual loads of select variables (SRP, TP, TN, TSS, CBOD 5 ) (Table 2) were converted to metric tons/km 2 yr -1 (annual load per unit watershed area) and then regressed against the various land cover classes using SAS statistical software (SAS 1996). Tributary stream watersheds were regressed separately from Sougahatchee Creek mainstream 16

26 watersheds. Biological metrics derived from the health of the macroinvertebrate community were also regressed against watershed land cover. RESULTS AND DISCUSSION Meteorological and Hydrological Conditions Meteorological conditions affect water quantity and water quality of streams and lakes. Stream discharge is directly related to the amount of precipitation on the watershed and precipitation also influences loading of dissolved and suspended substances from the watershed into the stream. Both years of this study, 2001 and 2002, were warmer and dryer than normal (Table 7). Annual rainfall measured at Auburn, Alabama was similar for the 2 years, cm in 2001 and cm in 2002, but distribution of the rainfall within years was quite different. In 2001, rainfall was unusually heavy in March, April and June followed by drought conditions from July through December (Table 7). In 2002, rainfall was distributed more evenly across months. The heavy spring rains of 2001 resulted in elevated mean daily discharge of Sougahatchee Creek at the gage station in March and April (Table 7). This spring flooding more than doubled the annual mean daily discharge of Sougahatchee Creek in 2001 (98.2 cfs) compared to 2002 (43.9 cfs). Attempts to measure nutrient and sediment loading of Sougahatchee Creek under drought conditions likely underestimated the contribution of nonpoint sources of pollution to the total load. The cities of Auburn and Opelika transfer water from other basins (Uphapee Creek and Chattahoochee River) via public water supply systems that ultimately augment the flow of Sougahatchee Creek. At times during the 2000 drought, stream discharge decreased 17

27 Table 7. Monthly and annual means for select meteorological variables and Sougahatchee Creek discharge measured during 2001 and Year Month Temp 1 ºC DFN* Rainfall 1 (cm) DFN* (cm) Mean Daily 1 Solar Radiation (Langleys) Mean Daily 2 Discharge (cfs) 2001 Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Annual Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Annual *Deviation from normal. 1 Auburn-AWIS (Agricultural Weather Information Service) 2 USGS-Gage Data (Sougahatchee Creek-Gage # ) 18

28 progressively downstream all the way to station 12 near Reeltown (unpublished data). The constant discharge of about 9.0 cfs of treated effluent in the Auburn/Opelika area provides a stable, dependable source of water to the creek when other sources such as rainfall runoff and groundwater are deficient. Estimated Discharge of Ungaged Streams To test the accuracy of estimating mean daily discharge of ungaged streams using the regression approach referred to in the Research Methods, we compared the estimated discharges of each stream (at sampling stations) per unit area of watershed with the measured discharge per unit area at the gaged station on Sougahatchee Creek (station 8 near Loachapoka). At 12 of the 18 sampling stations included in this analysis, the estimated discharge per unit area was within 25% of the gaged discharge per unit area (Table 8). At stations 6 (US Highway 280) and 7 (North Donahue) located in the urban reaches of Sougahatchee Creek, estimated discharge per unit area was 30% and 28%, respectively, higher than the measured discharge per unit area at the gaged station. This increase in discharge was likely a result of the abundance of impervious surfaces associated with urban areas upstream of stations 6 and 7 (Figure 1). At 4 of the 18 sampling stations (14, 15, 16 and 18) the estimated discharge per unit area was over 30% lower than the gaged discharge per unit area. These urban sites might have been expected to yield more runoff per unit area because of impervious urban land surfaces. One possible explanation was that storm sewers capture runoff and transport it out of the watershed, or within the watershed, but downstream of our sampling station. Another possibility was that a portion of the flow entered the groundwater and was lost, either temporarily or permanently, from the surface 19

29 20

30 waters of the stream. Only one tributary to Sougahatchee Creek, Pepperell Branch, had more than one sampling station (stations 15 and 16). Even though the area of the watershed increased by more than 50% (2.5 mi 2 ) from station 15 to station 16, the instantaneous discharge at station 16 measured during the 41 sampling trips was equal to or less than, the discharge at station 15, 32% of the time (Table 9). Discharge at station 15 was added to the discharge of treated waste effluent ( cfs) from the West Point Stevens plant at station 1, which entered Pepperell Branch between stations 15 and 16 (Figure 1). Most of the discharge deficits occurred under low-flow conditions. Mainstream Sougahatchee Creek Constituent Concentrations. The headwaters of Sougahatchee Creek are heavily influenced by activities and development in the urban areas of both Auburn and Opelika, Alabama (Figure 1). Treated effluent from two municipal wastewater treatment plants (WWTP), Opelika Westside (3.1 mgd) and Auburn Northside (1.4 mgd), and one industrial (textile) facility, West Point Stevens (1.3 mgd) enters Sougahatchee Creek in the upstream one-third of the basin. Waste from these facilities along with urban drainage and stormwater runoff from the cities of Auburn and Opelika cause an upstream peak in nutrient concentration. All nitrogen species peaked at mainstream station 6 (US Highway 280 bridge) and declined steadily downstream to station 12 near Reeltown, Alabama both years (Figure 2 and Appendix II, Tables 1 and 3). Carbonaceous biochemical oxygen demand (CBOD 5 ) followed a similar pattern (Figure 3). Soluble reactive phosphorus and total phosphorus peaked at the first mainstream sampling location (station 8) downstream from the Auburn Northside WWTP (station 3) where the mean phosphorus concentration in the effluent was usually considerably higher than that measured in other waste effluents 21

31 Table 9. Instantaneous discharge (cfs) for Pepperell Branch stations 15 and 16, West Point Stevens (WPS) effluent discharge, and the combined discharge of station 15 and the West Point Stevens effluent on 41 dates in 2001 and DATE St 15 WPS WPS + 15 St 16 01/16/ /31/ /06/ /22/ /05/ /14/ /21/ /11/ /25/ /30/ /20/ /31/ /30/01 * /27/ /18/01 * /07/ /27/ /05/01 * /12/01 * /10/02 * /31/02 * /06/ /14/ /20/ /06/ /12/ /21/ /04/ /17/02 * /08/02 * /05/02 * /18/02 * /29/02 * /19/ /04/ /16/ /05/02 * /11/ /21/ /05/ /18/02 * * Asterisk denotes dates that instantaneous discharge at station 16 was equal to or less than station 15 + WPS effluent.

32 23

33 24

34 25

35 26

36 (Figures 4 and 5 and Appendix II, Tables 1 and 3). From station 8, phosphorus concentrations declined downstream to station 12. Some minerals, like the total nitrogen compounds, peaked at station 6 then declined progressively downstream to station 12. Calcium, sodium, potassium, magnesium and boron followed this pattern (Figures 6 and 7 and Appendix II, Tables 2 and 4). These elements were in relatively high concentration in one or more of the point source effluents. Other minerals, like barium, aluminum, iron and manganese, showed no clear concentration pattern from headwaters to the mouth of Sougahatchee Creek (Figure 8 and Appendix II, Tables 2 and 4). In fact, iron and manganese were consistently higher in two tributaries (stations 4 and 13) to Sougahatchee Lake than at any other location (Figure 8 and Appendix II, Tables 2 and 4). These two streams drained a series of beaver ponds and it is possible that microbial reduction of iron (Fe III) and manganese (Mn IV) compounds caused the elevated concentrations of these metals in the water. These two sampling locations had mean dissolved oxygen (DO) concentrations and ranges that were lower than all other sites (Figure 9 and Appendix II, Tables 1 and 3). No other mainstream sites had DO concentrations lower than 5.0 mg/l (Appendix II, Tables 1 and 3). Annual mean concentrations of total suspended solids (TSS) in Sougahatchee Creek were variable both years with highest concentrations at station 10 in 2001 and station 8 in 2002 (Figure 10). Variation in TSS concentration was pronounced both within a station (as indicated by error bars in Figure 10) and between stations among years. WWTP s contributed relatively little sediment to Sougahatchee Creek. 27

37 28

38 29

39 30

40 31

41 32

42 Constituent Loading. When total load was higher than point source load at a station, estimates of mean annual point and nonpoint source contributions to total load were calculated by subtracting the point source load (measured at the point source) from total load (measured at the stream sampling location). This approach assumes that all of a constituent entering a stream from a source travels all the way downstream to the mouth, in this case station 12. This probably never occurs. Greatest losses of constituents occur during overbank flooding events when materials settle out of the water column and onto the flood plain and are lost to the stream indefinitely. Smaller losses occur when constituents (e.g. nutrients) are taken up by biota that then leave the ecosystem (e.g. emerging insects). CBOD 5 is reduced through oxidative processes within the stream. Because 2001 and 2002 were drought years, a minimum of overbank flooding occurred. However, these unquantified losses of constituents may result in an over estimation of point source contributions at downstream locations. Constituent loading was higher in 2001 than in 2002 because of the greater mean daily discharge that occurred that year (Table 7). In 2001, total nitrogen (TN) load increased in a downstream direction from less than 7.0 metric tons (mt) at station 4 to more than 233 mt at station 12 (Figure 11 and Appendix III, Table 1). The nonpoint source contribution of TN increased progressively downstream and the point source contribution decreased (Figure 11). In 2002, with lower hydraulic discharge, TN loading increased from less than 6 mt at station 4 to 130 mt at station 10 and then decreased downstream to 120 mt at station 12 (Figure 11 and Appendix III, Table 3). Nonpoint source TN contributions were generally lower in 2002 than in

43 34

44 The increased runoff and discharge that occurred in 2001 resulted in much higher nonpoint source contributions to the carbonaceous biochemical oxygen demand (CBOD 5 ) load than occurred in 2002 (Figure 12 and Appendix III, Tables 2 and 4). In 2001, nonpoint sources of CBOD 5 contributed well over one-half of the load at all mainstream stations. Improvements in waste treatment at West Point Stevens (Personal Communication, Eddie Lanier), resulted in a 53% reduction in CBOD 5 loading at station 1 between 2001 and 2002 (Figures 3 and 12). Soluble reactive phosphorus (SRP), the form of phosphorus most readily available for plant use, is usually found in higher concentrations in point sources, particularly municipal WWTP effluent, than in nonpoint sources (Figure 4). This fact, along with the tendency of phosphorus to associate with settlable solids resulted in SRP loads at stations downstream from the point sources that were less than the sum of the point sources (Figure 13 and Appendix III, Tables 1 and 3). This made it impossible to estimate the nonpoint source contribution of SRP load at stations There was an 80% reduction in the SRP load from West Point Stevens from 2001 to 2002 (Figures 4 and 13). Total phosphorus (TP) loading of Sougahatchee Creek was similar to trends observed for TN loading for 2001 and 2002 (Figure 14 and Appendix III, Tables 1 and 3). In 2001, with higher runoff and discharge conditions, TP annual loads increased in a downstream direction from 0.3 mt at station 4 to 33.3 mt at station 12. The nonpoint source contribution of TP increased progressively downstream and the point source contribution decreased (Figure 14). In 2002, annual TP loads increased from station 4 (0.2 mt) to station 10 (17.5 mt) and then decreased at stations 11 and 12 (16.0 mt). Nonpoint source contributions at the two downstream stations (11 and 12) were higher in 2001 than in

45 36

46 37

47 38

48 (Figure 14). TP load in the West Point Stevens waste effluent was 36% lower in 2002 than in 2001 (Figures 5 and 14). Sodium and potassium are two alkali metals that were found in relatively high concentrations in the West Point Stevens waste effluent (Figures 6 and 7). Common salts (e.g. chlorides) of these metals are extremely soluble and would tend to remain in solution throughout Sougahatchee Creek. In 2001 and 2002 West Point Stevens contributed most of the sodium load in the creek (Figure 15 and Appendix III, Tables 2 and 4). In 2001, with higher runoff and stream discharge, nonpoint sources contributed between 19 and 39% of the total load at mainstream stations 6 through 12 (Figure 15). With less runoff and discharge in 2002, nonpoint source contributions to sodium load varied between 3 and 19% at stations 7, 9, 11 and 12. Sodium load was less than the sum of the upstream point sources at stations 6, 8 and 10 (Figure 15). Sodium load increased 3.5 times in the Opelika WWTP effluent between 2001 and 2002 (Figures 6 and 15). Potassium is a macronutrient required by green plants for growth. Unlike other macronutrients (e.g. nitrogen and phosphorus) potassium is seldom limiting in aquatic systems because of its high water solubility and availability in decaying organic matter. Potassium loading at mainstream Sougahatchee Creek sampling stations increased progressively in a downstream direction during both 2001 and 2002 (Figure 16 and Appendix III, Tables 2 and 4). Nonpoint source contributions ranged between 50 and 67% in 2001 and between 15 and 46% in Concentrations of suspended solids in point source effluents were relatively low (Figure 10) and typically contributed <10% of the TSS load at any of the mainstream sampling stations in 2001 and 2002 (Figure 17 and Appendix III, Tables 1 and 3). 39

49 40

50 41

51 42

52 TSS loads at mainstream sampling stations did increase progressively in a downstream direction with a substantial increase between stations 9 and 10 during both years. Ropes Creek enters Sougahatchee Creek just upstream of station 10 and appears to be the cause of this increased TSS load. The likely source of this sediment was extensive forest clear cuts on the watershed of Ropes Creek and its tributaries (see Landscape Analysis Section, p 66). Point source TSS concentrations were lower in 2002 than in 2001, led by a 55% reduction in West Point Stevens effluent (Figure 10). Sougahatchee Creek Tributary Streams The 12 tributary streams that ranged in size from first order to third order varied greatly in estimated mean daily discharge (Figure 18 and Appendix IV, Table 1). The third order stream, Loblockee Creek (station 21), led all tributaries in estimated discharge. An unnamed second order tributary to Loblockee Creek (station 20) that drains the northern portion (upper station) of the Auburn University Fisheries Research Station had the second highest discharge during both years (Figure 18). Because these two streams converge prior to flowing into Sougahatchee Creek, Loblockee Creek provides more water (54%) to Sougahatchee Creek than all of the other 10 tributaries combined. Cane Creek (station 22) was second to Loblockee Creek with about 10% of the discharge from these 12 streams. The first order stream (station 17) draining the southern portion of the Auburn University Fisheries Research Station (lower station) had the lowest discharge of any of the 12 stations. Constituent Concentrations. Pepperell Branch was sampled at two locations, station 15 upstream of the West Point Stevens wastewater outfall (station 1) and station 16 downstream of the outfall. This point source effluent was relatively high in some of the water quality variables measured and therefore, station 16 frequently had the highest 43

53 44

54 concentrations of all of the tributary streams (Figure 19 and Appendix II, Tables 1 through 4). Total nitrogen concentrations were highest in urban streams (stations 14, 15 and 16) and generally decreased in downstream tributaries. With the exception of station 16, CBOD 5 ranged from 2.13 mg/l to 0.86 mg/l with lower concentrations in downstream tributaries (Figure 20; Appendix II, Tables 1 and 3). Higher phosphorus concentrations were found in streams draining urban areas (stations 14, 15 and 16) than in those draining forested and agricultural watersheds (Figures 21 and 22; Appendix II, Tables 1 and 3). The Opelika City Park stream (station 14) was particularly high. The main source of sodium and potassium in Sougahatchee Creek was the West Point Stevens waste effluent (station 1) that entered Pepperell Branch upstream of station 16 (Figures 23 and 24; Appendix II, Tables 2 and 4). Potassium was higher in tributaries draining the upper basin (stations 13-19) than in those draining the lower basin (stations 20-24). Other than station 16, stations 15 and 17 had higher maximum concentrations of sodium than all other streams (Appendix II, Tables 2 and 4). The stream draining from the southern portion (lower station) of the Auburn University Fisheries Research Station (station 17) may have experienced occasional peaks in sodium concentrations following application of salt (NaCl) to ponds for treatment of fish parasites or disease. Iron and manganese concentrations were higher at station 13, a tributary to Sougahatchee Lake, than in any other tributary stream (Figure 25; Appendix II, Tables 2 and 4). A discussion of the cause of the high iron and manganese concentrations at stations 4 (Figure 8; Appendix II, Tables 2 and 4) and 13 appears in the section on Mainstream Constituent Concentrations. As was the case at station 4 (Figure 9; Appendix II, Tables 2 and 4) mean annual dissolved oxygen concentrations were lower at station 13 than in all of the other tributary streams (Figure 26; Appendix II, Tables 1 and 3). Low DO 45

55 46

56 47

57 48

58 49

59 50

60 51

61 52

62 53

63 concentrations would favor microbial reduction of iron and manganese to a more soluble state. Unlike the mainstream Sougahatchee Creek stations, tributary streams frequently had minimum DO concentrations <5.0 mg/l (Appendix II, Tables 1 and 3; and Appendix IV, Table 1). Mean annual total suspended solids (TSS) concentrations were higher in 2002 at almost every station (except stations 13 and 19) than levels measured in 2001 (Figure 27; Appendix II, Tables 1 and 3). The fact that annual mean stream discharge measured at the USGS gage was over twice as high in 2001 compared to 2002 (Figure 18) indicated that we did a better job of targeting rain events in the tributary streams during 2002 than in Maximum TSS was much higher in 2002 than in 2001 (Appendix II, Tables 1 and 3). Highest TSS concentrations were found in upstream urban tributaries (Figure 27). Constituent Loading. The West Point Stevens wastewater outfall (station 1) enters Pepperell Branch between stations 15 and 16 and delivers relatively high concentrations of some constituents to the stream (Figure 28; Appendix III, Tables 1-4). In the bar graphs dealing with tributary loading an additional bar was added between stations 15 and 16 to represent the combined load of station 15 plus station 1, the West Point Stevens effluent (Figure 28). This point source addition resulted in Pepperell Branch being one of the leading contributors of soluble constituents to Sougahatchee Creek. Pepperell Branch was one of the four tributaries suspected of losing discharge (see Estimated Discharge of Ungaged Streams). In most cases constituent load was lower at station 16 than the combined loads of stations 15 and 1 (West Point Stevens) even though an additional 2.5 mi 2 of watershed drained into Pepperell Branch between stations 15 and 16. For all constituents, mean annual loading was higher in 2001 than in In 2001 station 16 had the highest mean annual total nitrogen (TN) loading followed by station 21 54

64 55

65 56

66 (Figure 28; Appendix III, Table 1). In 2002, station 21 had the highest TN loading followed by station 16. Stations 14 and 20 contributed significantly both years. Since station 20 is a tributary to Loblockee Creek (station 21), the sum of these two stations would be a conservative estimate of the Loblockee Creek total nitrogen contribution. Loblockee Creek (stations 20 and 21) was also the largest contributor of CBOD 5 load to Sougahatchee Creek followed by Pepperell Branch (station 16; Figure 29 and Appendix III, Tables 2 and 4). Mean annual soluble reactive phosphorus loads were higher in Pepperell Branch (station 16) during both years than in all other tributaries (Figure 30 and Appendix III, Tables 1 and 3). Total phosphorus load was higher in Pepperell Branch (station 16) than in Loblockee Creek (stations 20 and 21) in 2001 but Loblockee Creek (stations 20 and 21) had the highest load in 2002 (Figure 31and Appendix III, Tables 1 and 3). Sodium and potassium loads were higher in Pepperell Branch (station 16) both years than in any of the other tributaries (Figures 32 and 33; Appendix III, Tables 2 and 4). The West Point Stevens effluent was responsible for virtually all of the sodium and most of the potassium load at station 16 (Figures 23 and 24). Given the solubility of most sodium and potassium salts, it was difficult to understand the fate of the sodium and potassium loads measured in the West Point Stevens effluent from the time it enters Pepperell Branch just downstream of station 15 until it reaches station 16. This was particularly true in 2002 when about two-thirds of the sodium and potassium loads disappear in this 2.0-mile stream reach. Mean annual total suspended solids loads were higher in Loblockee Creek (stations 20 plus station 21) than in any of the other tributaries during both years (Figure 34; Appendix III, Tables 1 and 3). Cane Creek (station 22) was the second leading contributor of TSS to Sougahatchee Creek among the tributaries sampled (Figure 34; Appendix III, Tables 1 and 3). Loblockee 57

67 58

68 59

69 60

70 61

71 62

72 63

73 (stations 20 and 21) and Cane (station 22) creeks ranked first (54%) and second (10%) in estimated mean annual discharge for the 2 year period (Figure 18 and Appendix IV, Table 1). Among the smaller streams station 17, draining the lower Fisheries Station, and the Opelika City Park stream (station 14) were consistent contributors of TSS (Figure 34; Appendix III, Tables 1 and 3). Passive Sampling of Storm Events Following a rainfall event, runoff from the watershed will initially cause an increase in stream depth and flow velocity reaching a peak followed by a decline in depth and flow velocity to pre-rainfall conditions. A graphical plot of the changing hydraulic discharge of a stream during a rainfall event is referred to as a storm hydrograph. Estimates of constituent loading (e.g. TP or TSS) based on samples taken at different points (e.g. rising limb or falling limb) on the storm hydrograph can be quite variable (Hewlett 1982). Because of the number of stream sampling stations (21 stations) and the variation in stream order (four orders) among the streams, it was not possible to sample all stations at the same point on the storm hydrograph. Maximum loading of some constituents (e.g. TSS and TP) will usually occur on the rising limb of the storm hydrograph (Hewlett 1982). Passive water samplers that were placed at two locations on the mainstream of Sougahatchee Creek (stations 6 and 7) were designed to fill on the rising limb of the hydrograph during a storm event at depths of about 30 cm, 60 cm and 90 cm above base flow. Estimates of instantaneous concentrations and loads of TSS, TP and TN were calculated and compared to the range of instantaneous concentrations and loads measured at stations 6 and 7 during regular sampling at monthly or twice-monthly intervals. 64

74 There were about twice as many smaller storm events that filled the shallow (30 cm) samplers but not the mid-depth (60 cm) samplers and about twice as many events that filled the mid-depth samplers but not the deep (90 cm) samplers (Table 10). TSS concentrations at both stations 6 and 7 generally increased with stream discharge. TP concentrations did not change with increasing discharge except for the deep (90 cm) sampler at station 7. TN levels generally declined with increasing discharge (Table 10). At station 6 the instantaneous loads (TSS, TP and TN) captured with the passive samplers were always well within the range of instantaneous loads estimated during the regular sampling (Table 10). However, this only occurred because of unusually large loads that were measured during one of 51 regular sampling events at station 6. All of the other 50 regular sampling events resulted in loading estimates that were well below the maximum loads captured with the passive samplers. At station 7, maximum loads for all three variables estimated with the passive samplers were considerably higher than loads estimated during regular sampling. Sediment loads during storm events increased because of increasing TSS concentrations and increasing stream discharge. However, increases in TP and TN loads during storm events occurred primarily because of increases in stream discharge. Water collected with the passive samplers at stations 6 and 7 were analyzed for diazinon on eight dates in 2001 and Diazinon was not detected in any of the samples at detection limits that varied between 0.10 ppb and 1.2 ppb. Diazinon was reported to be the most frequently detected insecticide in surface water samples collected in streams draining urban areas of the Mobile River Basin. It was one of six insecticides that exceeded aquatic-life criteria (McPherson et al. 2003). 65

75 Table 10. Estimates of instantaneous concentrations and loads of total suspended solids (TSS), total phosphorus (TP) and total nitrogen (TN) based on water samples collected with passive samplers during storm events compared to the range of loads measured during regular sampling (monthly and twice monthly) at stations 6 and 7 in Sougahatchee Creek. Variable Station Depth* n Mean Discharge Passive Regular sampling load (cm) concentration (cfs) sampler load min max (mg/l) (mt/yr) (mt/yr) (mt/yr) TSS , , , , , , , , , , ,710 TP TN , , , ,684 *Depth above base flow. 66

76 Landscape Analysis A cursory examination of Figures 35 and 36 reveals changes in land cover that have occurred in the Sougahatchee Creek Basin during an 8-year period from 1993 to 2001, the first year of this study. In 2001, the basin as a whole had about 66% forest, 12% clearcut, 8% urban, 7% tilled agriculture, 6% grassland and 1% each water and wetland (Figure 36 and Table 11). The smaller tributary streams were much more variable with urban coverage ranging between 0.8% and 70.8% and forest coverage from 18.7% and 79.1% (Figure 37 and Table 11). For the entire Sougahatchee Basin, forest and grassland coverage declined 17.3% and 1.1%, respectively, from 1993 to 2001 (Table 12). Gaining coverage during this time span were clearcut/barren (10.4%), urban/suburban (4.2%) and tilled agriculture (3.4%). These trends document the effects of increased human activities in the basin as a result of population growth. Lee County is one of the fastest growing counties in Alabama. Landscape Effects on Stream Loading Regression analysis was used to relate watershed land cover to estimates of loading of nutrients, sediment and biochemical oxygen demand. The analysis was conducted for eight mainstream Sougahatchee Creek stations (4, 6 12) and repeated for the 12 tributary stations (4, 13-15, 17-24). Tributary station 16 was eliminated because a portion of the upstream discharge and load was apparently entering the ground water before reaching station 16 (Figures 28 through 34). Also station 16 was the only tributary site that received effluent from a waste treatment facility (West Point Stevens). The analysis of mainstream stations allowed us to evaluate the effects of point source and nonpoint source 67

77 68

78 69

79 Table 11. Percent land cover in 2001 for Sougahatchee Creek mainstream and tributary watersheds by station as determined from Landsat satellite imagery. Percent Land Cover Stations Urban Forest Clearcut Grassland Tilled Ag Water Wetland Mainstream Tributary

80 71