Sources, Instream Transport, and Trends of Nitrogen, Phosphorus, and Sediment in the Lower Tennessee River Basin, 1980-96


An objective of this analysis of historical data is to describe temporal trends in nutrient and sediment concentrations during the period 1980-96 and to interpret the trends with respect to changes in sources during this period. Although quantitative data on temporal variation of sources are sparse, a comparison of general information about source changes with observed trends of instream concentration is possible.

Trends of Inputs

The primary sources of nitrogen and phosphorus for which data are available for the LTEN River Basin are wastewater discharge, fertilizer, and livestock waste tables 3 and 4). The volume of wastewater discharge increased during the period 1980-96 as a result of population growth. Concentrations of nitrogen and phosphorus in wastewater decreased during this period, however, because of legislated control of municipal-effluent quality implemented through construction and upgrading of treatment plants to bring all dischargers to secondary or tertiary treatment standards. Concentrations of phosphorus in wastewater decreased dramatically starting around 1988, when reductions in the phosphate content of commercially-available detergents were made to reduce the phosphorus input to wastewater treatment plants (S. Fishel, Tennessee Department of Environment and Conservation, oral commun., 1997).

Significant changes have occurred in effluent concentration of certain constituents of total nitrogen; specifically in the relative amounts of reduced and oxidized forms of nitrogen. Because the reduced forms of nitrogen (organic nitrogen and ammonia) deplete instream oxygen levels and because ammonia is toxic to fish and aquatic life, advanced treatment processes have focused on converting ammonia and organic nitrogen to nitrate (U.S. Environmental Protection Agency, 1993). These processes have resulted in decreases in effluent loads of ammonia and increases in effluent loads of nitrate.

Changes in stream inputs of nitrogen and phosphorus from fertilizer and livestock waste between 1980 and 1996 are more difficult to estimate, even qualitatively. As with wastewater, these land-phase inputs increased during this period, based on pounds of fertilizer applied and animal census data; however, the extent to which improvements in nonpoint-source controls have offset the increased inputs by reducing delivery of land-phase to stream inputs is impossible to generalize.

Trends of Instream Concentration

Trends of instream concentration of nitrogen, phosphorus, and sediment were quantified with the multivariate log-linear regressions included in Cohn's Estimator program (Cohn, 1988). Trends were estimated by examining the statistical significance of the coefficient on time (β3 in equation 1). The direction (increasing or decreasing) of significant trends was determined from the sign of β3. Positive values of β3 indicate an increasing trend; negative values of β3 indicate a decreasing trend. Trend results are reported for two separate periods--the period of available record between 1980-96 and a common period of record (last column of table 8).

The trend results for the period of available record constitute a larger data set and span a longer time period, compared with results for the common period of record, and thus are more useful for at-site interpretations of instream trends with respect to trends in sources. The trend results for the common period of record, however, are more useful for comparing results among sites (fig. 19). The common period used for this second set of tests (generally from mid-1980's to mid-1990's) differed slightly among constituents but is consistent for results at all sites for a constituent. Trend tests could not be done for a common period of record for suspended sediment, however, because there were too few sites and the periods of available record generally did not overlap.

The time series of concentration residuals (from a regression against flow and season) for selected sites and constituents are shown in figure 20. Use of residuals, rather than actual concentrations, in these displays shows more clearly how concentration varies with time, independent of other influences (season and streamflow). A smoothed curve of the residuals calculated by locally weighted scatter-plot smoothing (LOWESS) is displayed with the residuals to illustrate the trend pattern. The trend estimate from the analysis of the common period of record is also shown on these plots in the inset boxes, for comparison.

All significant trends in nutrient concentrations during the common period of record (mid-1980's to mid-1990's) were decreasing trends, except for total nitrite plus nitrate, which increased at site 9 (Elk River below Tims Ford Dam). The constituents for which significant downward trends were most commonly observed were total ammonia and total nitrogen. Sites with decreasing trends in ammonia concentration during the period 1986-94 were also sites which ranked relatively high (compared with the other sites) in wastewater inputs compared to other inputs, and the timeframe of the observed trends spans the period of changes in wastewater loading. These lines of evidence suggest that the ammonia trends at site 1 (Clarks River at Almo), site 5 (Duck River at Williamsport), and site 7 (Shoal Creek at Highway 43) result from decreases in wastewater effluent concentrations. Concentrations of total phosphorus did not decrease during the period 1985-93 at the sites with decreasing ammonia trends, however, as might have been expected considering reductions in wastewater loading of phosphorus during this period.

The common-period trend results for site 18, Tennessee River below Raccoon Mountain, are apparently contradictory (fig. 19); no significant trend is in concentrations of nitrite plus nitrate or ammonia plus organic nitrogen, but a significant decreasing trend exists in total nitrogen concentration, which is calculated as the sum of concentrations of the first two constituents. This discrepancy can be explained by the slightly different time periods used as the common period for each constituent: 1986-93 for ammonia plus organic nitrogen, and 1985-94 for nitrite plus nitrate and for total nitrogen. Decreasing trends for site 18 for all three nitrogen constituents for the period of available record (1981-94, table 8) are reasonable: decreasing trends exist in concentrations of all three constituents.

The trend results include periods other than the common period used for spatial interpretation (table 8). Total ammonia increased at site 15 (Tennessee River at mile 23) during the period 1990-94, dissolved orthophosphorus increased at sites 12 and 19 (Town Creek near Geraldine and Scarham Creek near Kilpatrick) during the period 1988-96, and total nitrogen increased at site 14 (Tennessee River at Highway 60 near Paducah) during the period 1980-84. The increasing trends in dissolved orthophosphorus at the Town and Scarham Creek sites (12 and 19) were not expected, because the period of trend analysis (1988-96) corresponds to a period of many recognized improvements in management of poultry-waste runoff in the watershed for the sites. This suggests that the predominant input of phosphorus in these basins is from another source(s). A separate explanation, however, is that the improvements in source control may have had the desired effect of causing decreases in instream concentrations and loads of total phosphorus and particulate phosphorus, but with a corresponding conversion of part of the particulate-phase phosphorus to dissolved forms (for example, in sediment detention areas), thus causing the increasing trend in dissolved orthophosphorus (Holt and others, 1970). Unfortunately, this hypothesis cannot be tested because of insufficient data for total phosphorus at these sites.

It is important to emphasize that these trend results only describe the net change in concentrations between the start and end of the period of analysis. Temporary fluctuations in concentration, caused by temporary changes in sources, for example, are not detected by these trend tests. These fluctuations are evident, however, in the time series of concentration residuals (fig. 20). The time series and LOWESS-smoothed line of total ammonia and total nitrogen concentrations at site 9 (Elk River at Tims Ford Dam) suggest a temporary decline in the late 1980's, but then an increase and return to early 1980's levels, with no net change or trend. This pattern might be caused by increases in sources other than wastewater during this period, superposed on and offsetting declines in wastewater in the early part of the period; or this pattern might correspond with temporary reductions in nonpoint-source loads because of decreased runoff during the drought of 1985-88.

Next Title Page Table of Contents List of Illustrations List of Tables Conversion Factors and Vertical DatumGlossary

Last modified: Tue Feb 8 08:41:48 2000

Tennessee home page