REPORT ON C-81740 DETAILED IDENTIFICATION OF NUTRIENT SOURCES IN AN AGRICULTURAL
WATERSHED
1. INTRODUCTION
During the period September 1972 through April 1974, a fairly intensive study was made of the nitrate, phosphorus and suspended solids content of Fall
Creek and streams draining selected sub-watersheds of the Fall Creek watershed.
Some results of that study have been summarized and published [see below].
The summary,of that work indicated the following major areas of uncertainty: 1)
The amount of soluble phosphorus delivered in the winter of 1972-73 was considerably more than the amount of soluble phosphorus delivered in the winter of 1973-74. This difference can be explained equally by one of the following two hypotheses: (a) the reduced phosphorus content in the Creek was primarily the result of lower levels of phosphorus in domestic waste, due to the New York State ban on phosphorus in laundry detergents imposed in
June of 1973, or: (b) the storm patterns in the Fall of 1972 were responsible for leaching of large amounts of soluble phosphorus from freshly decomposing vegetation.
2) Most of the sub-watersheds initially selected for intensive studies were found to be somewhat unsatisfactory for estimating non-point sources of soluble phosphorus because of numerous point sources. Thus it was believed that study of another set of sub-watersheds would further improve the estimates of non-point sources of soluble phosphorus.
3) The nitrate content of streams draining sub-watersheds was found to be highly correlated with several watershed parameters. The correlations were not regarded as being conclusive and hypotheses suggested by them required substantiation by obtaining data from other small watersheds.
4) There was considerable evidence that substantial amounts of soluble phosphorus were converted to biologically unavailable forms during transport downstream. The implications of this result with regard to biological enrich- ment of the Creek are so important that confirmation appeared to be highly desirable.
The following report describes work subsequent to the above results and is intended to serve as an appendix to Chapter 3 in Nitrogen and Phosphorus:
Food Production, Waste and the Environment. (Keith Porter, Editor, Ann Arbor
Science Publishers, Inc. 1975). The overall objective of this work was to pursue the questions emerging from the first study as indicated above.
11. METHODS
The sampling and procedures of sample analysis were essentially identical to those in the studies cited above. The methods of data analysis were also similar.
The sampling locations selected for this phase of the study are shown in Figures Al and A2 (the prefix A will be used for all Figures and Tables in this report in order to distinguish them from those in the citation which may be referred to from time to time.) In particular, Figure A2 shows the subwatershed selected for study of a watershed in which there were only non- point sources.
III. RESULTS
A. The loading of suspended solids and phosphorus fractions for Fall
Creek at both Forest Home and Preeville and for Virgil Creek at Freeville.
Regressions of suspended solids and molybdate reactive phosphorus (MRP)

on discharge rate and rate of change of discharge rate were calculated as in the previous study (op. cit.). Coefficients estimated by the regressions for the three locations are presented in Tables Al through A6.
The data representing rate of discharge at locations I and 16 (Fall Creek at
Forest Home and Virgil Creek at Preeville) were obtained from hourly gage height readings reported by USGS. There were two periods when these estimates were known to be unreliable. In April 1974, an accumulation of ice at Location 16 made estimates for that month questionable. Similarly, during September 1975 there was torrential rain, and the discharge rate far exceeded the rating curve.
Unfortunately, continuous records of discharge rate at Location 15
(Fall Creek at Preeville) were not available, but staff gage readings were made each time a sample was taken. It was then determined, as illustrated in Figure
A3, that the discharge rate at Location 15 and at Location 1, when expressed on a unit surface area basis, were nearly identical. Thus, discharge rate at Location 15 was estimated as being equal to product of the ratio of the area of the two watersheds times the discharge rate at Location 1.
The regression analyses were applied to data according to month, or groups of months, as indicated in Tables Al through A6. This aggregation of the data was determined by two considerations: (a) the need to have a large enough group of samples to provide adequate representation of a large range of flow data and,(b) the need to allow seasonal differences to be reflected in the regression analysis. During months of high discharge rate, the sampling was intensive and was considered to be representative of the discharge for that month. For example, at Location I during the November- December 1974 period (Table Al), 42 samples were taken at discharge rates ranging from 1.4 to
50 m3/sec, while the range of discharge rates was from
1.2 to 50 m 3 /sec. ,During a period of high discharge in late February 1975, the range of sampling included a sample taken when the discharge rate was
89 m 3 /sec while the maximum discharge rate during the sampling period was
94 m 3 /sec.
The loading of suspended solids, MRP, and TS@'were calculated as described previously (op. cit.) and the pertinent loadings are listed in Tables
A7 through A12, together with estimates of reliability based on the previously described statistical analysis (op. cit.).
The average difference between TSP and MRP was found to be 12 microg P/l
(standard error of mean 0.5, based on 344 samples), which is equal to the value of this parameter found during the previous experimental period. The TSP loading was calculated as the sum of MRP loading plus the product of flow times average concentration of TSP-MRP, referred to above.
Since the April 1974 and September 1975 data at Location 16 were not available as already indicated, the comparisons of the three watersheds are based on the period May 1974 through August 1975. The total discharge and weighted mean concentrations are shown in Table A13.
First, the data in Table A13 illustrates the similarity in discharge among the various major sub-watersheds when expressed on a unit surface area basis.
Second, the amounts of suspended solids, MRP, and TSP per unit surface area vary by a factor of 2 even though discharge per unit surface area does not vary. The
Virgil Creek sub-watershed and apparently the area between Freeville and the
Forest Home station are major sources of suspended solids per unit surface area.
Probably most of the suspended solids in the Creek are not generated by human activity, but are the result of stream erosion in the Virgil Creek sub-watershed and between Freeville and Forest Home. Several areas of major bank undercutting and collapse are readily apparent on Fall Creek between Freeville and Forest
Home. Several such areas are evident also in the Virgil Creek sub-watershed and, in one region, Virgil Creek is actively cutting through a terminal moraine. Such erosion is not so evident in the Fall Creek sub-watershed above Freeville.

Total Soluble Phosphorus
Consistent with the results reported in the previous study period
(op. cit.), Virgil Creek is a major source of MRP, presumably as a consequence of the secondary sewage treatment plant at Dryden, as documented in the previous study period (op. cit.). The concentrations of MRP are twice those in the Fall Creek watershed above Freeville, which is similar in many other respects. The most remarkable aspect of the MRP data, as shown in
Table A13, is that the total amount of MRP at Forest Rome is virtually equal to the combined amount of MRP delivered by 15 and 16 at Freeville. Clearly this is a consequence of conversion of MRP to particulate forms since there are numerous point and diffuse sources between Freeville and Forest Home which would otherwise yield a net gain in MRP between the three sampling points. This is much clearer evidence of the transport losses referred to in the report on the previous study period (op. cit.).
The seasonal variation in the levels of MRP in the Creek are high- lighted in Table A14. The aggregated data also illustrates that during low flow months, the sum of loading of MRP at Locations 15 and 16 exceeds that at
Location I by a considerable margin. It would appear, therefore, that the immobilization of MRP on the suspended solids and subsequent deposition on the stream bed is greater during periods of low flow than during periods of high flow. During periods of high flow, only a portion of this deposited phosphorus is redissolved during the subsequent storm events when this sediment resuspended and carried downstream. (It is indeed unfortunate that the discharge data at Location 16 for the very intense storm period in late
September 1975 is unreliable. The stream discharge far exceeded the capacity of the bridge at the gaging site and thus there seems to be no means of reconstructing reasonable estimates of a fair amount of the discharge, although samples were collected and analysed.)
B. Quantification of types of sources contributing to the MRP and TSP loading in
Fall Creek.
1) Biogeochemical phosphorus
In the earlier study, the term biogeochemical phosphorus was applied to that phosphorus which is transported independently of human activity. To determine the loading of such geochemical phosphorus, a watershed with no human activity, watershed iii, was investigated during the study period reported here. The overall average concentration of MRP of the samples collected from the watershed was 6.9 microg P/liter (S.E. of 0.6 for 46 samples) and the corresponding value of the TSP was 16 +/- 3 microg P/l. This agrees well with the average value of
MRP of 6.0 microg P/I and TSP of 15 microg P/I reported in Table 3.7 (op. cit.) for 123 well and stream samples (including 9 from watershed 111). Since these values are clearly not significantly different from those reported in Table 3.7 and because the values reported in Table 3.7 are representative of a wider range of samples, the biogeochemical phosphorus loading is calculated here as in the previous report; namely Ekg MRP = (6) (E discharge in m 3x 10-6 and Zkg TSP = 15
(E discharge in m 3 x 10-6 ).
The results shown in Table A16 illustrate that at Location I the biogeo- chemical MRP and TSP, when expressed as percentages of the total, are very similar to those reported for the previous study period (Table 3.14, op. cit.); that is, at Location 1, approximately 40% of the MRP and 50% of the TSP could be attributed to biogeochemical sources. At Location 15, corresponding values were
45 and 60% respectively, reflecting the lower inputs of point source P, whereas at Location 16 with much larger upstream inputs of point source P the corresponding values were 22% and 37%.
2) Diffuse and point source inputs
The diffuse sources of inputs were estimated by the hydrograph analysis procedure described previously (op. cit.). The results representing
Location I for the 1974-75 winter are shown on Figure A4 and again in

Figure A5 together with the results reported previously for purposes of comparison. Evidently the relation between MRP concentration and the ratio of surface runoff to total discharge for the winters of 1973-74 and 1974-75 are very similar but differ substantially from corresponding relations for 1972-73.
The marked difference lends additional support to the hypothesis that the results of the 1972-73 winter are strongly influenced by the phosphorus derived from laundry detergents.
The results at Locations 15 and 16 are shown in Figures A6 and A7, respectively.
The results at Location 16 are similar to those at Location 1, although the correlation between the MRP concentration and the ratio of surface runoff to total discharge is only 0.53 for Location 16. This may reflect greater variations in the balance between point and non-point sources in the subwatershed upstream of 16 compared to the whole watershed. As can be seen from
Figure A6, the estimated maximum concentration of MRP(discharge composed entirely of surface runoff) at Location 15, were about 25% higher than those at
Locations I and 16. Also, the estimated rate of increase of MRP concentration at
Location 15 is much higher than those at Locations I and 16 as the proportion of surface runoff increases. This might be expected since the sub-watershed draining to Location 15 has proportionately fewer point sources, which are less a function of surface flow than the sub-watershed draining to 16 and the total watershed upstream of Location 1. Estimates of the loadings of MRP due to human activity, yet originating from diffuse sources, are shown in Table A15. In Table
A16, a summary of all estimated sources are shown. The MRP loadings from point sources was estimated as being the difference between total loadings and the sum of the biogeochemical and diffuse source loadings.
The preponderance of the non-biogeochemical loading in upper Fall Creek above
Preeville is associated with diffuse sources, while for Virgil Creek, the point source inputs are much the larger, since these include the discharge of the
Dryden sewage treatment plant. The Table also makes evident the apparent transmission losses of MRP between Preeville and Forest Home; that is, the difference between the sum of loading at Location I and the sum of loading at 15 and 16 is unreasonably small.
Concerning these estimates, it may be argued that the validity of the hydrograph analysis procedure as a means of estimating surface runoff and the diffuse source inputs is clearly questionable because of the influence of point sources
(see op. cit. for more detailed discussion). However, it does provide a means of estimating the maximum upper limit for point sources, and this should be kept in mind when considering the ensuing analysis. In previous work, the net effect of agriculture was estimated to increase the concentration of TSP in flow from active agricultural land by 18 microg TSP/l. Corresponding calculations are reported in Table A17, basedon the foregoing analysis. The results are in reasonable agreement with the value of 18 microg TSP/I (increases in concentration above biogeochemical levels due to diffuse source inputs) reported earlier. Again, the uncertainties in such a number are very great indeed. 3)
Sub-watershed studies The objective of studies of the small watersheds outlined in Figure A2 was to verify the results of the foregoing analysis. The data in
Table A18 summarizes the results of the sampling analysis and the descriptions of the sub-watersheds. In addition a staff gage was installed at Locations 122 and III and readings were taken when these locations were sampled.
Unfortunately,the USGS has not completed the calibration of these stations and hence no discharge data is available.
It may be noted that the concentrations reported in Table A18 seem unexpectedly high in some cases. Several explanations can be suggested.
First, there were several point sources, particularly just above Location 116 and in the watershed above 103 both of which were found to have high concentrations of MRP. However, no major point sources were located above 119,
120 and 122 which had low concentrations of MRP, although there is a barnlot

which appears to be outside 122, but which may in fact contribute significant amounts of drainage to 122. Most of the watersheds also contain houses and the sewage systems from some of these houses may discharge directly into streams.
Hence, even in these watersheds point sources may make large contributions.
Second, the values recorded in Table A18 may be misleading. The area is characterized by very porous glacial debris through which there could be considerable volumes of subsurface flow. If it is assumed that this water contains only 15 microg TSP/l, and that the runoff actually sampled was only a fraction of the total flow draining the area, then the average concentration of the total flow could significantly differ from the estimates shown in Table
A18.
IV. MAIN CONCLUSIONS
Main conclusions of this phase of the Fall Creek study are as follows:
1) Further evidence was obtained that the New York State ban on phosphatic detergents has markedly reduced the levels of phosphorus transported by Fall
Creek.
2) Further estimates of the loads of soluble phosphorus from non-point sources were obtained and agreed closely with those in the the previous study, i.e., diffuse sources due to human activity contribute abo ut 30% and 20% of the
MRP and TSP respectively at Location 1.
3) Evidence obtained during this study further supported the hypothesis that MRP is converted to particulate forms of phosphorus during transit downstream.