Phosphorus and nitrogen concentrations and loads at Illinois River south of Siloam Springs, Arkansas, 1997-1999

Water-quality sampling consisting of every other
month (bimonthly) routine sampling and storm event
sampling (six storms annually) is used to estimate
annual phosphorus and nitrogen loads at Illinois River
south of Siloam Springs, Arkansas. Hydrograph separation
allowed assessment of base-flow and surfacerunoff
nutrient relations and yield. Discharge and
nutrient relations indicate that water quality at Illinois
River south of Siloam Springs, Arkansas, is affected by
both point and nonpoint sources of contamination.
Base-flow phosphorus concentrations decreased with
increasing base-flow discharge indicating the dilution
of phosphorus in water from point sources. Nitrogen
concentrations increased with increasing base-flow
discharge, indicating a predominant ground-water
source. Nitrogen concentrations at higher base-flow
discharges often were greater than median concentrations
reported for ground water (from wells and
springs) in the Springfield Plateau aquifer. Total estimated
phosphorus and nitrogen annual loads for calendar
year 1997-1999 using the regression techniques
presented in this paper (35 samples) were similar to
estimated loads derived from integration techniques
(1,033 samples). Flow-weighted nutrient concentrations
and nutrient yields at the Illinois River site were
about 10 to 100 times greater than national averages for
undeveloped basins and at North Sylamore Creek and
Cossatot River (considered to be undeveloped basins in
Arkansas). Total phosphorus and soluble reactive
phosphorus were greater than 10 times and total nitrogen
and dissolved nitrite plus nitrate were greater than
10 to 100 times the national and regional averages for
undeveloped basins. These results demonstrate the utility
of a strategy whereby samples are collected every
other month and during selected storm events annually,
with use of regression models to estimate nutrient
loads. Annual loads of phosphorus and nitrogen estimated
using regression techniques could provide similar
results to estimates using integration techniques,
with much less investment.