The use of fencing to exclude pastured animals from streams has been recognized as an agricultural best-management practice. Streambank fencing was installed in a small basin within the Mill Creek Watershed of Lancaster County, Pa., during summer 1997 to evaluate the effectiveness of fencing on surface-water quality. A preliminary review of data collected during a pre-treatment, or calibration period (October 1993 through June 1997), and part of the post-treatment period (July 1997 through November 1998) has identified a varied instream nutrient response to streambank fencing.

Concentrations of total nitrogen (N) during low-flow periods were significantly reduced by 20 to 31 percent at treated relative to untreated sites, but the yield of total N during low-flow conditions did not change significantly. Low-flow concentrations and yields of total phosphorus (P) did not change significantly at the outlet of the treatment basin, but data from a tributary site (T-2) in the treatment basin showed a 19- to 79-percent increase in the concentration and yield of total P relative to those at untreated sites. The total-P increase was due to increased concentrations of dissolved P. The processes causing the decrease in the concentration of total N and an increase in the concentration of total P were related to stream discharge, which declined after fencing to about one-third lower than the period-of-record mean. Declines in stream discharge after fence installation were caused by lower than normal precipitation. As concentrations of dissolved oxygen decreased in the stream channel as flows decreased, there was increased potential for instream denitrification and solubilization of P from sediments in the stream channel. Vegetative uptake of nitrate could also have contributed to decreased N concentrations. There were few significant changes in concentrations and yields of nutrients during stormflow except for significant reductions of 16 percent for total-N concentrations and 26 percent for total-P concentrations at site T-2 relative to the site at the outlet of the control basin.

Suspended-sediment concentrations in the stream were significantly reduced by fencing. These reductions were partially caused by reduced cow access to the stream and hence reduced potential for the cows to destabilize streambanks through trampling. Development of a vegetative buffer along the stream channel after fence installation also helped to retain soil eroding from upgradient land. Reductions in suspended sediment during low flow ranged from 17 to 26 percent; stormflow reductions in suspended sediment ranged from 21 to 54 percent at treated relative to untreated sites. Suspended-sediment yields, however, were significantly reduced only at site T-2, where low-flow and stormflow yields were reduced by about 25 and 10 percent, respectively, relative to untreated sites.

Benthic-macroinvertebrate sampling has identified increased number of taxa in the treatment basin after fence installation. Relative to the control basin, there was about a 30-percent increase in the total number of taxa. This increase was most likely related to improved instream habitat as a result of channel revegetation.