Plot size was 6' x 15' with 6-ft alleys between plots within blocks and
 10-ft alleys between blocks. The experimental design was a randomized
 complete block with four replicates. Yield was estimated from the harvest of
 a 30-inch by 15-ft strip in each plot. Total green weight of forage from each
 plot was determined and moisture content was determined from a dried 400 gram
 sample of green forage. Dry matter yield and hay yield (16% moisture) were
 calculated using appropriate conversion factors. Nitrogen concentration of
 bermudagrass tissue was determined by micro-Kjeldahl procedures. Clipping
 frequency was about every four weeks.

 Maximum and minimum N-rates were determined by a two-step process: (1)
 least significant differences (Isd) for yield at P (probability) = 0.05 and P
 = 0.50 were calculated from the analysis of variance (AOV) error mean square
 and student's t (Steel and Torrie, 1960), (2) nitrogen fertilizer levels
 with 5 and 50% probabilities of no response (PNR) to additional fertilizer-N
 were determined from graphs of regression models (yield versus fertilizer-N)
 by subtracting Isd (0.05) and Isd (0.50) from maximum predicted yield,
 plotting horizontal lines from these points on the Y-axis to the regression
 line and vertical lines from the resulting points on the regression line to
 the X-axis, respectively. Minimum fertilizer-N corresponds to PNR (0.05) and
 maximum fertilizer-N corresponds to PNR (0.50).

 Regression equations for soil-pH versus rates of ammonium sulfate and
 ammonium nitrate were calculated using the minitab data analysis system
 (Minitab, Inc. 1985). Predicted soil-pH changes were calculated from
 regression equations for annual application of 450 lb of N/acre and 48 lb of
 S/acre with specified applications of dolomite.

 RESULTS AND DISCUSSION

 Yields of bermudagrass hay were above 14 tons/acre in 1986 (Fig. 1.).
Lower yields in 1987 and 1988 were due to lower rainfall amounts during the
growing season each year, compared to 1986. Highest yield in 1987 was about
13 tons/acre and it was near 11.6 tons/acre in 1988. Residual soil nitrogen
was higher in 1986 than in the following years as indicated by a yield of 9.5
tons/acre in 1986 with zero fertilizer-N and near 6 tons/acre with zero N in
1986-87.

 A linear yield response of bermudagrass to nitrogen was observed in 1986
(regression not shown). However, mean yields for 600 and 800 lb of N/acre
were about the same. Maximum and minimum N-rates were calculated for 1987 and
1988 (Fig. 2.). The minimum N-rate was 350 lb/acre each year, because the
probability of no response (PNR) to additional fertilizer N was only 5%.
Therefore, the risk of not getting a response to 350 Ib of N per acre was very
low. Maximum N-rate was 500 lb/acre in 1987 and 450 lb/acre in 1988, because
the PNR to additional N was 50%. These data suggest that 350 lb of N/acre be
applied to unirrigated coastal bermudagrass and no more than 500 lb of N/acre
be applied to irrigated coastal bermudagrass for maximum N recovery efficiency
and minimum environmental impact.

 Nitrogen content of the third harvest of coastal bermudagrass ranged from
1.74% with no applied N to 2.36% with 800 lb of N/acre (Fig. 3.). However,
the 200 lb of N/acre treatment contained 2.33% N, indicating that protein
concentration was not increased by high N-rates. Nitrogen concentration of