Preliminary Investigation of Nitrogen Rate Influence on Irrigated Bermudagrass Forage Production

Abstract

Bermudagrass (Cynodon dactylon) forage production recommendations are often developed in natural environments with available water limitations, often resulting in highly variable responses and lower average responses. As farmland ownership changes and agriculture and irrigation technologies become more affordable the amount of irrigated hay production has increased. While much of the agronomic management does not differ between rain-fed and irrigated environments, nutrient use and uptake dynamics may. This requires a reevaluation and potential adjustment of current recommendations to allow for increased yield potential of irrigated production systems without detrimental impacts on the system. The objective of this study was to identify the need for further investigation of nitrogen application rates for forage bermudagrass production under irrigated conditions. Nitrogen applications of 0 to 280 kg N ha⁻¹, in 56 kg increments, were applied at spring green-up and following the first and second harvests. Dry matter biomass, crude protein, and total digestible nutrients increased with increasing nitrogen application rate, while yield and profit maximizing rates both exceeded the typical recommended rate for bermudagrass hay production. The responses noted for increased nitrogen application rates indicate the need for further investigation of N requirements of non-moisture-limited hay production.

1. Introduction

Bermudagrass (Cynodon dactylon) is a cornerstone warm-season perennial forage across the southern regions of the United States of America because of its persistence, rapid regrowth, and adaptability to a wide range of soils and management intensities. Nitrogen (N) is the number one most limiting nutrient in most cropping systems, including bermudagrass, as it leads to the productivity and nutritive value. Numerous studies note Bermudagrass forage to be highly responsive to N applications, and plateaus vary by environment, management, and cultivar. Most of the current literature shows production maximization between 100 and 300 kg N ha⁻¹ yr⁻¹, with some high-input or irrigated settings exhibiting continued response at much higher rates [1,2,3,4,5,6,7].

Much research shows that cultivar decision plays a key role in the magnitude and efficiency of N response. Cultivars differ in growth habit, disease tolerance, and forage quality, which in turn influence the performance and fertilizer efficiency, as shown by Burns et al. (2009) [8]. A multi-year grazing trial that comparing ‘Coastal’ and ‘Tifton 44’ across three N rates (100, 200, and 300 kg N ha⁻¹ yr⁻¹) showed higher nutritive value and improved steer performance for ‘Tifton 44’ at a given N rate [8]. While research has shown that cultivar affects the N response of bermudagrass, environmental factors, particularly moisture availability, can also have a major influence.

Moisture availability is a major co-driver of crop response to N applications. Under rainfed conditions, episodic water deficits constrain leaf area expansion, N assimilation, and regrowth, often masking the yield potential associated with higher N rates. In contrast, studies that actively avoid water stress via timely irrigation report steeper N response and higher plateaus. For example, field work in the Southeastern Coastal Plain showed that maintaining soil water potential above −30 kPa with supplemental irrigation increased bermudagrass hay yields and shifted the response to N under both 4- and 8-week cutting intervals [9]. Even recent research in turfgrass highlights the relationship between N response and water use, citing diminished production under decreased N rates [10]. Ashley et al. (1965) [11] highlights an increase in production and N response of bermudagrass under irrigation compared to non-irrigated. However, this contradicts previous work by Overman et al. (1990) [12], which showed that irrigation improved production, but did not relate to N response. Similarly, research has identified potential residual N influence at high fertilization rates, which may influence subsequent production [11]. While these findings are contradictory, they emphasize the importance of understanding N response in irrigated environments.

At high fertilization intensities and frequent harvests, bermudagrass can sustain very large annual yields, but diminishing marginal returns and reduced N recovery are common at the upper end of N supply [13]. Despite extensive work on bermudagrass fertility, few studies deliberately eliminate moisture stress and isolate N response in an intensive multi-harvest system. We hypothesize the current N rate recommendations may be insufficient for maximizing bermudagrass production under irrigation. Therefore, our objective of this study was to identify the need for further investigation of nitrogen application rates for forage bermudagrass production in irrigated production environments.

2. Materials and Methods

2.1. Field Study

This study was conducted as a preliminary investigation over a single site year in 2024 at an on-farm location near Marmaduke, AR (36°8′34.17″ N, −90°19′37.68″ W) on a Beulah fine sandy loam with a Midland hybrid bermudagrass. The experiment was established as a randomized complete block design with 5 N application rate treatments, based on the removal rate of 25 kg N per Mg of dry matter [9], and one no-N control plot, replicated 4 times. Nitrogen application rates were 56, 112, 168, 224, and 280 kg N ha⁻¹, applied as urea ((NH₂)₂CO, 46% N) at trial establishment, and following the first and second harvests (

Table 1. Nitrogen rate treatment structure with nitrogen rates in kg N ha⁻¹ for each application. Superscript in treatment name identifies the number of times the rate was applied during the season.

	Application
Treatment	1st	2nd	3rd
	--------------------- kg N ha−1---------------------
0×3	0	0	0
56×3	56	56	56
112×3	112	112	112
168×3	168	168	168
224×3	224	224	224
280×3	280	280	280

). After the first harvest, additional treatments of a single application of 112 and 224 kg N ha⁻¹ were added within each replication to compare the potential influence of soil residual N, and these applications were only made once (

Table 2. Additional nitrogen rate treatments, added following the first harvest. N rates presented in kg N ha⁻¹ for the single application. Superscript in treatment name identifies the number of times the rate was applied during the season.

	Application
Treatment	1st	2nd	3rd
	--------------------- kg N ha−1 ---------------------
112×1	0	112	0
224×1	0	224	0

). Irrigation management was applied on an as-needed basis by the landowner, and early-season precipitation reduced the possibility of moisture limitations. Meteorological data was collected from a National Oceanic and Atmospheric Administration (NOAA) weather station located approximately 12.5 miles away from the research location (36°12′31.02″ N, −90°32′4.74″ W). Precipitation early in the season delayed the need for irrigation until August 7th (Figure 1). Cumulative precipitation over the whole season was 329 mm, with only a single irrigation event of 25.4 mm during this study period. Each harvest cycle received a total of 134, 147, and 76.9 mm of precipitation and irrigation in the first, second and third harvest cycles, respectively (

Table 3. Dates of fertilizer application, harvest, and total days of growth for each respective harvest cycle.

	Harvest Cycle
Event	1st	2nd	3rd
Application	6 May	22 June	2 August
Harvest	11 June	25 July	30 August
Days of Growth	36	33	28
Cumulative Precip. (mm)	147	134	32.5
Irrigation (mm)	0	0	25.4

) Mechanical failure on the irrigation system prevented the possibility of multiple irrigation events in the final harvest cycle, resulting in only one irrigation event of 25.4 mm.

2.2. Soil Analysis

Prior to trial establishment, 0–15 cm composite soil samples (Fifteen 2.5 cm cores per composite) were collected from the trial area. Soil pH was determined by pH electrode measurement of a 1:2 soil/water solution after a 30 min equilibration period. Phosphorus (P) and potassium (K) were extracted with the Mehlich 3 extraction solution [14] and analyzed using inductively coupled plasma-optical emission spectrometry [ICP-AES; SPECTRO Analytical Instruments; GmbH, Kleve, Germany]. Organic matter was analyzed using the weight loss on ignition method [15]. Soil analysis results estimated the soil to be a siltloam soil with a pH of 5.9, 114 ppm phosphorus, 211 ppm potassium, and 1.8% organic matter.

2.3. Biomass Harvest

Three harvest cycles of approximately 28–36 days between treatment application and harvest were achieved (Table 3). Biomass harvest of each plot was accomplished using a commercial mower (Walker Manufacturing, Fort Collins, CO, USA) by collecting the weight of all biomass greater than 10 cm in height from 1.1 m by 6.1 m area. Sub-samples were collected to be analyzed for moisture content and nutritive value and were dried for 72 h in a forced-air dryer at 55 °C prior to grinding to pass a 2 mm sieve. Forage yield was reported as Mg ha⁻¹ of dry matter (DM), calculated from wet weight in field and using the percent moisture derived from the sub-samples to determine DM. Forage nutritive analysis of crude protein (CP) was determined using a dry combustion carbon/nitrogen analyzer (CN 628, LECO Corporation, St. Joseph, MI, USA). Acid detergent fiber (ADF) concentration was analyzed on an Ankom 200 Fiber Analyzer (Ankom Technology, Macedon, NY, USA), using the methods outlined in [16,17]. Total digestible nutrient (TDN) concentration is calculated using ADF and CP using Equation (1). Crude protein and TDN were converted to a yield parameter by multiplying each by the DM yield (Equations (2) and (3), respectively). Crude protein yield is presented in kg ha⁻¹ to provide a more representative quantity of production, often less than 1000 kg ha⁻¹.

TDN (%) = 73.7 + 0.463(%CP) − 0.595(%ADF)

(1)

Crude Protein Yield (kg ha⁻¹) = %CP × DM (kg ha⁻¹)

(2)

TDN Yield (Mg ha⁻¹) = %TDN × DM (Mg ha⁻¹)

(3)

2.4. Statistical Analysis

Statistical analysis was conducted using PROC GLM in SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA). Dry matter, CP, and TDN yields were analyzed for N rate and harvest cycle influence. Mean separation was conducted using Fisher’s t-test for least significant difference (LSD) analysis at an alpha of 0.05. Single degree of freedom contrasts was used to determine residual N influence between corresponding N Rate treatments within the second and third harvest results for each parameter (Table 3), where the contrast shows the treatment being compared, and the estimate shows the resulting difference between the two treatments. If the estimate is a positive value, the first treatment in the contrast produced more yield, and vice versa for a negative estimate.

3. Results

3.1. Forage Production

Bermudagrass DM and TDN yields were influenced by the interaction between N rate and harvest cycle (p < 0.0001), and CP yield was influenced by the N rate (p < 0.0001) and harvest (p < 0.0001) main effects only. Therefore, the DM and TDN yields were sliced by the harvest cycle and analyzed for treatment response within the harvest cycle, while CP yields will be presented as averages of each treatment at each harvest, but analyzed by the main effects only (Figure 2).

The application of N in the first harvest cycle did not influence DM yield (p = 0.7907), averaging 6.1 Mg DM ha⁻¹. In the second harvest cycle, the application of N increased DM yield by an average of 4.3 Mg DM ha⁻¹, regardless of rate, compared to the check which yielded an average of 1.6 Mg DM ha⁻¹. However, when 224 kg N ha⁻¹ was applied, DM yield increased by 1.1 Mg DM ha⁻¹, greater than the application of 56 kg N ha⁻¹. By the third harvest cycle, N rate response levels out around 112 kg N ha⁻¹. The application of 112 kg N ha⁻¹ or more increased yield to an average of 3.2 Mg DM ha⁻¹, which is 2.7 and 0.8 Mg DM ha⁻¹ greater than the check and 56 kg N ha⁻¹ rates, respectively.

Single degrees of freedom contrasts from the second harvest cycle, showing a significant loss of 0.89 Mg DM ha⁻¹ when 224 kg N ha⁻¹ was applied in the first harvest and second harvest compared to only in the first harvest. A similar non-significant trend was observed for the 112 kg N ha⁻¹ as well (

Table 4. Single degree of freedom contrasts estimates and p values of the comparisons between single and multiple applications of the same rate of each yield parameter in harvests 2 and 3.

Variable	Harvest	Contrast	Estimate	p-Value
Dry Matter Yield	2	112×3 v 112×1	−0.782	0.0664
		224×3 v 224×1	−0.893	0.0382
	3	112×3 v 112×1	1.252	<0.0001
		224×3 v 224×1	0.462	0.0296
Crude Protein Yield	2	112×3 v 112×1	−29.5	0.7256
		224×3 v 224×1	−10.9	0.8967
	3	112×3 v 112×1	269	<0.0001
		224×3 v 224×1	143	0.0005
TDN Yield	2	112×3 v 112×1	−0.381	0.1719
		224×3 v 224×1	−0.482	0.0877
	3	112×3 v 112×1	0.828	<0.0001
		224×3 v 224×1	0.311	0.0208

). The third harvest contrast reports significant DM yield increases for both rates when applied to all harvest cycles compared to only the second harvest cycle. Unfortunately, no research has been identified specifically utilizing single applications to identify residual N carry-over in a multi-harvest system.

Crude protein yield was only influenced by the main effects of treatment and harvest.

The application of N increased CP yield by at least 303 kg ha⁻¹ compared to 212 kg CP ha⁻¹ when no N was applied. The application of 112 kg N ha⁻¹ or more increased CP yield by an average of 206 kg CP ha⁻¹ compared to the 516 kg CP ha⁻¹ produced when only 56 kg N ha⁻¹ was applied. The application of 224 kg N ha⁻¹-maximized CP at an average of 815 kg ha⁻¹, however, was not significantly different from 168 and 289 kg N ha⁻¹ application rates. Single degree of freedom contrasts did not show significant influence of residual nitrogen of CP concentrations in either the second or third harvest cycles. Crude protein yield decreased from the first and second harvests, averaging 711 kg CP ha⁻¹ to 431 kg CP ha⁻¹ in the third harvest. Single degrees of freedom contrasts for CP yield were not analyzed due to a lack of interaction between the main effect treatment and harvest.

Total digestible nutrient yield was only influenced by the N rate in the second and third harvest cycles, with the first harvest cycle averaging 3.7 Mg TDN ha⁻¹. The application of N in the second harvest cycle increased TDN yield from 0.97 Mg ha⁻¹ to an average of 3.6 Mg ha⁻¹ regardless of application rate. TDN yield of the final harvest cycle was increased by the application of N from 0.32 Mg ha⁻¹ to at least 1.5 Mg ha⁻¹. The application of 112 kg N ha⁻¹ or more increased TDN by an average of 2.1 Mg ha⁻¹, 0.6 Mg ha⁻¹ greater than 56 kg N ha⁻¹. Single degree of freedom contrasts from the second harvest cycle showed no significance of the N application sequence on TDN yield (Table 4). Similarly to DM yield, the third harvest contrast showed that TDN yield was significantly increased for both rates when applied in all harvest cycles compared to only in the second harvest cycle.

3.2. Quadratic Plateau Model and Optimum Nitrogen Rates

Quadratic plateau models described the relationships between N rate and DM, CP, and TDN yield across harvests (

Table 5. Quadratic plateau model coefficients, agronomic and economic optimum nitrogen rate estimations of nitrogen rate (kg ha⁻¹, yield (Mg or kg ha⁻¹) and net return (USD ha⁻¹) for dry matter yield (Mg ha⁻¹), crude protein yield (kg ha⁻¹) and total digestible nutrient (TDN) yield (Mg ha⁻¹).

		Coefficients			Agronomic Optimum			Economic Optimum
Variable	Harvest	Intercept	Linear	Quadratic	Rate	Yield	Return	Rate	Yield	Return
Dry Matter Yield	1	5.77	5.65 × 10−3	−1.60 × 10−5	181	6.28	USD 791.18	0	5.77	USD 951.40
	2	2.13	4.63 × 10−2	−1.20 × 10−4	197	6.69	USD 857.17	165	6.56	USD 877.32
	3	0.699	2.94 × 10−2	−7.50 × 10−5	196	3.58	USD 353.01	147	3.40	USD 382.58
Crude Protein Yield	1	435	3.57	−6.72 × 10−3	266	910	USD 1233.48	208	888	USD 1272.24
	2	191	6.37	−1.40 × 10−2	227	915	USD 1316.64	202	905	USD 1332.54
	3	76.1	4.91	−1.15 × 10−2	213	598	USD 789.02	183	588	USD 807.16
TDN Yield	1	3.38	4.28 × 10−3	−1.00 × 10−5	206	3.82	USD 1632.55	76	3.65	USD 1720.03
	2	1.30	2.79 × 10−2	−7.00 × 10−5	199	4.08	USD 1791.11	182	4.06	USD 1802.26
	3	0.426	1.86 × 10−2	−4.70 × 10−5	197	2.26	USD 891.33	172	2.23	USD 906.86

). Model intercepts, which indicate the expected yield when no N is applied, ranged from 0.70 to 5.77 Mg ha⁻¹ for DM yield, 76.1 to 435 kg ha ⁻¹ for CP yield, and 0.43 to 3.38 Mg ha⁻¹ for TDN yield. Linear coefficients were positive in all cases, whereas quadratic coefficients were negative, indicating yield responses to N increased in a curvilinear fashion with a plateau. This means yield increases with each increment of N, but as the N rate increases, the incremental increase in yield decreases until a point at which yield no longer increases and response plateaus. The plateau point for each nitrogen rate model is identified as the rate in Table 5; the yield or return is the resulting response at the plateau point. The model parameters were used to develop agronomic optimum and economic optimum N rate estimations. Nitrogen fertilizer prices was assigned at USD 1.35, USD 1.26, and USD 1.21 kg N⁻¹ for harvest cycles 1, 2, and 3, respectively, collected from personal communications on application dates, and DM yield was assigned a value of USD 165.35 Mg DM⁻¹. Crude protein and TDN yield were assigned values of USD 1.75 kg⁻¹ CP and USD 500 Mg⁻¹ TDN based upon the price of USD 350 per Mg of 20% supplementation feed [18].

Agronomic optimum N rates (AONRs), based on modeled yield, varied among harvests within each parameter. Dry matter yield was optimized at 181, 197, and 196 kg N ha⁻¹ for harvests 1, 2 and 3, respectively, producing 6.28, 6.69, and 3.38 Mg DM ha⁻¹. Crude protein yield required more N to optimize production, with 226, 227, and 213 kg N ha⁻¹ producing 910, 915, 598, and 915 kg CP ha⁻¹, respectively, in harvests 1, 2, and 3, respectively. Total digestible nutrient yields were optimized at 206, 199, and 197 kg N ha⁻¹ in harvests 1, 2 and 3, respectively, producing yields of 3.82, 4.08, and 2.26 Mg TDN ha⁻¹, respectively. Modeled yield shows that in harvests 2 and 3, DM and TDN yields are maximized by similar N rates, while CP yields always required more N regardless of the harvest. In the less responsive first harvest TDN required 25 kg ha⁻¹ more N than DM, and CP yield required 60 kg N ha⁻¹ more than TDN.

Economic optimum N rates were consistently lower than AONRs for all variables, except for the DM yield at harvest 1, where the economic optimum was at 0 kg N ha⁻¹ due to high check yield and low incremental return. Harvest 2 and 3 returns were maximized at 165, and 145 kg N ha⁻¹ respectively, returning USD 877.32 and USD 382.58 ha⁻¹ respectively. Crude protein EONRs was maximized at 208 kg N ha⁻¹ in harvest 1, 202 kg N ha⁻¹ in harvest 2, and 187 kg N ha⁻¹ in harvest 3, returning USD 1272.24 ha⁻¹, USD 1332.54 ha⁻¹, and USD 807.16 ha⁻¹, respectively. Returns for TDN production were USD 1720.03, USD 1802.26 and USD 906.86, respectively, for 76, 182, and 172 kg N ha⁻¹ in harvests 1, 2, and 3, respectively. While N rate estimates required for maximizing net returns were much lower than the yield maximizing estimated N rates, the yields produced at the EONR and AONR were within approximately 5 percent of each other for all variables and harvests, except DM yield in the first harvest, which was 8 percent different.

4. Discussion

4.1. Forage Production

Dry matter production in the first harvest cycle did not respond to N application, which we attribute to decreased growth due to cool, moist temperatures early in the 2024 season compared to during harvest cycles 2 and 3. This lack of response of DM in the first harvest resulted in limited response of TDN production in the first harvest as well. The increases in DM, CP, and TDN yields that were observed are echoed by the results of many studies, which showcase the influence of N on production. Specifically, DM yield results support Sohm et al. (2014) [4], who noted yield increases up to their max rate of 672 kg N ha⁻¹ under irrigated conditions. This, along with others [2,11] contradicts our DM response, noting increases in production when N rates > 672 kg N ha⁻¹ yr⁻¹ in a rainfed and irrigation system alike. Meanwhile, other studies, such as the one conducted by McFarland et al. (2007) [19], note little benefit of N rates greater the 112 kg N ha⁻¹ in rainfed conditions of Texas. Massey et al. (2011) [6] do not achieve similar seasonal application rates as ours or other studies’ but notice differing trends year after year in rain-fed systems. This highlights the influence of moisture variations on N response of bermudagrass, lending to the need for more research under irrigated conditions.

Quality production such as TDN and CP are not typically presented on a per-acre production basis; however, similar trends can be observed in production. The increase in CP production, as a factor of DM production, is similar to Sohm (2014) [4] who noted increased CP concentration as N rate increases. Similarly, our study notes that the N rate increased the CP yield across harvests; this is largely due to the CP concentration response to N. Total digestible nutrient yields are not commonly reported by other studies; however, one study performed by Kering et al. (2011) [5] found responses of TDN concentrations to N, noting that TDN increased with N up to 224 kg N ha⁻¹. Our study only noted increases up to 112 kg N ha⁻¹ in the final harvest, whereas harvests 1 and 2 were only impacted by the application of N regardless of rate. The lack of response to N in the first two harvests could largely be influenced by the limited DM response, coupled with small deviations in TDN concentrations.

Single applications are not commonly utilized by other studies to monitor for influences of residual N from previous applications; however, this study included these due to the lack of response in the first harvest. Single degree of freedom contrasts for the second harvests showing no additional yield is gained from the multiple applications compared to only receiving one application, while the third showed that yield was gained in all parameters when the multiple applications were made in comparison to the single application. One study [11] did note that subsequent production was influenced by N rates, where higher N around 1000 kg ha⁻¹ application led to greater production; however these rates are greater than what was used in our study. Together, the results of the contrasts show that residual N influences were not observed in this trial.

4.2. Quadratic Plateau Model and Optimum Nitrogen Rates

Quadratic Plateau model results showcase a curvilinear response to nitrogen, similar to other studies [4]; however, the achievement of a plateau in this study is dissimilar to what many studies reported. The achievement of a plateau allows for the full understanding of the nutrient needs of the crop. The plateau points were depicted by the AONR, which showcases the N rate where production is maximized. The EONR shows the rate that gives the highest profit based on the cost of application, not necessarily the highest yield. While the N rates needed to maximize profit in this study were lower than needed for the yield, the yield differences at the rates are often very similar. These preliminary results indicate that targeting economic optimum production may be able to achieve similar yields as non-limiting N rates under irrigated production. However, more research is needed to refine the optimum rates and identify relationships between agronomic and economic optimum rates.

5. Conclusions

The findings from this preliminary investigation indicate that the nitrogen application rate and timing exert a clear influence on bermudagrass DM and TDN yields under non-moisture-limited conditions, with less pronounced but still significant effects on CP yield. Overall, the quadratic model response to the N rate showed the economic optimum N rate is able to produce 95% of the agronomic maximum yield while maximizing the return on investment for every parameter. However, rates required to achieve agronomic and economic maximum DM yield are insufficient for achieving agronomically or economically maximized CP or TDN yields.

These patterns support the need for further, more comprehensive investigation of N rate effects in bermudagrass systems where moisture is not limited. Future research should explore multi-year datasets, soil nutrient strategies, and revisit residual N influences to refine N recommendations. Such work will help producers balance maximum forage yield and quality with economic returns, while optimizing nutrient use efficiency in Arkansas bermudagrass production systems.