Field experiments.

Experiments were conducted at the Athens Turfgrass Research and Education Center (ATREC) in Athens, GA (lat. 33.54°N, long. 83.22°W) and at the University of Georgia Facilities Maintenance Division (FMD) complex in Athens, GA (lat. 33.92°N, long. 83.37°W). The soil at ATREC was a Cecil sandy clay loam (fine, kaolinitic, thermic Typic Kanhapludults) with a pH of 5.5 and an organic matter content of 1.2%, whereas the soil at FMD was a Cecil sandy clay loam with a pH of 5.9 and an organic matter content of 1.7%. Seedbed preparation at ATREC consisted of cultivating the research site in two directions (perpendicular to one another) with a tractor-mounted rototiller (Bush Hog, Selma, AL) to a depth of 10.2 cm, and grading to provide a smooth planting bed of desired contour. A hand tiller (Stihl, Waiblingen, Germany) was used to a depth of 10.2 cm at the FMD location before grading to prepare the seedbed for planting. Plots measured 1.5 × 1.5 m and were arranged in a randomized complete block design with four replications at both locations.

‘Sahara II’ common bermudagrass (uncoated/untreated) was seeded at 98 kg⋅ha^–1 on 28 July 2021 at both locations. A seeding box (1.5 × 1.5 m) was placed on the corners of each plot, and seed was applied evenly across the soil surface using a shaker jar. Treatments were applied immediately to the soil after seeding and consisted of no treatment, fertilizer (5N–5P₂O₅–5K₂O) (EarthWorks Natural Organic Products, Martins Creek, PA) at 24 kg⋅ha^–1 N at seeding, fertilizer at seeding + 3 weeks after seeding (WAS), K. variicola 137-1036 (KV) at an inoculant rate of 1.87 L⋅ha^–1 [4 × 10⁸ colony-forming units (cfu)/mL] and a carrier volume of 3000 L⋅ha^–1 at seeding, KV at seeding + 3 WAS, fertilizer at seeding + KV 3 WAS, and fertilizer + KV at seeding + KV 3 WAS. Fertilizer was applied by hand using a shaker jar. Microbial inoculants were applied with distilled water using a watering can. Plots received ≈0.3 cm of water through an overhead irrigation system after treatment application and were covered with germination cloth (A.M. Leonard, Inc., Piqua, OH) to promote seed germination and prevent desiccation. Plants were mowed weekly (starting at 4 WAS) to a height of 3.81 cm with a walk-behind rotary mower (American Honda Motor Co., Alpharetta, GA), with clippings collected and removed. Clippings were removed to avoid interference with grid count data collection. About 2.5 to 4 cm of water per week was applied at ATREC and FMD through an overhead irrigation system (well water).

Measurements of CO₂ efflux (CE) (measured in micromoles per square meter per second) were recorded at trial initiation and 5 and 8 WAS with a LI-COR 8100 automated system (LI-COR, Inc., Lincoln, NE) to provide plant root and soil microbial respiration as well as to determine the overall metabolic activity of each system (Jensen et al., 1996; Ryan and Law, 2005). The setup included an infrared gas analyzer connected to the LI-COR 8100 device via an RS-232 serial cable. The LI-COR 8100 was also connected to a laptop computer via an ethernet cable to run the LI-8100A 4.0.0 software for sample collection. A 20-cm infrared gas analyzer survey chamber was placed on top of a polyvinyl chloride collar (diameter, 20 cm) that was inserted randomly into each plot to contain the sampling area (offset, 7.5 cm) for measurement. To prevent any irrelevant CO₂ buildup in the chamber, a 60-s prepurge and 45-s postpurge was conducted before and after each measurement, respectively. After closing the chamber, a dead-band period of 40 s was used before measurements were initiated to obtain a constant rate of efflux. An observation length of 60 s was used for each measurement. Change over time (Δ) for CE was determined for 5 and 8 WAS by comparing values to initial measurements.

Greenhouse experiments.

Experiments were conducted at the ATREC greenhouse complex (lat. 33.54°N, long. 83.22°W) in Athens, GA, during Summer 2021. On 27 July 2021, ‘Sahara II’ common bermudagrass (uncoated/untreated) was seeded at a rate of 98 kg⋅ha^–1 into circular pots (diameter, 15.2 cm) containing a 2:1 mixture of Cecil sandy clay loam (fine, kaolinitic, thermic Typic Kanhapludults) and Wakulla sand (siliceous, thermic Psammentic Hapludults). Seed was applied evenly to the soil surface and top-dressed lightly to increase seed-to-soil contact and reduce desiccation. Two pots were established per treatment per experiment replication to conduct a time-lapse destructive harvest at 3 and 6 WAS. Two experimental runs were conducted simultaneously in separate greenhouses using a randomized complete block design with five replications.

Treatments were applied on 27 July 2021 and were the same as described previously for field experiments. Fertilizer (5N–5P₂O₅–5K₂O) was applied by hand using a shaker jar at a rate of 24 kg⋅ha^–1 N. A stock solution of the microbial inoculant KV was prepared with distilled water and applied to the soil with a 50-mL syringe at an inoculation rate of 1.87 L⋅ha^–1 (4 × 10⁸ cfu/mL) and a carrier volume of 3000 L⋅ha^–1. Sequential applications of treatments were made on 17 Aug. 2021. Pots were irrigated immediately after inoculant and fertilizer applications with ≈0.3 cm of water to incorporate treatments into the soil. An untreated check was included for comparison. The pots was watered using an overhead irrigation system calibrated to deliver 3.8 cm of water/week. Natural light was supplemented with artificial light (metal halide) to remain at 500 μmol⋅m^–2⋅s^–1 photosynthetic photon flux (measured at the canopy) in a 12-h day to approximate summer light intensity and photoperiod. Conditions in the climate-controlled greenhouse were maintained at day/night temperatures of 32 °C/26 °C. Experimental blocks were arranged along a gradient created by the greenhouse cooling pads and associated fans.

Pots were harvested destructively at 3 and 6 WAS. Roots and shoots were separated from each other, washed of all soil, dried in an oven for 48 h at 110 °C, and weighed to determine root and shoot biomass (in grams) at 3 and 6 WAS.

Data analysis.

Analysis was conducted separately for TC, TQ, NDVI, and percentage of turfgrass cover at 3, 5, and 8 WAS, as well as CE and ΔCE at 5 and 8 WAS for field experiments. In addition, a separate analysis was also conducted for root and shoot weights at 3 and 6 WAS for greenhouse experiments. Analysis of variance was performed using PROC MIXED with the appropriate expected mean square values described by McIntosh (1983) in SAS (SAS version 9.2 for Windows; SAS Institute, Cary, NC). Means were separated according to Fisher’s protected least significant difference test at α = 0.05. Data were arcsine square root–transformed to stabilize variance as described by Bowley (2008). Transformed and nontransformed data were analyzed, and interpretations were not different; therefore, nontransformed means are presented for clarity.

Field experiments.

Experimental run-by-treatment interactions for field experiments were not significant (P = 0.57). Therefore, data from individual runs were pooled and are presented accordingly. At 3 WAS, the greatest percentage of turfgrass cover was observed in response to fertilizer + KV at seeding (64.5%) followed by fertilizer at seeding (48.5%– 56%; Table 1). The microbial inoculant applied alone at seeding resulted in 25.5% to 33.5% cover, which was slightly greater than the untreated check 3 WAS (23%). Sequential applications of fertilizer, fertilizer + KV at seeding + KV 3 WAS, and fertilizer at seeding + KV 3 WAS exhibited the greatest percentage of turfgrass cover 5 WAS (77%, 76%, and 67%, respectively; Table 1). The single application of fertilizer at seeding resulted in slightly less of a percentage of turfgrass cover 5 WAS (63%), whereas sequential applications of KV, KV at seeding, and the untreated check were statistically similar (47%, 42%, and 41%, respectively). At 8 WAS, all treatments that received at least one application of fertilizer resulted in a similar percentage of turfgrass cover (88.5%–99%), whereas treatments that consisted only of KV (59%–68.5%) had similar cover as the untreated check (65%; Table 1).

Table 1.

Establishment (percentage of turfgrass cover) of seeded ‘Sahara II’ common bermudagrass [Cynodon dactylon (L.) Pers.] in response to fertility and microbial inoculants applied during Summer 2021 in Athens, GA.

View Table

At 3 WAS, the greatest NDVI (0.52–0.58) and TQ (3.6–4.1) ratings were observed in response to treatments that supplied fertilizer at seeding (Table 2). Three of the four fertilizers-at-seeding treatments also resulted in the greatest TC (5.1–5.6). The microbial inoculant applied alone at seeding resulted in a similar NDVI (0.38–0.39), TC (4.1–4.3), and TQ (2.7–3.1) as the untreated check (0.36, 0.39, and 2.6, respectively). Sequential fertilizer applications, fertilizer + KV at seeding + KV 3 WAS, fertilizer at seeding + KV 3 WAS, and fertilizer at seeding treatments resulted in the greatest NDVI (0.60–0.65) and TC (5.6–6.1) 5 WAS (Table 2). The greatest TQ 5 WAS also observed in response to sequential fertilizer applications (6.1), fertilizer + KV at seeding + KV 3 WAS (6.1), and fertilizer at seeding + KV 3 WAS (5.8). At 5 WAS, KV-alone treatments resulted in a similar NDVI (0.51–0.54), TC (4.6–4.7), and TQ (3.6–3.9) as the untreated check (0.50, 4.7, and 3.4, respectively). A similar trend was observed 8 WAS with the greatest NDVI, TC, and TQ in response to sequential fertilizer applications, fertilizer + KV at seeding + KV 3 WAS, fertilizer at seeding + KV 3 WAS, and fertilizer at seeding (Table 2). Treatments receiving KV alone still resulted in similar NDVI, TC, and TQ as the untreated check 8 WAS.

Table 2.

Normalized difference vegetation index (NDVI), turfgrass color (TC), and turfgrass quality (TQ) of seeded ‘Sahara II’ common bermudagrass [Cynodon dactylon (L.) Pers.] in response to fertility and microbial inoculants applied during Summer 2021 in Athens, GA.

View Table

Treatments receiving KV 3 WAS resulted in the greatest ΔCE 5 WAS, with fertilizer at seeding + KV 3 WAS exhibiting the greatest ΔCE (5.3 μmol⋅m^–2⋅s^–1; Table 3). The microbial inoculant at seeding (3.4 μmol⋅m^–2⋅s^–1) and treatments only supplying fertilizer (3.1 μmol⋅m^–2⋅s^–1) were statistically similar to the ΔCE of the untreated check (2.1 μmol⋅m^–2⋅s^–1). At 8 WAS, fertilizer at seeding + KV 3 WAS and fertilizer + KV at seeding + KV 3 WAS resulted in the greatest ΔCE (4.5–4.7 μmol⋅m^–2⋅s^–1); however, the ΔCE of the untreated check (3.2 μmol⋅m^–2⋅s^–1) was statistically similar (Table 3). All other treatments exhibited ΔCE ≤ 2.8 μmol⋅m^–2⋅s^–1.

Table 3.

Carbon efflux of seeded ‘Sahara II’ common bermudagrass [Cynodon dactylon (L.) Pers.] in response to fertility and microbial inoculants applied during Summer 2021 in Athens, GA.

View Table

Greenhouse experiments.

Experimental run-by-treatment interactions for greenhouse experiments were significant (P = 0.03). Therefore, data were not pooled across experimental runs and results for each experiment are presented separately.

In experimental run 1, treatments that contained initial fertilizer applied alone exhibited the greatest root weights (0.43–0.56 g) 3 WAS (Table 4). Treatments that contained initial KV applications resulted in the lowest root weights (0.32–0.41 g) and were statistically similar to the untreated check (0.28 g) 3 WAS. The greatest shoot weights (1.15–1.45 g) 3 WAS were observed in response to all treatments that contained initial fertilizer applications. Treatments that supplied initial KV alone exhibited the lowest shoot weights (0.73–0.79 g), which were statistically similar to the untreated check (0.68 g). At 6 WAS, the greatest root weight was observed in response to sequential fertilizer applications (5.32 g), fertilizer + KV at seeding + KV 3 WAS (3.84 g), and single applications of fertilizer (3.58 g; Table 4). Treatments containing KV alone resulted in the lowest root weights (2.16–2.45 g), which were less than the untreated check (2.57 g) 6 WAS. Sequential applications of fertilizer resulted in the greatest shoot weight (9.15 g) 6 WAS followed by fertilizer + KV at seeding + KV 3 WAS (7.68 g) and a single application of fertilizer at seeding (6.27 g). All other treatments resulted in shoot weights ≤ 4.95 g and were statistically similar to the untreated check (4.82 g) 6 WAS.

Table 4.

Root and shoot weight of seeded ‘Sahara II’ common bermudagrass [Cynodon dactylon (L.) Pers.] in response to fertility and microbial inoculants applied in the greenhouse during Spring 2021 in Athens, GA.

View Table

The greatest root weight (0.57–0.71 g) 3 WAS in experimental run 2 was observed in response to treatments in which fertilizer was applied at seeding (Table 4). Treatments that supplied KV initially resulted in root weights of 0.24 to 0.36 g and were statistically comparable to the untreated check (0.30 g) 3 WAS. A similar trend was observed with respect to shoot weight 3 WAS. Single and sequential applications of fertilizer resulted in the greatest root weights (3.15–3.83 g) 6 WAS, whereas the other two treatments containing fertilizer resulted in similar root weights (2.44–2.57 g) as the untreated check (1.99 g; Table 4). The least root weights (1.59–1.64 g) were produced 6 WAS in response to treatments supplying KV only. Single and sequential applications of fertilizer resulted in the greatest shoot weights (7.71–8.21 g) 6 WAS; however, fertilizer at seeding + KV 3 WAS produced a comparable root weight (5.42 g). All other treatments resulted in statistically similar or lower root weights 6 WAS as the untreated check.

Treatments containing fertilizer consistently had greater NDVI, TC, and TQ than treatments that contained KV only. Establishment of bermudagrass plots in response to treatments containing fertilizer were nearly at full cover at the conclusion of the study (8 WAS), whereas treatments containing KV only were ≤ 70% cover and similar to the untreated check. However, the greatest ΔCE was often observed in the field in response to treatments that supplied KV 3 WAS. In the greenhouse, the greatest root and shoot weights were typically observed in response to treatments containing fertilizer, whereas KV-alone treatments resulted in root and shoot weights either similar to or less than the untreated check. Similarly, Peacock and Daniel (1992) observed a significantly greater hybrid bermudagrass growth rate in the greenhouse in response to urea fertilizer (3187 mg⋅m^–2⋅d^–1) than inoculation with Bacillus spp. (1903 mg⋅m^–2⋅d^–1) 17 d after treatment. Only a 17% increase in bermudagrass shoot growth with no increase in root growth was observed by Baltensperger et al. (1978) in response to inoculation with Azospirillum and Azotobacter spp. under low N fertility in the greenhouse. Contrarily, Coy et al. (2014) reported enhancements in root and shoot growth of ‘Tifway 419’ hybrid bermudagrass after inoculation with blends of Bacillus, Paenibacillus, and Brevibacillus spp. Although bermudagrass did not receive fertility during the trial, Coy et al. (2014) supplied bermudagrass with fertility for 3 consecutive weeks before trial initiation. Furthermore, mature hybrid bermudagrass transplants used by Coy et al. (2014) may have been more conducive to colonization than the common bermudagrass established from seed in our research. Increases in root and shoot weight of other C-4 grasses [pearl millet, Pennisetum glaucum (L.) R. Br.; foxtail millet, Setaria italica (L.) P. Beauv.] in response to Azospirillum spp. have been documented; however, all these grasses exhibit annual growth habits (Di Ciocco and Cáceres, 1994; Mane et al., 2000; Rafi et al., 2012).

Most turfgrass research documenting benefits from use of microbial inoculants has examined the response of cool-season grasses planted in sterile media and grown in controlled environments (Acikgoz et al., 2016; DeLuca et al., 2020). Therefore, microbial inoculants are often evaluated without competition from native microorganisms or under local environmental conditions. Coy et al. (2019) documented nitrogenase activity and microbial persistence in root and shoot tissue of common bermudagrass after field inoculation with Bacillus spp. However, bacterial strains used in their research were collected previously from native populations and isolated by Auburn University’s Department of Entomology and Plant Pathology, ensuring some adaptation to indigenous environments and competition with endemic microorganisms. The K. variicola bacterial inoculant used in our research has been gene-edited to continue N fixation even in the presence of high soil N (Bloch et al., 2020). Furthermore, it has also been gene-edited to decrease glutamine synthase production and activity, thus allowing bacteria to fix N in the presence of high soil inorganic N. Although these traits confer advantages, bacteria contained within this inoculant were isolated originally from the mucilage of aerial roots of an indigenous landrace of maize grown in Mexico (Van Deynze et al., 2018). Therefore, microorganisms adapted to our soil and environmental conditions may pose enough competition to reduce potential for colonization and N fixation from our K. variicola bacteria. In addition, although properly stored, a reduction in cfu’s of the liquid K. variicola formulation used in our research may have occurred between manufacturing and application. Freeze-dried materials that are rehydrated before treatment do not experience these potential shortcomings.

Intensely managed turfgrass requires large amounts of management inputs (fertilizer, irrigation, pesticides) and cultural practices (aerification, top-dressing) to sustain adequate growth and functionality. However, these same inputs often influence microbial activity, persistence, and species diversity through the alteration of the physical and chemical characteristics of the soil. In addition, turfgrass species may play a critical role in the potential for symbiotic relationships between plant roots and soil microorganisms. Research conducted by our group examined only one or two microbial inoculation events and monitored bermudagrass response over a short period of time. Coy et al. (2014) reported enhancement in root and shoot growth in response to weekly microbial inoculation over a 5-week period. Furthermore, documented growth enhancement of agronomic crops observed in response to K. variicola applications may be a result of reduced microbial competition from recurrent tillage events not conducted in turfgrass environments. Future research should examine the impact of more frequent and numerous inoculation events on turfgrass performance over multiple growing seasons to determine the feasibility of colonization and potential adoption of more sustainable turfgrass N management.