Cover crops are multifunctional tools that mitigate environmental impacts of agriculture, enhance resilience to weather extremes, and suppress weeds and arthropod pests. Cover crops provide non-crop food and habitat resources that attract natural enemies of pests, but their outcomes for pest management are less clear in regions where keystone mutualisms between red imported fire ants and aphids dominate.

Here, we manipulate ant exclusion treatments and cover crop that vary in food/habitat resources treatments (living mulches and terminated cover crops), and examine responses of ants, aphids, and other herbivores and predators in a cotton agroecosystem.

Living mulches reduced both ants and aphids in the crop canopy by 97% and 93%, respectively, relative to bare soil treatments, and terminated cover crops reduced them as well by a lesser degree (~50%). Non-aphid herbivores occurred in low densities system-wide, and increased in living mulches, while native predators had variable responses to cover crops and ant exclusion. Cover crops had no effect on prey removal in the crop canopy, but living mulches tripled rates of weed seed biocontrol relative to bare soils.

Cover crops elicited a shift in fire ant foraging from cotton foliage downward to the soil-surface, preventing competitive exclusion by keystone ant/aphid mutualists that dominate crop monocultures. Cover crops altered the system-wide impacts of fire ants: reducing ecosystem disservices (i.e. aphid tending), and enhancing ecosystem services (i.e. weed seed biocontrol). These results provide incentives for cover crop adoption as a regenerative practice in large-scale commercial agriculture.

Our experiment compared four cover crop treatments in a conventional, reduced-tillage cotton (Gossypium hirsutum) agroecosystem: 1) white clover living mulch (Trifolium repens ‘Durana^®’), 2) terminated cereal rye (Secale cereale ‘Wrens Abruzzi’), 3) terminated crimson clover (Trifolium incarnatum ‘Dixie’), and 4) a bare ground treatment (Fig. 1). Terminated cover crop treatments were selected because they are commonly implemented conservation practices in agronomic crops regionally (Hill et al., 2021; Vaan et al., 2018, Weisberger et al. 2024)), and encouraged in control of Amaranthus palmeri which herbicide sites of action are rendered inadequate without an additional tool for weed control (Weisberger et al. 2024). The living mulch was selected because Durana^® clover has been optimized recently as a perennial groundcover in field crops with great promise for improving soil health and augmenting nitrogen fertility budgets (Hill et al. 2021; Sanders et al. 2017). The four cover crop treatments were randomized in four replicated blocks (16 main plots total) that were 10.9 m wide and 16.5 m long. There was no space between plots within each block, and 11 m alleys separating blocks were planted to cereal rye. Each main plot contained 12 rows of cotton plants with 90 cm row spacing. Within each main plot, we manipulated two ant manipulation subplot treatments (‘ant exclusion’, and control). Each plot also contained a subplot where A. palmeri seeds were sown, to examine effects of cover crops on suppression of the herbicide-resistant weeds (reported separately in Weisberger et al. 2024).

The cover crop trial was originally established on October 3rd, 2019 in a cotton field and was chisel plowed and disked for seedbed preparation that same day. Cover crop treatments were planted on October 18th, 2019 with a Great Plains ^Ⓡ no-till drill at the following rates: Cereal rye at 112 kg/ha, large seed box D2-65); crimson clover was planted at 22 kg/ha with a small seed box - 85), and white clover living mulch planted at 22 kg/ha with a small seed box - 80). Cereal rye and crimson clover were replanted every year in mid-November (after cotton harvest) for three years at the same rate. After year 3 of cotton harvest, living mulch was replanted on November 21st, 2022, and again in Feb 2023 due to poor germination. Cereal rye was terminated with a roller crimper on May 10, 2023, and glyphosate and dicamba were applied to cereal rye, crimson clover, and bare ground plots at 3.12 kg a.i. / ha), as well as within cotton rows in 0.2 m bands in living mulch treatments (Weisberger et al., 2024). Cotton cultivar ‘DG 3615 B3XF’, a Bt hybrid with herbicide tolerance, was planted with the no-till drill seeder five days after herbicide treatments were applied on May 24, 2023, at 38,000 seeds/ha (about 12 cm apart within rows), and follow-up herbicide treatments were applied on June 16 and July 12, 2023 (glufosinate at 2.3 L / ha).

On June 29th, 2023, two ant manipulation subplot treatments were established within each plot: 1) 'ant exclusion’ treatments, where ants were physically prevented from accessing four plants in a row, and ‘control’ treatments, where ants were allowed access to the plants. In ant exclusion treatments, a band of tape 5 cm wide was secured just below the bottom-most leaves at the bases of four consecutive plants. The tape was covered in Tanglefoot^Ⓡ sticky barrier similar to experimental ant exclusion treatments applied on wild cotton plants by Rudgers (2004). Data were not collected on the outer two plants in the ant-manipulation subplots to reduce effects of ants immigrating from neighboring plants. Each plot contained two pseudoreplicates of ant manipulation treatments (four subplots total), which were selected randomly and were >6 m apart. All pseudoreplicates were pooled at the plot level prior to analysis.

Insect sampling

Visual surveys: One week after ant exclusion treatments were established, we began visual surveys of insects on cotton plants. No data were collected from cotton plants in the outer two rows to minimize border effects. We conducted visual surveys on the two inner plants of the ant manipulation subplots by counting and identifying all insects on the leaves, stems, and flowers on the entire cotton plant. We identified all insects at the family level on-site. Visual surveys occurred weekly (three sample dates) on July 7, July 14, & July 21, 2023.  

Vacuum sampling: To examine cover crop effects on highly mobile pests and natural enemies not easily sampled in visual surveys, we also vacuum sampled ten cotton plants outside the ant manipulation subplots with a leaf blower modified in suction mode with a funnel and a mesh bag for capturing insects. We did not vacuum our experimental ant manipulation subplots, to avoid disturbing that experimental treatment. Ten plants in a single row of each plot were suctioned for three seconds each. Insects were transferred to 70% ethanol and stored for later identification. This survey was conducted once on July 7, 2023.

Pitfall traps: To examine how cover crops influenced ants at the soil surface, pitfall traps were set at two randomly selected locations in each plot for 72 h on July 14, 21, and 28. Because many traps were flooded after rain events or lost to wildlife damage in the second and third weeks, we only analyzed pitfall samples from July 14, when no traps were missing. Pitfall traps consisted of a 1 L deli cup sunk into the soil flush with the soil surface. After samples were collected, insects were stored in 70% ethanol for later identification.

Estimating biological control

To estimate insect predation, we secured 25 frozen fly pupae to 2 cm x 3 cm cardstock with double-sided carpet tape, covered the remaining sticky residue with sand, and attached them to cotton leaves with paper clips. We attached two cards per plot, one on a plant with the ant exclusion treatment, and one on a control plant. Pupae cards were collected after approximately 24 hrs. This survey occurred on July 13, and July 25, 2023.