Providing evidence to improve practice

Action: Exclude ants that protect pests

Key messages

Parasitism: One of two replicated, controlled studies (one also randomised) from Japan and the USA found greater parasitism of pests by natural enemies when ants were excluded from trees. The other study found greater parasitism at one site but no effect at another.
Natural enemies: Five studies (including four randomised, replicated, controlled trials) from Japan, Switzerland and the USA found effects varied between natural enemy species and groups, sampling dates, sites, crop varieties and ground cover types beneath trees.
Pests: Three of seven studies (including four randomised, replicated, controlled trials) found fewer pests and another found fewer pests at times of peak abundance only. One study found mixed effects depending on date and other actions taken simultaneously (predator attractant and ground cover treatments). One study found no effect.
Damage and tree growth: One study found no effect on damage to tree foliage but one study found greater tree growth.
Ants: Six studies found that glue or pesticide barriers reduced ant numbers in tree or vine canopies. One study found that citrus oil barriers had no effect.

Crops studied were cherimoyas, cherry, grape, grapefruitorange, pecan and satsuma mandarin.

Supporting evidence from individual studies


A randomised, replicated, controlled study in 1984-1985 in California, USA (Haney et al. 1987) found fewer spider mite destroyers Stethorus picipes in navel orange Citrus sinensis trees with insecticide barriers to exclude ants (0.2-0.8 destroyers/sticky card) vs. control trees (15-32 destroyers) in autumn, but no difference in summer and winter. The same effect was found in grapefruit Citrus paradisi, but not in Valencia oranges. Predatory mite Euseius tularensis numbers were similar between treatments. Fewer citrus red mites Panonychus citri occurred in trees with vs. without ant barriers (2-65 vs. 17-173 mites/tree, respectively) in orange orchards during peak numbers in late summer and the same effect was found in a grapefruit orchard during an autumn peak (approximately 75 vs. 200 mites/tree). Argentine ants Linepithema humile were successfully excluded from orange trees with insecticide barriers (0 ants/minute/tree with barriers applied vs. 14-158 in control trees) but a limonene citrus oil barrier had no effect (19-219 ants/minute/tree). Plots of orange trees (in two orchards) were assigned to insecticide (90 ml of 1% chlorpyrifos applied at the base of trees), citrus oil (135 ml of 15% limonene) or no-barrier treatments. Plots of grapefruit (in one orchard) were assigned to insecticide or no-barrier treatments only.


A randomised, replicated, controlled study in 1984 in California, USA (Phillips et al. 1987) tested three ant exclusion techniques and found lower Argentine ant Linepithema humile infestation when sticky band barriers (ant activity rating of 0.52) and chlorpyrifos insecticide (rating of 1.0) were applied than when baited traps (rating of 1.39) or no ant exclusion (rating of 1.43) were applied to cherimoyas Annona cherimola trees. Mealybug (namely Pseudococcus longispinus) infestation was closely related to Argentine ant activity but was similar (when tested statistically) between the ant exclusion treatments (infestation ratings of 0.1, 0.3, 0.5 and 0.7 for sticky band, chlorphyrifos, baited trap and no exclusion treatments, respectively). Sticky band barriers comprised of Tangle-trap aerosol sprayed in a band (of 3 inch-width) around tree trunks. In the insecticide treatment, chlorpyrifos was applied to the base of the trunk and surrounding 12 inches of soil. The baited trap contained a sugar/carboxymethylcellulose bait and Amdro pesticide. Treatments were replicated four times with three cherimoyas trees/replicate. Ant activity was rated 0 (< 1 ant/minute passing a point on the trunk) through to 5 (51-100 ants/minute). Mealybug infestation was rated 0 (no mealybugs) to 3 (over half of fruit surface area infested).


A randomised, replicated, controlled study in 1993 in Extremadura, Spain (Perez et al. 1995) found more natural predators on cherry Prunus sp. trees with ant-excluding glue (averaging 466-827 predators/100,000 aphids) than on trees treated with insecticide (42-238 predators) in June-July, and more than on untreated trees (94 predators/100,000 aphids) in June. Numbers were similar between treatments on other dates. Predators included ladybirds (Coccinellidae), flies (Chamaemyiidae and Syrphidae) and lacewings (Chrysopa sp.). Fewer aphids (Aphidoidea) occurred on trees with glue barriers (2,799-78,517 aphids/tree) and insecticide treatments (27-28,487 aphids) than on untreated trees (61,470-269,310 aphids) in May-July. Damage to foliage in October was similar in trees with glue barriers (249 shoots affected/tree), a March insecticide treatment (138 shoots) and no treatment (415 shoots), but an April insecticide treatment resulted in less damage (87 shoots). Glue barriers reduced ant (Formicidae) numbers vs. untreated and insecticide-treated trees (0-1 vs. 4-24 ants) in May-June but numbers later became similar when ants gained access to canopies via weeds and farm tools. Four treatments were replicated four times (one tree/treatment/replicate): glue applied around tree trunks, pirimicarb application (100 g/Hl) to tree canopies in March, pirimicarb application in April, and an untreated control.


A replicated, controlled study in 1989-1991 in Wakayama Prefecture, Japan (Itioka & Inoue 1996) found more parasitism of red wax scale Ceroplastes rubens by first generation wasps Anicetus beneficus on twigs where black garden ants Lasius niger were excluded with glue barriers (14% scale insects parasitised) than on twigs without glue barriers (5% parasitised). Second generation wasps parasitised marginally more scales (13.5%) on twigs with than without barriers (8%). Red wax scale survival rate was lower on twigs with (2.1%) than without (4.6%) ant barriers. Twigs with barriers had 1,582-3,122 young scales and 13-64 egg-laying adult scales, compared with 2,791-4,028 young and 166-187 egg-laying adults on twigs without barriers. Scale population increase was 10 times less on twigs with than without barriers over two years. A pair of one-year old twigs was selected on each of 12 satsuma mandarin Citrus unshiu trees in a 1 ha orchard area in June 1989. One twig in each pair received glue at the base of the stem to exclude ants. Scales were counted in August-October 1989 and in May-June 1990 to assess parasitism, which was determined by body colour. The experiment was repeated on nearby twigs (in the same trees) in 1990-1991.


A randomised, replicated, controlled study in 1993-1994 in four pecan Carya illinoinensis orchards in Alabama and Georgia, USA (Dutcher et al. 1999) found that beneficial insects were not affected by insecticide barriers to exclude ants in 1993. Ladybeetle (Coccinellidae) numbers were similar or showed inconsistent differences between trees with and without ant barriers in 1994, but at one site more were found in trees with (0.35-0.50 ladybeetles/bud) than without ant barriers (0.17-0.32 ladybeetles) in mid-May, when cover crops were also planted under trees. Barriers had no or inconsistent effects on aphids, for example fewer blackmargined aphids Monellia caryella occurred in trees with than without ant barriers on three sampling dates at one site, but the opposite was found on two other dates. Evidence from one of two sites showed that numbers of blackmargined aphids during the spring peak in 1994 were lowest when combining ant barrier treatments with cover crops and a foliage spray to attract natural enemies. Barriers successfully prevented red imported fire ant Solenopsis invicta from accessing pecan trees. Ant barriers were chlorpyrifos sprays (1 kg/ha) applied in spring in a 1 m-width band around tree trunks. Barriers were re-applied if ants were observed overcoming the barrier.


A randomised, replicated, controlled study in 1998-1999 in California, USA (Daane et al. 2007) found greater parasitism of mealybugs Pseudococcus spp. by parasitoid wasps (Encyrtidae) in grapevines Vitis vinifera with vs. grapevines without ant barriers (21-68% vs. 10-11% parasitised, respectively) in the Central Coast region. There was no effect in the North Coast region (0.00-0.02% vs. 0.004-0.005%). Fewer mealybug destroyers Cryptolaemus montrouzieri (introduced before the study) occurred in vines with vs. without ant barriers (0.07-0.09 vs. 0.29-0.31 mealybug destroyers/vine) in the Central Coast. The same effect occurred in the North Coast in 1999 but not 1998. Lacewing (Chrysopidae) numbers were unaffected in both regions (0.08-0.24 vs. 0.05-0.20 lacewings/vine). Fewer obscure mealybugs Pseudococcus viburni occurred in vines with vs. without ant barriers in the Central Coast (12-59 vs. 129-303 mealybugs/vine, respectively) and grape mealybugs Pseudococcus maritimus were similarly affected in the North Coast (6-28 vs. 54-69 mealybugs). Fewer Argentine ants Linepithema humile occurred in vines with barriers (0.0-0.9 ants/2 minutes/vine) vs. vines without barriers (25-39 ants). Barriers were made by stripping bark from vine trunks and covering the exposed wood with duct tape coated in Tanglefoot Pest Barrier (re-applied when necessary). Three vineyards across two regions were studied, with six replicates/site.


A replicated, controlled study in 2008-2009 near Bern, Switzerland (Stutz & Entling 2011) found fewer earwigs Forficula auricularia (enemies of black cherry aphid Myzus cerasi) on trees with glue barriers to exclude ants (0.2 earwigs/tree) than trees without barriers (2.1 earwigs). More hoverfly (Syrphidae) eggs and larvae were found on trees with than without barriers in mid-May (1.9-3.3 vs. 0.6-2.1 eggs or larvae/twig, respectively) but the opposite occurred in late May-early June (0.0-1.0 vs. 0.6-1.9 eggs or larvae/twig). Barriers reduced the total number of ladybirds (Coccinellidae) counted across the season (late April to early June 2009) but differences between treatments were not consistent across individual sampling dates. Fewer ants (Formicidae) occurred on trees with barriers at all sampling dates (0.0-0.2 ants/twig with barriers vs. 0.7-8.2 ants without barriers) and aphids were also fewer from mid-May to early June (0-25 vs. 90-360 aphids/twig). Wild cherry trees Prunus avium with barriers grew more new wood than trees without barriers. Four-year-old cherry trees were planted at 30 sites in spring 2008 and black cherry aphids were released onto four trees/site in mid-April 2009. A 7 cm-width glue ring was attached around the main stem of two trees and renewed monthly to exclude ants.

Referenced papers

Please cite as:

Wright, H.L., Ashpole, J.E., Dicks, L.V., Hutchison, J., McCormack, C.G. & Sutherland, W.J. (2017) Some Aspects of Enhancing Natural Pest Control. Pages 359-381 in: W.J. Sutherland, L.V. Dicks, N. Ockendon & R.K. Smith (eds) What Works in Conservation 2017. Open Book Publishers, Cambridge, UK