Companion planting and what garden research says actually works
Companion planting can reduce pest pressure in some settings, but controlled studies support the broader benefits of plant diversity more consistently than the specific crop pairings promoted in garden charts. A meta-analysis of 552 experiments reported in 45 articles found that diversified plantings suppressed herbivores, raised numbers of their natural enemies and reduced crop damage more than undiversified plots (Letourneau et al., 2011).
This explainer separates the part that has been tested from the part that has been copied between charts: what the pooled studies actually measured, why mixed plantings interfere with insects, which garden pairing has been investigated for a plausible chemical mechanism, where the yield evidence is strongest, and which claims have no measurement behind them. Further coverage of crops sits in the [plants section](/plants/).
What the pairing charts rest on
The familiar chart, with its columns of friends and enemies, is a compilation rather than a dataset. Entries circulate between gardening sources without a trial attached, and the strongest of them, marigolds beside tomatoes, was described in print for decades before anyone measured it in a glasshouse (Conboy et al., 2019). That gap between the claim and the measurement is the subject of this article.
The scientific literature does not use the phrase in the same way. It studies diversification: intercropping, polycultures, trap crops, flowering strips, and the behaviour of insects moving through mixed vegetation. Those terms describe what is actually tested, and the results map onto the garden practice only in places. The same questions at the scale of a whole farmed landscape are set out in [biodiversity and insect pest management](/environment/biodiversity-insect-pest-management-agroecosystems-environment/).
The distinction matters because the two make different promises. A chart promises that a named partner protects a named crop. The research mostly shows that mixing plants of any suitable kind changes how insects find a crop, which is a weaker and more general claim.
There is also a practical reason the two diverge. A garden bed is small, mixed and observed by eye, while the studies work with plots large enough to separate an effect from the ordinary variation between seasons. Effects that are real but modest, which is what the pooled results describe, are close to invisible at the scale of one bed in one summer.
What the pooled studies measured
The broadest synthesis gathered 552 experiments reported in 45 articles over a decade and tested the standard explanations against them. Herbivore suppression, enhancement of natural enemies and suppression of crop damage were all significantly stronger on diversified crops than on crops with little or no diversification, with support for intercropping schemes, the inclusion of flowering plants, and plants chosen to repel herbivores or draw them away (Letourneau et al., 2011).
A second meta-analysis asked the harder question, whether those gains come free. It selected only studies that recorded both pest control and yield, 26 studies and 301 observations, and found win-win relationships between per-plant yield and biocontrol in polycultures that minimised competition between plants through substitutive planting. Where planting was additive, at high density, biocontrol improved with no difference in yield per unit area when a legume was the second crop (Iverson et al., 2014).
The evidence favours mixing plants, not matching partners. Charts promise the second and cite the first.
Read together, the two reviews give a defensible position: diversification reliably changes insect numbers, and whether it also costs yield depends on how the planting is arranged rather than on which species are paired. Letourneau and colleagues also reported a small but significantly negative mean effect on crop yield, partly because a pest-suppressive mix replaces plants of the main crop. Neither review supports the chart's implied precision.
Why mixed plantings interfere with insects
Two explanations have dominated since the 1970s. The resource concentration hypothesis came out of a three-year study of collards grown either in pure stands or in single rows surrounded by meadow vegetation, sampled on 20 dates; specialist herbivores were less abundant where the host plants were harder to find and to stay on (Root, 1973). The enemies hypothesis holds that mixed vegetation supports more predators and parasitoids.
A later theory set both aside in favour of insect behaviour. Searching insects land indiscriminately on green objects, including non-hosts, and whether they stay depends on what they find after landing; in diverse vegetation a larger share of landings are inappropriate, so fewer insects settle and lay eggs (Finch & Collier, 2000). The practical consequence is that green ground cover can interfere with host finding in the pests this theory was built on, which is awkward for a chart built on specific partners.
Trap cropping gives the clearest behavioural evidence. In a trial using mixtures of brassica trap crops at constant total area, only the three-species mixture improved the yield of adjacent broccoli, and it outperformed a monoculture of the single most attractive trap species. Flea beetle numbers did not differ between the diverse and simple trap plantings, which points to a change in pest behaviour rather than in pest abundance (Parker et al., 2016).
The marigold and tomato pairing, tested
This is one of the few familiar garden pairings investigated experimentally for a plausible chemical mechanism. Two large glasshouse trials tested marigolds against glasshouse whitefly on tomato. Where marigolds grew alongside tomato through the season, whitefly population development over a 48-day infestation was significantly slowed. Reducing the proportion of marigolds and adding other non-host plants, basil, nasturtium and Chinese cabbage, also reduced whitefly numbers, while adding attractive plants around the plot perimeter had little effect (Conboy et al., 2019).
Laboratory experiments identified airborne limonene from French marigolds as a probable mechanism of whitefly repellency. Tests with limonene dispensers provided further support for its role, although the findings do not establish it as the only factor involved (Conboy et al., 2019). The second trial tested marigolds introduced after whitefly numbers had already built up, which is how many gardeners would reach for them, and there the effect was minimal: once whitefly had built up only the limonene dispensers still worked.
What the result does not license is the general form of the claim. It concerns one pest, one crop and an enclosed glasshouse, and the finding that unrelated non-hosts also helped points back to the landing behaviour described above rather than to anything specific about marigolds.
The glasshouse setting deserves one more note, because it is doing work in the result. The experiments took place in enclosed glasshouses, where exposure to plant volatiles may differ from outdoor conditions. The study did not establish whether the same level of protection would occur in open gardens (Conboy et al., 2019). The trial tested a mechanism rather than a gardening instruction, and the difference between the two is the distance between an identified compound and a chart entry (Conboy et al., 2019).
Where the yield evidence is strongest
Intercropping has the firmest numbers, and they come from agriculture rather than from gardens. Across 939 observations in 126 studies, intercrops returned a mean land equivalent ratio of 1.30, with a confidence interval from 1.27 to 1.32, meaning that the intercrops used 23% less land than monocultures producing the same output. The intercrops produced 38% more gross energy and 33% more gross income, and the advantage held under both irrigated and rain-fed conditions (Martin-Guay et al., 2018).
Study | What was measured | Reported result |
Letourneau et al. (2011) | 552 experiments in 45 articles | Herbivore suppression, enemy enhancement and damage suppression stronger on diversified crops |
Iverson et al. (2014) | 26 studies, 301 observations | Win-win yield and biocontrol under substitutive planting |
Martin-Guay et al. (2018) | 939 observations, 126 studies | Mean land equivalent ratio 1.30, 38% more gross energy |
Conboy et al. (2019) | Two glasshouse trials, tomato and whitefly | Marigolds slowed whitefly build-up; limonene identified as a probable repellent mechanism |
Parker et al. (2016) | Brassica trap crop mixtures at constant area | Only the three-species mix raised broccoli yield |
Note: these are field and glasshouse measurements on named crops and pests, not endorsements of particular pairings for a garden bed. The land equivalent ratio describes land use efficiency across a whole intercrop, not the yield of one plant beside another.
The three sisters, maize with bean and squash, has been examined at the level of roots rather than folklore. In a root-architecture simulation of the first 40 days of growth, nitrate uptake and biomass production were up to 7% greater in the polycultures than in the corresponding monocultures, but only when differences in root architecture were taken into account, and the gain was independent of nitrogen fixation by the bean. Competition between roots had negligible effect on phosphorus or potassium uptake (Postma & Lynch, 2012).
That is a modest and conditional advantage, and it belongs to mobile soil resources. It is also a reminder that the mechanism behind a traditional planting is often not the one the tradition proposes, since the nitrogen explanation usually offered for the bean was not what produced the measured gain.
A later study of the same system looked at how roots forage rather than at architecture alone. Yield advantage in the three sisters tracked niche complementarity between the three root systems, which places the benefit in how the species divide the soil rather than in any pairwise relationship between them (Zhang et al., 2014).
What is claimed without measurement
Several chart entries are not covered by the trials in these reviews at all. Claims that a herb improves the flavour of a neighbouring fruit, that particular pairs must never share a bed, or that a single border plant protects a whole plot, do not appear in the reviews above, and the reviews are the places such results would be collected if they existed (Letourneau et al., 2011; Iverson et al., 2014).
Nematode suppression by marigolds sits between the two camps. It has a literature, and the effect depends on the marigold species and cultivar, the nematode species and how the planting is managed, which is a long way from the single line a chart gives it (Hooks et al., 2010). Closely related work on other cover crops against root-knot nematodes is summarised in the review of [calendula, crotalaria and oats against root-knot nematodes](/agriculture/calendula-crotalaria-oats-root-knot-nematodes-naranjilla-agriculture/).
The reviews of companion plants for aphid management reach a similar conclusion in plainer terms. Volatiles from companion plants can disturb host location, companion plants can shelter natural enemies, and many factors have to line up for the strategy to work, so it is treated as one tactic to combine with others rather than as a replacement (Ben-Issa et al., 2017).
The aphid review is also explicit about the conditions. Companion plant volatiles have to reach the pest at the moment it is searching, the companion has to be unsuitable as a host itself, and the shelter it provides has to suit the natural enemies that are present, which is a longer list of requirements than a chart entry implies (Ben-Issa et al., 2017). Where those conditions are not met, the same planting produces no measurable effect, and a garden has no straightforward way to tell the two situations apart.
None of this makes the practice pointless. It makes the expectation the problem: the measured benefit is a shift in the odds across a season, not the clean protection a pairing table promises, and it arrives through mechanisms that are indifferent to which plant provides the interference.
What the evidence supports and what it does not
Supported: diversified planting suppresses herbivores and raises natural enemy numbers across a large body of experiments; intercropping uses land more efficiently than monoculture; marigolds slow whitefly build-up on tomato in glasshouse trials, with limonene identified as a probable mechanism; and trap crop mixtures can protect an adjacent crop (Letourneau et al., 2011; Martin-Guay et al., 2018; Conboy et al., 2019; Parker et al., 2016). Across 939 observations, intercrops returned a mean land equivalent ratio of 1.30 (Martin-Guay et al., 2018).
Not supported: the precision of the pairing chart. The mechanism most consistent with the behavioural evidence is indiscriminate landing in mixed vegetation, which predicts that many unrelated non-host plants will work and that no particular partner is required (Finch & Collier, 2000; Conboy et al., 2019). The economics of changing a whole system are a separate question, examined in the review of [organic against conventional farming](/agriculture/organic-vs-conventional-farming-economics-ecology-agriculture/).
Unsettled: whether the garden-scale version reproduces the field-scale result. Almost all the quantitative work comes from farm plots or glasshouses with a single focal crop and pest, and the trade-off analysis shows that outcomes depend on planting arrangement and density rather than on species identity (Iverson et al., 2014). The honest summary is that mixing plants is well supported and the chart is not.
Frequently asked questions
Do marigolds really protect tomatoes?
Against glasshouse whitefly there is evidence. One glasshouse trial found that marigolds grown alongside tomato slowed whitefly population development over a 48-day infestation. Laboratory experiments identified airborne limonene as a probable contributor to this effect. Other non-host plants reduced numbers too, which weakens the claim that marigolds are special. (Conboy et al., 2019)
Does companion planting increase yield?
At field scale, mixing crops does. Across 939 observations the mean land equivalent ratio was 1.30, so the intercrops used 23% less land than monocultures producing the same output. That figure describes the intercrop as a whole, not the yield of one plant placed beside a chosen partner. (Martin-Guay et al., 2018)
Why do the charts disagree with each other?
Because most entries were never measured. The research literature studies diversification rather than named pairs, and the reviews that would collect pairing results report general effects of mixing instead. Where two charts conflict, there is usually no trial that could settle the disagreement.
Is it an alternative to other pest control?
Reviews do not present it that way. Companion plants can disturb how a pest locates its host and can shelter natural enemies, but many conditions have to align, so the literature treats the approach as one tactic to be combined with others rather than as a standalone method.
Does the three sisters planting actually work?
In a root-growth simulation, modestly. Nitrate uptake and biomass were up to 7% higher in the polyculture than in monocultures, but only once differences in root architecture were accounted for, and the gain did not come from nitrogen fixed by the bean, which is the usual explanation given. (Postma and Lynch, 2012)
Sources
- Letourneau DK, Armbrecht I, Rivera BS, et al. Does plant diversity benefit agroecosystems? A synthetic review. Ecological Applications. 2011;21(1):9–21. DOI 10.1890/09-2026.1.
- Iverson AL, Marin LE, Ennis KK, et al. Do polycultures promote win-wins or trade-offs in agricultural ecosystem services? A meta-analysis. Journal of Applied Ecology. 2014;51(6):1593–1602. DOI 10.1111/1365-2664.12334.
- Conboy NJA, McDaniel T, Ormerod A, et al. Companion planting with French marigolds protects tomato plants from glasshouse whiteflies through the emission of airborne limonene. PLOS ONE. 2019;14(3):e0213071. DOI 10.1371/journal.pone.0213071.
- Root RB. Organization of a plant-arthropod association in simple and diverse habitats: the fauna of collards (Brassica oleracea). Ecological Monographs. 1973;43(1):95–124. DOI 10.2307/1942161.
- Finch S, Collier RH. Host-plant selection by insects, a theory based on appropriate and inappropriate landings by pest insects of cruciferous plants. Entomologia Experimentalis et Applicata. 2000;96(2):91–102. DOI 10.1046/j.1570-7458.2000.00684.x.
- Martin-Guay MO, Paquette A, Dupras J, Rivest D. The new Green Revolution: sustainable intensification of agriculture by intercropping. Science of the Total Environment. 2018;615:767–772. DOI 10.1016/j.scitotenv.2017.10.024.
- Postma JA, Lynch JP. Complementarity in root architecture for nutrient uptake in ancient maize/bean and maize/bean/squash polycultures. Annals of Botany. 2012;110(2):521–534. DOI 10.1093/aob/mcs082.
- Zhang C, Postma JA, York LM, Lynch JP. Root foraging elicits niche complementarity-dependent yield advantage in the ancient ‘three sisters’ (maize/bean/squash) polyculture. Annals of Botany. 2014;114(8):1719–1733. DOI 10.1093/aob/mcu191.
- Hooks CRR, Wang KH, Ploeg A, McSorley R. Using marigold (Tagetes spp.) as a cover crop to protect crops from plant-parasitic nematodes. Applied Soil Ecology. 2010;46(3):307–320. DOI 10.1016/j.apsoil.2010.09.005.
- Parker JE, Crowder DW, Eigenbrode SD, Snyder WE. Trap crop diversity enhances crop yield. Agriculture, Ecosystems and Environment. 2016;232:254–262. DOI 10.1016/j.agee.2016.08.011.
- Ben-Issa R, Gomez L, Gautier H. Companion plants for aphid pest management. Insects. 2017;8(4):112. DOI 10.3390/insects8040112.
Related
Calendula, crotalaria and oats for root-knot nematodes in naranjilla
Organic vs Conventional Farming: Economics and Ecology
Biodiversity for Better Insect Pest Control in Agroecosystems
New Zealand Ferns and Allied Plants: A 1989 Field Guide
How to Grow Hydnophytum and Myrmecodia Ant-Plants
Powdery mildew and why this fungus spreads across plant leaves
Nitrogen fixation and how legumes and soil bacteria feed plants