Introduction

Crop species diversification, such as intercropping, is receiving increasing attention as ecological practice to sustainably improve agricultural systems1,2,3. Intercropping systems (Box 1)—i.e., the cultivation of two or more crop species within the same field coexisting for at least part of their growing season4—can have major benefits for yields, land-use efficiency, nutrient and fertilizer-use efficiency, pest and disease suppression and soil fertility compared to monocropping systems5,6,7,8. However, there is also enormous variation in the effects of intercropping on these ecosystem benefits, partly determined by climatic condition, soil status, cropping design, crop species and cultivar used for each crop (see various meta-analyses5,9,10,11,12,13). More research is needed to decipher the effect of these determinants and understand the mechanisms underlying the benefits of intercropping. This knowledge can support and inform farmers and breeders on the potential of intercropping as a sustainable agricultural practice.

One important determinant of intercrop performance is the combination of crop species or cultivars within the intercrop. In intercrops, the diversity of crop species within the field also increases the diversity of environmental conditions within the field in time and space compared to monocrops (Fig. 1). This diversity in conditions can induce a variety of structural and physiological responses within crop plants, called plastic responses (Box 2), causing phenotypic differences between plants from different growth systems, referred to as phenotypic plasticity 14,15. Plastic responses affect crop growth and therefore makes predicting the performance of crop–crop combinations challenging. Depending on the situation, phenotypic plasticity is often referred to as phenotypic differences in intercrop compared to monocrop, while plastic responses can refer to phenotypic differences induced by environmental signals from intercrop systems that are compared to signals perceived by a plant growing alone without intra or interspecific interactions (see Box 2). Over the last decades, agronomists and ecologists have been studying the effects of diversifying ecosystems but have largely ignored the role of plasticity in determining crop and vegetation performance. At the same time, eco-physiologists have been focusing on plasticity in relation to heterogenous environments but have not been including species diverse systems. Here, we want to bridge these fields and illustrate that plastic responses could be important determinants of the performance of diverse systems such as intercropping fields.

Fig. 1: Relationship between heterogeneity in field conditions and the associated benefit of plasticity, against the level of interspecific interactions in the field, for five different cropping systems.
figure 1

Monocrops have no interspecific interactions and a relatively low variation in environmental conditions within the field due to a homogeneous canopy and root distribution. Limited benefit is to be expected from plasticity because all plants need to express the similar traits. Relay strip intercrops (clock icon represents a time difference between sowing and harvesting between the component crops) have increased interspecific interactions due to presence of two crop species, even though both species grow without the companion species for some time. Heterogeneity in field conditions is very high due to species differences and temporal segregation, and so is the potential benefit of plasticity. Strip intercrops in which the co-growth period is maximal have higher levels of interspecific interactions and decreased heterogeneity in the field than relay systems. Therefore the benefits of plasticity in non-relay strip intercrops are lower than in the relay strip intercrop. Row intercrops have a higher level of interspecific interactions than the strip systems and therefore a lower heterogeneity in the field, by which the benefit of plasticity is also lower. Full mixtures represent maximal levels of interspecific interactions due to species mixing between and within rows. Heterogeneity in conditions is higher than in monocrops but lower than in row intercrops. Benefits of being plastic are therefore low but higher than in monocrops. In this categorization, with ‘heterogeneity’ we refer to the environmental conditions induced by crop species, not intrinsic environmental conditions of soil structures or nutrients that are present in any agricultural field irrespective of cultivation design. Plasticity can be beneficial in monocrop fields with high intrinsic soil heterogeneity or under fluctuating weather events, which is outside the scope of this story.

Full size image

Plastic responses within diverse crop systems can relate to changes (in time and space) in light conditions and resource availability but also to changes in the composition of soil microbial communities, weeds and insect populations. However, the extent to which crop plants respond to changes in these conditions and what this would mean for the functioning and performance of intercrops is not clear. On the one hand, plastic responses could increase competition for resources and thus suppress one of the component species. On the other hand, the plastic responses could enable plants to take advantage of favorable conditions and increase resource uptake which improves growth of all component species. Within this review we outline (i) the relationship between plasticity, the degree on interspecific interactions and the heterogeneity in conditions within an intercrop, (ii) the current evidence for plasticity observed in crop plants in intercropping systems and the consequences for their performance and (iii) outline the diversity of signals that occur in intercropping that could induce plastic responses.

Intercropping systems and heterogeneity in environmental conditions

The term intercropping is used for a variety of mixed crop designs8, and each of these designs comes with its own specific environmental heterogeneity in time and space (Fig. 1). Here, we refer to environmental heterogeneity induced by the cultivated crop species, not intrinsic environmental heterogeneity of soil structure or nutrients that are present in any agricultural field irrespective of cultivation design. Full mixtures, in which the component species are fully mixed and sown simultaneously, have a high level of interspecific plant-plant interactions. At the field scale however, heterogeneity of environmental conditions is relatively low since the aboveground canopy and belowground rooting is homogeneous in its pattern. Each individual plant of one species will experience similar conditions due to the homogenous mixing of species. If all plants of the same species experience similar conditions, plastic responses are less important, as plants should all express similar trait values. Compared with full mixtures, interspecific interactions are lower in alternating row designs because of the intraspecific interactions within the row, but there is usually higher canopy and root environmental heterogeneity due to the alternating rows of species that differ in phenotype. Strip intercropping systems have a considerably lower level of interspecific interactions than row intercrops depending on the number of rows within one species strip, since interspecific interactions happen only at the strip edges where two plant species meet. Strips have large heterogeneity in environmental conditions in space due to different combinations of rows with similar or different species. In strip intercropping, plants in the inner rows of a strip experience different conditions than plants in the border rows, and this experience of different conditions is even larger in relay strip intercropping designs. If individual plants within one field experience different conditions, plastic responses could enable them to express different trait values in both situations. Therefore the ability of species to show plastic responses can be more beneficial when individuals within the field experience different environmental conditions. So far, only a few studies have quantified different phenotypes within one field and related this to environmental heterogeneity within the field16,17, and no studies have been able to compare whether more plasticity would occur in more or less heterogeneous intercrops. The heterogeneity of environmental conditions is partly related to abiotic conditions such as light and nutrient availability. However, biotic conditions such as the presence of microbial, weed and insect communities are also more heterogeneous in intercropping systems. This heterogeneity in environmental conditions that characterizes intercrops can induce various plastic responses which enables plants to take advantage of beneficial conditions or deal with harmful conditions. Overall, high heterogeneity of environmental conditions in time and space in the field is a reason why phenotypic plasticity could be beneficial for crop species growing in intercropping.

Evidence of plasticity in intercrops and its consequences for performance

In (relay) strip intercropping designs, plants growing in the rows directly adjacent to a row of the companion species (i.e. border rows) experience different conditions than plants in the middle of the strip and in a monoculture. These local conditions can be beneficial, such as improved abiotic conditions like light intensity or nutrient availability, leading to plastic responses such as increased tillering and leaf area growth in cereals; responses that allow the plants to actively increase resource acquisition. For instance, in wheat-maize relay intercrop systems with wheat being sown first, wheat plants in the borders of the strips next to absent or young maize seedlings produce up to four times the number of fertile tillers, produce twice larger flag leaves and orient their leaves towards the maize strip compared to plants growing directly next to conspecifics only. Such plastic responses can also be observed belowground, with one species reaching higher root length densities across the soil profile when grown next to another species than when growing in monoculture stands, such as wheat with maize18, maize with soybean or peanut19 and wheat with soybean20. These plastic responses can be categorized as resource-foraging responses, with opportunities for extra resources offered by the intercrop context.

By contrast, the conditions created by intercrops can also be potentially harmful to one of the component species, such as heavy shading or competition for soil resources. Similarly, these conditions may elicit plastic responses, but in this case the plastic responses may allow the plants to mitigate the detrimental effect of competition and escape adverse conditions created by the companion crop. Soybean plants experience considerable competition for light by neighboring maize plants in case the strips of both species have been sown simultaneously, and this leads to typical competition avoidance responses by soybean such as increased stem length and higher specific leaf area, in an attempt to avoid the shading and maximize photosynthetic surface area21,22,23. Also, leaves on maize plants in similar maize-soybean systems may increase their specific leaf area in order to improve light capture24. Such plants also show further physiological responses to competition in intercrops, like a reduction in carotenoids and an increase in chlorophyll b, which improves the efficiency of carbon assimilation at low light.

Foraging for resources and escaping competition are typical for strip systems, in which plants in the border rows show strongest deviations from typical crop phenotypes surrounded by the same species in middle rows or monoculture systems. However, in row cropping and full mixtures, plastic response related to foraging of resources or avoiding competition are also relevant. If crop plants increase the difference in root placement through plastic responses, belowground competition is reduced which could increase resource uptake and improve crop growth. In addition, in mixtures of species with differential stress tolerance the more tolerant species can, through plasticity, better utilize the greater resource availability created by the lesser growth of the less tolerant species hence enhancing yield stability. Crop species have been observed to express trait plasticity (e.g. fewer nodes, or increased tiller numbers) in fully mixed legume-cereal intercrops compared to sole crops, resulting in greater trait space in intercrop than in sole cropping25. However, this observed phenotypic plasticity has not been linked to increased foraging or decreased competition responses, which makes their effects on crop and system performance ambiguous. Plastic responses may also be caused by changed levels of intraspecific competition. This is the case for instance in specific strip intercrop designs in which plants of one species are sown at a higher density within their strips than in sole crops24. Overall, comparing crop traits from various mono- and intercrop systems, determining crop plasticity and quantifying how plasticity influences crop performance in intercrops has not yet been done, but is needed to identify the potential of plasticity for intercropping. This in turn is needed to understand the observed differences in crop growth and performance in different intercrop systems26,27,28,29.

All plastic responses may impact the performance of the plants in an intercrop, which can have consequences for the performance of the whole intercrop. Relay intercrop systems with a relatively short overlap in growth duration between the component species, provide extra resources to the component crops, to which they can respond plastically resulting in increased productivity. In wheat-cotton intercrops, the species have a co-growth period of less than two months, resulting in the later-sown species (cotton) to make full use of the space left once the first-sown species (wheat) has been harvested, by almost completely covering the empty companion strip30. This results in radiation interception advantages over sole crops that lead to land-use efficiencies of up to 1.5 higher than sole crops. Also, systems with a longer co-growth period than cotton-wheat, such as maize-wheat relay intercrops, may be more resource-efficient than the respective sole crops, part of which can be attributed to resource-foraging responses16. If competition-avoidance responses mitigate the negative effects of a dominant component species on a subordinate species, overall system performance may surpass sole crops if the dominant species does well.

The exact contribution of plasticity to the performance of an intercrop, or of a component species of an intercrop, cannot be easily quantified experimentally. Next to plastic responses, intercrop performance benefits may arise, in relay systems, from the longer total growth season and the associated higher total amount of resources available. Benefits may also come from the effects of species composition on pollinator, insect herbivore and parasitoid communities, on soil micro-organisms communities, among others4,8,31. To extract and quantify the effect of plasticity on intercrop functioning and performance, plant modeling has been instrumental32,33. Running virtual experiments, using computer simulations in which the performance of plants with phenotypes representative of plastic responses to mono and intercrop conditions are compared, gives the opportunity to isolate the contribution of the plastic responses to crop performance. Specifically for light, this approach has proven useful to both resource-foraging and competition avoidance responses17,34 and allows for virtual explorations of the plant trait space to find high-performing phenotypes35 with particular plastic responses especially suitable for intercrops, above- and/or belowground.

Diversity of signals in intercrops

In the previous section, we considered that intercrops have different environmental conditions than monocrops, and that the constituent plants respond to these conditions. Here, we explore which signals associated with these environments may induce plastic responses. As mentioned before, the heterogeneity of environmental conditions that plants experience in intercrops may differ profoundly from those experienced in monocrops and hence the type, magnitude and spatial and temporal distribution of specific signals that plants receive may also differ considerably. Yet surprisingly little is known about all possible signals in intercrops that could induce plastic responses and result in phenotypic plasticity. We propose that environmental signals could be categorized into three groups (Fig. 2). The first two groups consist of direct abiotic signals that are either generic or specific. Abiotic signals are considered a direct link between two component plants in the environment. Generic signals are independent of the plant species that produce them and only the magnitude and distribution could depend on the intra-and interspecific interactions. Specific signals say something about the identity of the sender. The third group consists of biotic signals that involve organisms other than the interacting crops, and these signals are therefore indirect.

Fig. 2: Identification of signals that can be direct generic or specific abiotic signals or indirect signals via biotic interactions.
figure 2

Direct generic signals are related to abiotic factors such as chemical compounds like exudates and volatiles and light quality such as the red to far-red ratio (R:FR), resources such as nutrients and mechanical stimuli. Direct specific signals are chemical compounds in exudates or as volatiles that have a specific composition that can signal a specific crop species. Indirect biotic signals are signals from a crop species that are mediated by microbes, invertebrates and arable flora within the system.

Full size image

Generic direct abiotic signals

The most obvious of these are direct signals from resource availability. As species differ in both architectural and physiological traits, they also differ in the amount and spatial distribution of resource uptake. Hence, the amount and distribution of light, water and nutrients will be different in intercrops than in monocrops and signals from this distribution can be generic for all species. Direct plastic responses to changes in resource availability are now textbook material36 and need no further discussion here. However, plants can also affect resource distribution by modifying the abiotic environment in other ways, for instance, the ability of plant species to mobilize phosphorus through root-secreted protons, organic acids (carboxylates) and acid phosphatase37. There is substantial interspecific variation in this phosphorus-mobilizing capacity. Several legume species like faba bean38 and soybean39, tend to have a relatively high phosphorus mobilizing capacity. In a root choice experiment Zhang et al.39 showed that tomato plants allocated roots towards soybean plants but not towards other species. As soybean was also the only species increasing soil mobile phosphorus, it suggests that the plastic response of tomato was linked to the phosphorus mobilizing effect of soybean. However, it is not clear what the exact signal might have been; were tomato roots simply responding to a positive phosphorus gradient towards soybean or did they detect and respond to soybean-produced exudates?

Plants also produce non-resource-linked generic cues such as changes in light quality or microclimate. The best-known example are the changes in light spectral distribution – the reduction in the ratio of red to far-red light (R:FR) and in blue light – associated with selective light absorption by leaves40. Plastic responses to these changes include internode elongation, leaf hyponasty and reductions in branching and are known as shade-avoidance41,42. As plant species differ in canopy size and structure the temporal and spatial dynamics of these light signals depend on the species that plants interact with. Plants also produce relatively generic volatiles (e.g. ethylene) and exudates (e.g. primary metabolites). Plants e.g., respond to low levels of ethylene through shade-avoidance type responses, and variation in these responses is known to strongly influence the competitive balance between plants43. Finally, plants can also respond to mechanical stimuli induced by wind or direct touch44. Compared to monocrops, plant within intercrops can experience different wind speeds; tall stature edge plants can experience increased wind speeds, while small inner plants can experience lower speeds45. Plant-induced touch is important for climbing species that use tendrils to capture structures and for plants that are being climbed on. However, to our knowledge no study has quantified to which extent mechanical-induced plasticity occurred in intercropping systems, it is also very challenging to detangling the contribution of these signals from other signals in an intercropping field.

Specific direct abiotic signals

There is evidence that within species, plants can detect the extent to which other plants are genetically related to themselves, so called kin detection. Plants have been found to produce smaller, less-branched root systems46,47 with lower water and nutrient uptake rates48, and less shade-induced stem elongation49 when plants interacted with kin than when they interacted with non-kin. Furthermore, exposure of Arabidopsis thaliana, rice and Deschampsia caespitosa plants to root exudates of either more-closely or more-distantly related con-specific plants induced similar responses as the responses to non-kin50. Similarly, exposure of barley plants to aboveground volatile organic compounds (VOC) taken from plants of a different variety induced greater allocation to roots than exposure to VOCs from the plants of the same variety51. In addition, exposure of Artemisia tridentata plants to VOCs showed stronger defense responses when these VOCs came from more-related plants than from less-related ones52. Together these results support the idea that root exudates and/or VOCs signal direct information about relatedness. This is likely associated with the actual VOC/exudate compounds or their blends being produced by some plants but not by others. However, which specific chemicals or combinations of chemicals present in exudates or VOCs are involved in the actual recognition process is still completely unknown.

Intriguingly, as far as we know, very little research has been done to determine whether identity recognition via VOCs or exudates also plays a role in the interaction between (crop) species. Hall et al.53 exposed plants of soybean and wheat to exudates and VOCs derived either from conspecifics or from the agricultural weed common ragweed (Ambrosia artemesiifolia). Wheat and soybean produced less biomass and leaves with lower chlorophyll contents when they were exposed to ragweed VOCs, but exposure to their own VOCs had no effects. Furthermore, in a root-choice experiment where root systems of wheat and soybean plants were divided over two compartments containing either their own exudates or ragweed exudates, they preferentially avoided the compartment with ragweed exudates. These experiments suggest that wheat and soybean can detect the difference between conspecific and heterospecific VOCs/exudates, but not whether plants can detect the differences in VOCs/exudates between different species or functional groups, and, as with the abovementioned kin recognition, what combination of chemicals are involved in the actual recognition.

Indirect biotic signals

Plant species can differ substantially in their interaction with other organisms—bacteria, fungi, nematodes, insects or other non-crop plants (i.e., weeds)—in the community both in terms of the species with which they interact and the nature of the interaction54,55. Hence, species composition and population structures of the agro-ecological community may differ considerably between monocrops and intercrops. These organisms, in turn, can differ in the effects they can have on plants including resource competition, decomposition, symbioses, pathogenic effects and herbivory and predation. Furthermore, these organisms can also be used by plants to communicate with each other (e.g. via mycorrhizal networks56) or release organic compounds that have a signaling function towards plants (e.g. in the rhizosphere57). The extent to which plants respond to non-crop organisms, use common mycorrhizal networks or respond to organic compounds from other organisms is a research topic on its own. Yet, the extent to which crop plants show plasticity induced by these biotic signals in intercrops is still unknown.

Clearly, various types of signals could manifest in intercropping systems at the same time, all inducing plastic responses in crop plants. Disentangling the different signals and the associated responses is crucial to understand the phenotypic plasticity observed of crops, and to recognize species interactions and their associated performance in intercrops.

A first step to determine the extent to which plants respond to direct abiotic or indirect biotic signals could be based on so-called training/feedback experiments in which in the feedback stage plants are grown on soil where another species was previously grown (training stage) and which contains soil biota associated with that species58 but also contains the remains of nutrients and/or exudates in the soil. This setup could be modified to separate the two by adding two treatments59: a ‘biotic’ treatment inoculating a standard sterilized soil with a small amount of the training soil hence creating the biotic but not abiotic composition of training soil, and an ‘abiotic’ treatment using sterilizing training soil for feedback hence providing the abiotic but not the biotic composition. If applied to intercropping one could determine the extent to which plasticity and changes in performance are associated with either abiotic and biotic differences between mono- and intercropped soil. But it still would not pinpoint the specific environmental signals that would induce plasticity. To this end, this approach would have to be elaborated. This could involve manipulative experiments e.g. modifying nutrient composition, pH or biotic characteristics of the training and/or feedback soils, combined with using a wide variety of species in the training phase. Use of genomic60 or metabolomic approaches61 could respectively help identify the role of different (functional groups of) organisms and chemical compounds and determine if signals could be generic or specific. But given the probably very large number of factors that could play a role the size and complexity of experiments involved will certainly be an issue.

Concluding remarks from plasticity to sustainable agriculture

As shown in Fig. 1, the heterogeneity in environments and thus the relative importance of phenotypic plasticity depends on planting pattern of intercrops; it being more important in (relay) strip intercropping with more variation in inter vs intraspecific interactions than in full mixtures. However, no direct proof of this concept is yet available. The current proof of phenotypic plasticity and its consequences for crop performance also mainly comes from strip intercropping studies. This raises the question whether different intercrop designs would require crop genotypes with different levels or types of plasticity. In addition, the observation that plasticity could have various roles (resource-foraging versus competition-avoidance) and consequences (affecting yield or resource-use efficiency) in intercropping has implications for the design and implementation of intercropping as means towards more sustainable crop production. Overall, a key question here is to what extent plastic responses make species interactions more positive (weaker competition or stronger facilitation) or more negative. In the former case breeders would tend to favor more plastic genotypes and in the latter case less plastic ones. If plasticity of a given crop variety leading to more positive interactions would be general, being beneficial irrespective of companion species or planting pattern, this could help simplify breeding for diversity62, since the need to develop specific varieties for different intercropping systems would be reduced. This would be more likely to occur if the most important signals for plastic responses are generic rather than specific. On other hand, if plastic responses of a crop variety leads to both positive and negative interactions depending on the companion species and/or the cropping design, breeding efforts become more complicated and specific. In this case, the important signals inducing plastic responses tend to be specific, and plasticity and consequences of interactions in the presence of one species are different than plasticity and its consequences in the presence of another species. Identifying general and specific mixing abilities of crop cultivars is a first step towards breeding for beneficial intercropping systems63. A next step could be to identify if and how plasticity has a role in shaping these general and specific mixing abilities of cultivars. Eventually, for breeding purposes and farmers practice, it could be ideal if plasticity would enable a cultivar to have a high general mixing ability which means that it would be a beneficial companion for many different crop species.

For monocropping the importance of phenotypic plasticity is increasingly considered in relation to resilience to environmental stressors such as climate extremes64. The rationale here is that if plasticity is adaptive in the sense of increasing plant performance under a variety of environmental conditions, it could help make crops more stress resistant65. Nevertheless, the opposite, more plastic crop varieties being less stress tolerant, has also been found66. In intercrops this resilience could be modified by plastic responses to interspecific interactions. If two crop species differ in resistance to a certain stressor with one being severely affected and other less, than the latter species is partially released from competition and a more plastic plant type would potentially be better able to utilize this greater availability of resources. This means that the role of plasticity in intercropping can be beneficial under benign, but maybe even more under stressful conditions where in both cases the resource-foraging and competition-avoidance responses will improve and stabilize intercropping performance.