Plant diversity and overyielding: insights from belowground facilitation of intercropping in agriculture

Authors

  • Long Li,

    Corresponding author
    1. Key Laboratory of Plant and Soil Interactions, Ministry of Education, China, College of Resources and Environmental Sciences, China Agricultural University, Beijing, China
    • Author for correspondence:

      Long Li

      Tel: +86 10 6273 4684

      Email: lilong@cau.edu.cn

    Search for more papers by this author
  • David Tilman,

    1. Department of Ecology, Evolution and Behavior, University of Minnesota, St Paul, MN, USA
    Search for more papers by this author
  • Hans Lambers,

    1. School of Plant Biology and Institute of Agriculture, The University of Western Australia, Crawley (Perth), WA, Australia
    Search for more papers by this author
  • Fu-Suo Zhang

    1. Key Laboratory of Plant and Soil Interactions, Ministry of Education, China, College of Resources and Environmental Sciences, China Agricultural University, Beijing, China
    Search for more papers by this author

Summary

Despite increasing evidence that plant diversity in experimental systems may enhance ecosystem productivity, the mechanisms causing this overyielding remain debated. Here, we review studies of overyielding observed in agricultural intercropping systems, and show that a potentially important mechanism underlying such facilitation is the ability of some crop species to chemically mobilize otherwise-unavailable forms of one or more limiting soil nutrients such as phosphorus (P) and micronutrients (iron (Fe), zinc (Zn) and manganese (Mn)). Phosphorus-mobilizing crop species improve P nutrition for themselves and neighboring non-P-mobilizing species by releasing acid phosphatases, protons and/or carboxylates into the rhizosphere which increases the concentration of soluble inorganic P in soil. Similarly, on calcareous soils with a very low availability of Fe and Zn, Fe- and Zn-mobilizing species, such as graminaceous monocotyledonous and cluster-rooted species, benefit themselves, and also reduce Fe or Zn deficiency in neighboring species, by releasing chelating substances. Based on this review, we hypothesize that mobilization-based facilitative interactions may be an unsuspected, but potentially important mechanism enhancing productivity in both natural ecosystems and biodiversity experiments. We discuss cases in which nutrient mobilization might be occurring in natural ecosystems, and suggest that the nutrient mobilization hypothesis merits formal testing in natural ecosystems.

Introduction

There is an ongoing debate on mechanisms underlying a positive relationship between productivity and plant diversity which has been observed in a number of experimental ecosystems, such as grasslands (Tilman et al., 1996, 2001; Hooper & Vitousek, 1998; Hector et al., 1999; Schläpfer & Schmid, 1999), forests (Lovelock & Ewel, 2005), a natural phytoplankton community (Ptacnik et al., 2008), as well as in a meta-analysis of both aquatic and terrestrial ecosystems (Cardinale et al., 2006). The debate on the mechanisms accounting for such positive relationships between ecosystem productivity and species diversity has focused on three aspects: the ‘sampling effect’ (Huston, 1997), ‘complementarity’ and ‘facilitation’ (Hector et al., 1999). Under the sampling effect (or selection effect) hypothesis, when communities are assembled at random from a pool of species, more diverse mixtures have a higher probability of containing a species with higher productivity which may result in overyielding if it becomes dominant (Huston, 1997). The complementarity hypothesis states that complementary resource use and niche differentiation in space and time, or facilitation leading to increased resource availability, reduce interspecific competition and could lead to greater acquisition of limiting resources (Tilman et al., 2001). Generally, the complementarity effect includes both niche differentiation and facilitation, because distinguishing between them is difficult in practice (Loreau & Hector, 2001). There is some evidence for niche differentiation in space and time, for instance, via different rooting depths or in the time of the year for maximum nutrient acquisition (Hector et al., 2002; Fargione & Tilman, 2005). Direct plant–plant facilitation, as defined by Callaway (1995), is a beneficial effect of one individual on another due to, for example, the amelioration of harsh environmental conditions, including increased availability of a limiting resource. Alternatively, facilitation may act indirectly in other ways, such as elimination of potential competitors, introduction of beneficial organisms (i.e. soil bacteria, mycorrhizal fungi), introduction of pollinators or protection from herbivores.

There is limited direct experimental evidence for mechanisms of interspecific facilitation in ecological biodiversity experiments. A review by Brooker et al. (2008) presented plant–pollinator interactions, positive impacts of N2-fixing legumes on soil nitrogen (N) availability, the capacity for resource sharing through mycorrhizal networks, and classic ‘nurse plant’ effects as the major mechanisms of ecological facilitation in plant communities. However, our review of recent studies of overyielding in agricultural intercropping systems reveals additional mechanisms of interspecific facilitation that may well be highly relevant to natural plant communities. These mechanisms involve belowground processes whereby a plant of one species increases the availability of soil phosphorus (P), iron (Fe), zinc (Zn) or some other non-N nutrient that is limiting in a habitat, and thus, besides benefiting itself, also increases the growth of another species. Ecological studies of semi-natural and natural systems have not investigated such mechanisms (but see Muler et al., 2014).

Intercropping – that is, growing individuals of at least two crop species in close proximity at (about) the same time – is associated with interspecific interactions, including above- and belowground competition, and can lead to facilitation and overyielding in tropical and temperate habitats (Vandermeer, 1989). Research on rhizosphere processes and nutrient utilization in intercropping systems has provided a wealth of physiological evidence for interspecific facilitation between species (Horst & Waschkies, 1987; Ae et al., 1990; Zuo et al., 2000; El Dessougi et al., 2003; Li et al., 2003, 2007). The mechanisms underlying overyielding of intercropping systems, as documented in the last decade, are generally attributed to niche complementarity (Hector et al., 1999) and positive interspecific interactions (facilitation) in resource use (Hauggaard-Nielsen & Jensen, 2005; Li et al., 2007). This paper reviews and synthesizes recent work on interspecific facilitation of nutrient acquisition in intercropping systems, highlights possible mechanisms for increased productivity with increasing plant diversity, and suggests approaches for studying facilitative rhizosphere mechanisms in terrestrial ecosystems.

Mechanisms underlying interspecific facilitation through enhanced acquisition of soil resources and evidence for their involvement in overyielding in intercropping

Plant communities often comprise species with differing functional traits, including P-mobilizing and non-P-mobilizing species (Lambers et al., 2008), Fe- and Zn-mobilizing (i.e. many graminaceous monocotyledonous) and nonmobilizing (many dicotyledonous and nongraminaceous monocotyledonous) species (Kraemer et al., 2006), and nitrogen-fixing and non-nitrogen-fixing species (Fornara & Tilman, 2008). Here, we define ‘mobilizing’ plant species as those that are able to convert a given sparingly available soil nutrient into an available form that can be readily taken up. A nonmobilizing plant species is incapable of such conversion. These proposed mechanisms differ from previously suggested ones on interspecific facilitation (Callaway, 1995; Brooker et al., 2008).

Phosphorus

We propose that one mechanism of interspecific facilitation operates through enhanced P acquisition, that is, P-mobilizing plants increase the P-acquisition and growth of non-P-mobilizing species via increased availability of soil P in the rhizosphere.

Although plants often perform poorly on soils with low concentrations of available P, some of these soils contain a considerable amount of P that is unavailable to most species. Phosphorus sorption occurs with both organic and inorganic P. Organic P may account for 30–70% of the total P pool in agricultural soils, and up to 80–95% in nutrient-poor grasslands, peat soils, and forest soils (Dalal, 1978; Macklon et al., 1994; Turner, 2006). Plants do not take up organic P directly; rather, organic P is first hydrolyzed by microbial or root-released phosphatases. An association with P-mobilizing plant species can facilitate non-P-mobilizing species by mobilization of either organic P or insoluble inorganic P, through release of carboxylates, protons and enzymes by roots of P-mobilizing species (Fig. 1a).

Figure 1.

Schematic representation of mechanisms underlying nutrient-mobilization-based facilitation in intercropping systems. (a) Interspecific facilitation of phosphorus (P) acquisition by association of P-mobilizing and non-P-mobilizing plant species. Phosphorus-mobilizing species (left) can mobilize either sparingly soluble inorganic in soil by exudation of carboxylates or protons, or organic P by releasing acid phosphatases, which hydrolyze soil organic P into soluble inorganic P. The soluble inorganic P can be utilized by both P-mobilizing and non-P-mobilizing plant species. (b) Possible mechanism of interspecific facilitation of iron (Fe) and zinc (Zn) nutrition in a plant community with dicotyledonous or nongraminaceous monocotyledonous (Strategy I for Fe acquisition, non-Fe-/or Zn- mobilizing) species and graminaceous monocotyledonous (strategy II for Fe acquisition, Fe-/Zn- mobilizing) species.

Crop species vary widely in their capacity to mobilize or otherwise access sparingly soluble inorganic P (Pearse et al., 2006). Phosphorus acquisition of non-P-mobilizing species may benefit from the presence of P-mobilizing species when they grow close together. Interspecific facilitation of P acquisition by P-mobilizing species has been observed in several intercropping systems in glasshouse pot experiments: wheat (Triticum aestivum)/lupine (Lupinus albus, a P-mobilizing species) (Horst & Waschkies, 1987), sorghum (Sorghum bicolor)/pigeon pea (Cajanus cajan, mobilizing species) (Ae et al., 1990), wheat/chickpea (Cicer arietinum, mobilizing species) (Li et al., 2003), peanut (Arachis hypogaea, mobilizing species)/maize (Zea mays) (El Dessougi et al., 2003), and maize/chickpea (mobilizing species) intercropping (Li et al., 2004). It has also been shown for faba bean (Vicia faba, mobilizing species)/maize intercropping under field conditions (Li et al., 2007) (Table 1). The mechanisms underlying interspecific facilitation include greater P acquisition from organic P sources (Li et al., 2003, 2004) and enhanced P acquisition from inorganic P in oxide and hydroxide complexes or in other unavailable chemical forms (Horst & Waschkies, 1987; Ae et al., 1990; Li et al., 2007). This can occur when P-mobilizing plant species release protons (in alkaline soils only) and/or carboxylates (in all soils) into the rhizosphere to solubilize phosphates that are then taken up by both the P-mobilizing species and the non-P-mobilizing species in intercropping systems on soils with a low availability of P.

Table 1. Evidence for facilitation of phosphorus (P), manganese (Mn), iron (Fe) and zinc (Zn) acquisition, and direct evidence for rhizosphere effects between species
AuthorsType of evidenceDirect evidence for rhizosphere difference
Gardner & Boundy (1983), Horst & Waschkies (1987)White lupine (Lupinus albus) increased P uptake by intercropped wheat (Triticum aestivum)Yes
Ae et al. (1990)Pigeon pea (Cajanus cajan) enhanced P uptake by sorghum (Sorghum bicolor)Yes
El Dessougi et al. (2003)Peanut (Arachis hypogaea) increased P uptake in the presence of maize (Zea mays)Yes
Li et al. (2003)Chickpea (Cicer arietinum) facilitated P uptake by wheat (Triticum aestivum) under organic P supply, because chickpea acidified its rhizosphereYes
Li et al. (2004)Chickpea facilitated P uptake by maize under organic P supply. Chickpea exuded more acid phosphatase, and hydrolysed organic P into soluble inorganic P in soilYes
Li et al. (2007)Faba bean (Vicia faba) facilitated P uptake by maize (Zea mays L.), because faba bean acidified its rhizosphere, and exuded malate and citrate into its rhizosphere, which mobilized insoluble inorganic P in soilYes
Oelmann et al. (2007)Plant diversity enhanced nitrogen and P cycling in grassland plant speciesNo
Karanika et al. (2007)Aboveground phytomass and total P increased with increasing species richness. No selection effect, but complementarity in grassland plant speciesNo
Firn et al. (2007)Soil P availability was not determined simply by the number of species, but more likely depended on specific traits of the species present in a tropical tree plantationNo
Hooper & Vitousek (1998)Available P in soil increased with increasing plant functional group diversity in grassland plant speciesNo
Zuo et al. (2000), Inal et al. (2007), Xiong et al. (2013)Fe and Zn concentrations in seeds, shoots and roots of intercropped peanut were increased by its association with maizeYes
Kamal et al. (2000)Both extractable Fe and Zn in calcareous soil, and Fe and Zn concentration in guava (Psidium guajava) (Fe-inefficient species) were improved when the plants were intercropped with maizeYes
Muler et al. (2014)Banksia attenuata facilitated Mn uptake and growth of Scholtzia involucrata in a glasshouse experimentYes

Faba bean facilitates P acquisition of associated maize under both field and glasshouse conditions. Overyielding was 46% for maize and 26% for faba bean in a 4-yr field intercropping experiment; intercropping enhances P acquisition by faba bean and maize, compared with that by the respective species in monoculture (Li et al., 2007). Under normal intercropping practice, there is temporal and spatial niche differentiation in resource acquisition for these species, with faba bean being sown and harvested earlier than maize. However, glasshouse experiments, where temporal and spatial niche complementarity are diminished by sowing and harvesting both species at the same time, and with roots exploiting the same soil volume, have shown interspecific facilitation between faba bean and maize in terms of P acquisition and growth (Li et al., 2007). The physiological mechanism involved in this facilitation is that faba bean releases protons, malate and citrate into the rhizosphere, thus mobilizing insoluble soil P (Li et al., 2007). Malate and citrate are carboxylates that mobilize P bound to calcium in calcareous soil and P bound to oxides and hydroxides of aluminum (Al) and Fe in acid soils (Hinsinger, 2001).

Chickpea, a species that effectively accesses organic P, facilitates P acquisition by associated non-P-mobilizing wheat (Li et al., 2003) and maize (Li et al., 2004) when grown in intercropping systems. The physiological mechanism behind this facilitation is the exudation of acid phosphatases from chickpea roots (Li et al., 2004) which hydrolyze organic P into inorganic P (Lambers et al., 2008) and benefit both species when growing together. Many crop species (e.g. Lupinus albus, Trifolium subterraneum) can also access nucleic acids, phospholipids, glucose 1-phosphate and glycerophosphate in soil, due to the release of phosphatases from their roots (Lambers et al., 2008). This may be an important mechanism of facilitation and overyielding in P-limited ecosystems.

Iron/zinc

Iron is an essential micronutrient for plants, but predominantly exists in most soils in the Fe(III) form which is poorly soluble and hence insufficiently available to many plants, especially dicotyledonous and nongraminaceous monocotyledonous species. As a further constraint, high bicarbonate concentrations often result in lower Fe availability in soils. Dicotyledonous and nongraminaceous monocotyledonous species with strategy I for Fe absorption promote the release of Fe from minerals and organic complexes by proton efflux into the rhizosphere and enzymatic reduction via membrane-bound Fe reductases in rhizodermal plasma membranes (Kraemer et al., 2006). For other species without such adaptations, symptoms of chlorosis (indicating Fe deficiency) are commonly reported in soils with high pH and high CaCO3 concentrations (Robin et al., 2008). Graminaceous monocotyledonous species, however, which deploy strategy II, are often resistant to Fe deficiency. They release phytosiderophores, which are chelators with a very high affinity for Fe(III), allowing Fe(III) to diffuse towards the root surface, where the entire Fe–phytosiderophore complex is absorbed (Ma, 2005).

Zhang et al. (1991) demonstrated that phytosiderophore exudation in barley (Hordeum vulgare), a graminaceous monocotyledonous species, is not only induced by Fe deficiency, but also by Zn deficiency, resulting in enhanced availability of Zn. When dicotyledonous species and graminaceous monocotyledonous species grow together, the graminaceous species likely enhance the mobility of Fe(III) in soil, and thus enhance the availability of Fe for dicotyledonous species; similarly, the availability of Zn might be enhanced in Zn-impoverished soils (Fig. 1b).

Some non-Fe-mobilizing species, for example, guava (Psidium guajava) and peanut, frequently show iron chlorosis when growing in calcareous soils (Kamal et al., 2000; Zuo et al., 2000; Ma et al., 2003). The condition can be improved greatly when these plants are intercropped with Fe-mobilizing graminaceous monocotyledonous species (Kamal et al., 2000; Zuo et al., 2000; Inal et al., 2007). Glasshouse rhizo-box studies have demonstrated that the improvement in Fe nutrition of peanut or guava intercropped with maize is caused mainly by rhizosphere interactions (Kamal et al., 2000; Zuo et al., 2000). When peanut and maize grow together, phytosiderophores released by maize roots mobilize Fe(III), which may then diffuse from close to the roots of maize to the root surface of peanuts. At the root surface of peanuts, Fe(III) can be reduced, and thus the release of phytosiderophores from maize roots may improve the Fe nutrition of non-Fe-mobilizing plants. Intercropping increases Zn extractability in calcareous soils and the Zn concentration in guava leaves (Kamal et al., 2000). Similar results have been obtained for the peanut/maize intercropping system, increasing both Fe and Zn acquisition (Inal et al., 2007). Xiong et al. (2013) suggested that the Fe(III)- phytosiderophores deoxymugineic acid released by maize is absorbed directly by neighboring peanuts in the peanut/maize intercropping system. Table 1 summarizes evidence for facilitation of acquisition of P, Fe, and Zn.

Might these nutrient-mobilizing mechanisms apply to natural ecosystems?

Let us now consider the potential applicability to natural ecosystems of the nutrient-mobilizing overyielding mechanisms reviewed above. We do so to suggest some cases in which reported patterns in natural ecosystems seem consistent with the nutrient-mobilization hypothesis. We stress at the outset that our goal is to demonstrate plausibility. Further work is needed to test for the actual occurrence and importance of nutrient mobilization.

Nutrient limitation in natural ecosystems

Soil nutrient conditions in many natural ecosystems are less favorable than in most managed systems. Indeed, nutrient limitation is the norm in native soils, especially in alkaline soils, including many Aridisols and some Entisols, which are characterized by poor availability of P and transition metals (Fe, Cu, Mn and Zn) (Lynch & Clair, 2004). Approximately 25–30% of the world's land is calcareous in the surface horizon, potentially causing Fe deficiency in susceptible plants, and Fe deficiency could occur in many regions with Arenosols, Calcisols, Chernozems, Cypsisols, Kastanozems, Regosols, Solonchaks, Solonetz and Vertisols (Hiradate et al., 2007).

Soils from southwestern Australia are highly impoverished in nutrients, including P, Mn and Zn, because of nutrient-poor parent material and/or a very old landscape, where nutrients have been removed through leaching, erosion and occlusion (Laliberté et al., 2012). Otero et al. (2013) also considered that Fe, Mn, copper (Cu) and cobalt (Co) are limiting nutrients for plant productivity in Antarctic soils (Byers Peninsula, maritime Antarctica).

Because about a third of all soils have insufficient available P, with many tropical acidic soils being highly P impoverished (Batjes, 1997), P mobilization and facilitation may be of importance among the native plant species in these ecosystems. It is similarly plausible that the mechanisms of Mn, Fe and Zn mobilization observed in intercropping systems play potentially important roles in generating overyielding in natural ecosystems in which these nutrients are limiting.

The above discussion suggests that many nutrient-mobilizing species are legumes; however, many species with cluster roots or dauciform roots are also very efficient at nutrient mobilization (Lambers et al., 2008). Moreover, there is a wide variation among legumes, with some releasing large amounts of carboxylates and others very little (Pearse et al., 2006). Given the success of some legumes in habitats where P is poorly available (Lambers et al., 2013), it is highly likely that the their nutrient-mobilization mechanisms play important roles in overyielding in a range of natural and semi-natural plant communities.

Primary evidence with the nutrient-mobilization hypothesis in natural and semi-natural ecosystems

In grassland ecosystems, plant communities with greater species diversity showed increased acquisition of P under both field conditions (Oelmann et al., 2007) and controlled conditions (Karanika et al., 2007). Species combinations significantly affected the complementarity effect and the selection effect, and particular species consistently benefited from some species combinations (Karanika et al., 2007). For instance, Poa pratensis and Loliun perenne showed an intermediate performance in monoculture, but overyielding (producing more biomass) in a mixture with other species (Karanika et al., 2007). These results suggest the existence of differences in resource acquisition among species, and greater mobilization of a poorly available nutrient with certain species combinations. Experimental results for a California grassland showed that bicarbonate-extractable soil P increased significantly with increasing plant functional group diversity (Hooper & Vitousek, 1998). The authors ascribed this increase in P availability to an increase of soil organic P with increasing richness of higher plant functional groups (Hooper & Vitousek, 1998).

In serpentine wetlands in California, Delphinium uliginosum (a rare endemic) is locally more abundant, across a large portion of its range, when the moss Didymodon tophaceus is present (Freestone, 2006). Although mainly attributed to facilitated recruitment, the occurrence of D. uliginosum was also negatively related to soil pH (Freestone, 2006). These mosses have structures growing into the soil which likely release protons and carboxylates and mobilize insoluble soil nutrients, potentially representing a ‘rhizosphere’ effect of moss on D. uliginosum. Long-term fertilization of a wooded meadow on calcareous soils in Estonia showed that mosses had a higher P requirement than vascular plants, but responded less to P addition than vascular plants did (Niinemets & Kull, 2005). This suggests that the mosses have some capacity to mobilize poorly available nutrients in soils.

In soils from Byers Peninsula in maritime Antarctica, the concentrations of the most labile and mobile fractions (amorphous Fe oxyhydroxides, easily reducible Mn and metals soluble in sodium pyrophosphate) were higher in the sites colonized by vegetation dominated by mosses than in sites without plants (Otero et al., 2013). These authors also found that flowering plants (Deschampsia antarctica and Colobanthus quitensis) were growing together with mosses (Otero et al., 2013). Is there any facilitation of micronutrient acquisition between the flowering plants and mosses? Further research is needed to answer this and related questions.

In a heathland ecosystem in the Netherlands, the growth of three dominant species (Molinia caerulea, Calluna vulgaris and Erica tetralix) is limited by different nutrients. In the same habitat, Erica tetralix is limited by N, Calluna vulgaris by P, and Molinia caerulea by both N and P (Roem et al., 2002). This suggests differences in P requirement or P mobilization among co-occurring plant species, and that Erica tetralix may be a P-mobilizing species, possibly due to its association with ericoid mycorrhizal fungi. Arbuscular mycorrhizal fungi (AMF) can result in a more efficient exploitation of the soil nutrients, thus influencing plant communities. Plant–AMF mutualism promotes plant–plant facilitations in a plant community (Montesinos-Navarro et al., 2012).

Some species in P-impoverished habitats form dauciform roots (many sedges, Cyperaceae) or cluster roots (some Betulaceae, Casuarinaceae, Eleagnaceae, Fabaceae, Myricaceae and Proteaceae) (Lambers et al., 2006). Dauciform roots (Shane et al., 2006) and cluster roots (Lambers et al., 2008) exude carboxylates and mobilize sparingly soluble P in soil. These P-mobilizing plant species enhance P acquisition for themselves. They may also benefit their neighbors where P is limiting but, to date, this has not been tested.

Recent work has provided evidence for the facilitation of nutrient acquisition. Banksia attenuata, a cluster-rooted species native to Western Australia occurs on severely P-impoverished soils, co-existing with species that lack such specialized roots, for example, Scholtzia involucrata (Muler et al., 2014). In a glasshouse experiment, cluster roots of B. attenuata facilitated the acquisition of Mn by neighboring S. involucrata by making micronutrients more available (Muler et al., 2014).

In natural ecosystems, graminaceous monocotyledonous species usually co-exists with dicotyledonous species. Is there any facilitation to Fe acquisition of dicotyledons from graminaceous monocotyledons in a plant community? To our knowledge, there is very limited information on Fe acquisition in the two plant groups in natural ecosystems, but we might expect facilitation in these systems based on findings from the intercropping of peanut/maize (Zuo et al., 2000) and guava/maize (Inal et al., 2007).

Cropping systems and natural ecosystems are not analogous. Crop species may be grown together in an intercropping system, with positive interspecific interactions occurring over relatively short timescales, whereas in natural ecosystems interactions may last much longer. Furthermore, intercropping imposes a fixed plant arrangement, whereas plant-to-plant distances are variable in natural terrestrial plant communities. Interplant distance is important, as shown, for example, by diminished chlorosis particularly in those border rows of intercropped peanuts nearest to maize (Zuo et al., 2000). Therefore, further work is needed to determine how the intensity of interspecific facilitation varies between intercropping and natural ecosystems.

Conclusions

Our review and synthesis shows that facilitation of P, Fe, Mn and Zn acquisition is a potentially important cause of overyielding in annual intercropping systems. A more productive intercropping system, compared with monocropping, generally removes more nutrients, which will need to be replaced. However, there are no similarly detailed studies of facilitation of P and micronutrient mobilization and utilization in grasslands or other terrestrial ecosystems.

Demonstrating that the nutrient-mobilization hypothesis of overyielding is relevant to natural ecosystems will require the concordance of at least three lines of evidence. First, if, for instance, P is limiting, it will be necessary to identify P-mobilizing species and non-P-mobilizing species in a given natural plant community by glasshouse and field experiments (Hayes et al., 2014). Second, we would need to determine the differences in mobilization rates of otherwise poorly available inorganic and organic P in soil among plant species, in terms of rates of exudation of protons, carboxylates and phosphatases from roots during growth and development. Leaf Mn accumulation in plants in habitats with low P availability probably reflects a reliance on carboxylate release for P acquisition, and this may provide a simple method for screening potentially P-mobilizing species (Hayes et al., 2014) under field conditions. Third, we need to determine if P-nutrition facilitation and overyielding actually do occur when a P-mobilizing plant species grows together with a non-P-moblizing plant species under field conditions. Similar studies are needed when Fe or other micronutrients are limiting in natural ecosystems.

In summary, our review of intercropping and overyielding leads us to hypothesize that facilitation mediated by mobilization of soil P, Mn, Fe and Zn potentially plays an important role in the overyielding observed in experimental mixtures of herbaceous perennials in plant diversity experiments.

Acknowledgements

This work was supported by the National Basic Research Program of China (973 Program) (Project no. 2011CB100405), the National Natural Science Foundation of China (Project nos. 30870406, 31121062, 31210103906). L.L. was supported by the Endeavour Executive Award of DEEWR (Department of Education, Employment, and Workplace Relations), Australia. H.L. was supported by the Australian Research Council (ARC). We appreciate the constructive comments made by three anonymous reviewers on an earlier version of this manuscript.

Ancillary