Plant–soil feedbacks: the past, the present and future challenges

Primary and secondary succession

During early successional stages, facilitation processes, operating either directly between plants or via soil organisms, are pivotal to ecosystem development (Brooker et al. 2008). Early successional stages are often associated with positive plant–soil feedbacks (Reynolds et al. 2003), for example, due to symbiotic nitrogen (N) fixation and plant associations with mycorrhizal fungi. A classic example of positive plant–soil feedbacks in primary succession involves the symbiosis between plants and nitrogen-fixing bacteria, which increases soil N (Chapin et al. 1994). The success of this symbiosis during primary succession is due to the fact that most parent material is devoid of N. Nitrogen-fixing legumes, such as Lupinus lepidus, which have colonized Mount St. Helens after the most recent eruption, can form ‘islands of fertility’ that promote conditions for growth of later successional plant species (Titus & del Moral 1998; Corti et al. 2002).

The initial benefit of symbioses may become a disadvantage when early successional plant species develop negative soil feedback, which reduces their competitive ability against later successional plant species that are tolerant of the pathogens accumulated by earlier successional species. Such processes may also occur without early successional plant species with symbiotic nitrogen fixation. A classic example of such negative plant–soil feedbacks is that of the foredune grass Ammophila arenaria (marram grass). This grass is most vigorous in mobile dunes, because the colonization of wind-deposited beach sand enables it to temporarily escape from soil-borne enemies. When dunes become stabilized, A. arenaria degenerates because its roots are continuously exposed to soil pathogens (Van der Putten, Van Dijk & Troelstra 1988). This is an example of negative direct plant–soil feedbacks (see Box 1). However, this soil feedback effect has an indirect positive effect on later successional plant species, because they are tolerant to the pathogens of their predecessor (Van der Putten, Van Dijk & Peters 1993) (Fig. 1a). Negative direct plant–soil feedbacks may also occur with the later successional nitrogen-fixing dune shrub Hippophaë rhamnoides (Oremus & Otten 1981). In this case, decreased symbiosis and increased pathogenicity have both been identified, but most evidence points towards increased pathogenicity (Oremus & Otten 1981).

Plant community composition in early stages of secondary succession can also change rapidly due to negative plant–soil feedbacks. In a series of old fields that differed in time since abandonment, early successional (weedy) plant species demonstrated negative feedback effects, whereas later successional species had positive plant–soil feedbacks effects (Kardol, Bezemer & Van der Putten 2006). This suggests that during early secondary succession, effects of symbiotic mutualists (such as arbuscular mycorrhizal fungi) may have a smaller effect than soil-borne pathogens, but this is reversed in later successional stages (Janos 1980). Interestingly, similar effects have been observed in experiments where the soil biota consisted mainly of microbial pathogens and root-feeding nematodes (Van der Putten, Van Dijk & Peters 1993; Kardol, Bezemer & Van der Putten 2006), as well as in experiments where composition and numbers of larger soil organisms, such as microarthropods and insects were manipulated (Brown & Gange 1992; De Deyn et al. 2003; Schädler et al. 2004). While the proportional contribution of these different soil biota to the overall net plant–soil feedbacks effect is still unclear, they are considered to have substantial potential in ecological restoration (Eviner & Hawkes 2008; Harris 2009; Kardol & Wardle 2010).

Historical contingency, legacies and priority effects

Every plant species causes changes in soil conditions, which can remain as ‘soil carry-over effects’ (Bartelt-Ryser et al. 2005), often referred to ‘legacy effects’, or ‘historical contingencies’. These effects can persist in the soil for a period of time after a plant species has disappeared. The length of this legacy depends on the nature of the change in soil conditions (Van der Putten 2003). Legacy effects potentially have long-term consequences for plant community diversity and productivity (Grman & Suding 2010; Kulmatiski & Beard 2011). Microcosm studies, for example, have shown that microbial soil legacies left by early-successional grasses have more pronounced effects on mid-successional grasses than on mid-successional forbs (Kardol et al. 2007). Priority effects have also been demonstrated in microcosm studies where introducing certain plant species, or species groups, earlier than others can modify plant species composition; these effects can last for several years (Ejrnæs, Bruun & Graae 2006; Körner et al. 2008).

Priority effects have been suggested to be critical to the long-term dynamics of California grasslands, where initial dominance of exotic species with low dependence on mycorrhizal fungi leads to their sustained dominance. On the other hand, initial dominance of native species with high dependencies on mycorrhizal fungi will result in high densities of fungi and sustained dominance of native species (Vogelsang & Bever 2009). Some studies have related greenhouse-based legacy effects to priority changes under field conditions. For example, in abandoned arable fields, the weighted feedback effect of the early successional forb Jacobaea vulgaris on later successional plant species becomes increasingly positive with time since abandonment, suggesting that feedback of this forb influences which plant species may become dominant through priority effects and, consequently, the course of secondary succession (Van de Voorde, Van der Putten & Bezemer 2011). If weeds or invasive exotic species have created soil legacies, then mere removal of these species will not be sufficient for successful restoration of pre-invasion plant communities (Suding, Gross & Houseman 2004; Suding & Hobbs 2009; Yelenik & Levine 2010).

Plant abundance and rarity: Janzen-Connell processes

The Janzen-Connell pattern of reduced success of conspecifics in proximity to adults can be a dominant force, which structures tropical forests (Terborgh 2012). While originally thought to result from seed predators and above-ground pathogens, the pattern of reduced success of conspecifics near adult trees may also result from localized negative soil feedbacks (Fig. 1b). The first test of this possibility was carried out by Augspurger & Kelly (1984), who examined whether mature trees accumulate soil-borne pathogens that may prevent conspecific seedling establishment. Recent work has confirmed that reduced success of conspecific seedlings near adults (conspecific negative density dependence) in the tropics can be attributed to local scale plant–soil feedbacks (Mangan et al. 2010). Clear evidence of soil feedbacks driving the Janzen-Connell pattern in temperate forest was shown by Packer & Clay (2000). Failed establishment of black cherry (Prunus serotina) seedlings under conspecific adults was due to pathogenic soil microbes (Pythium sp.). Subsequent work has suggested that the decline in performance and abundance of seedlings near conspecific adult trees is common in both tropical and temperate forests (Comita et al. 2010; Johnson et al. 2012; Terborgh 2012). As the strength of negative density dependence of conspecific tree species in temperate forests was found to correlate with regional species richness and to increase towards the tropics, it has been suggested that plant–soil feedbacks may also contribute to latitudinal gradients in plant species richness (Johnson et al. 2012).

Within forested systems, the strength and direction of the soil feedback correlates with the relative abundance of tree species (Mangan et al. 2010). Dominant tree species had less negative plant–soil feedbacks than less-abundant species. A similar strong correlation was observed in a Canadian old field where frequently occurring plant species exhibited neutral to positive direct soil feedback, whereas less-abundant species had negative soil feedbacks (Klironomos 2002) (Fig. 1c). The consistency of this pattern in both forests and grasslands suggests that plant–soil feedbacks may provide an explanation for plant rarity. However, in a semi-arid grassland study, no such relationship between species abundance and soil feedback was found (Reinhart 2012).

Biological invasions

One of the main reasons why exotic species can become so invasive in their new range is that they have become released from specialist natural enemies that control abundance in the native range (Fig. 1d). In a study in a Canadian old field, introduced exotic plant species had neutral to positive plant–soil feedbacks comparable to dominant native plant species (Klironomos 2002). This finding was taken to suggest that exotic plant species could become invasive by not being exposed to negative plant–soil feedbacks, but it does not exclude the possibility that the exotic species were dominants in their native range where they may have had neutral plant–soil feedbacks as well. Nevertheless, subsequent studies have shown that invasive exotic plant species can indeed switch from negative plant–soil feedbacks in their native range to neutral, or even positive feedback in the range where they have become introduced (Reinhart et al. 2003; Callaway et al. 2004).

Positive plant–soil feedbacks effects may result from enhanced effectiveness of symbiotic mutualists (Fitter 2005) and decomposer organisms and/or reduced exposure to specialized soil pathogens. Most invasive exotic species are early successionals (Blumenthal 2005), indicating that they may benefit more from soil-borne enemy release than from exposure to generalist symbionts (Kardol, Bezemer & Van der Putten 2006). In North America for example, introduced species appear to have less dependence on mycorrhizal fungi than native species (Pringle et al. 2009). However, there are also a variety of other mechanisms involving plant-soil feedback explaining why exotic plant species can become so invasive. For example, some exotic species show negative plant–soil feedbacks in the new range while still being invasive. This phenomenon could be explained by accumulation of local pathogens to which the exotic species are less susceptible than natives (Eppinga et al. 2006). Support for this ‘accumulation of local pathogens’ mechanism has been provided by research on Chromolaena odorata in India (Mangla, Inderjit & Callaway 2008). Other possible mechanisms may relate to plants altering soil physico-chemical properties in the invaded range, which might provide direct positive plant–soil feedbacks to the exotics and negative feedback to natives.

It has been proposed that exotic plant species can lose invasiveness (Simberloff & Gibbons 2004; Hawkes 2007), and some recent studies have shown that this may result from altered plant–soil feedbacks with time since introduction (Fig. 1e). For example, in a study from New Zealand, the soil feedback of 12 introduced exotic species was negatively correlated with their residence time (Diez et al. 2010). The cause of increased negative plant–soil feedbacks has not yet been elucidated. In another example, the soil-mediated effects of the glucosinolates produced by the introduced forb Alliaria petiolata in the USA were found to decrease with time since introduction (Lankau et al. 2009). Most exotics, however, do not become invasive at all. Little is known about plant–soil feedbacks patterns of these non-invasive exotics.

Climate change: direct impacts and range shifts

Both temperature increases and altered precipitation patterns are likely to influence plant–soil feedbacks, with potentially far-reaching consequences for ecosystem functioning. Climate change is likely to impact the soil organisms directly because warming of the soil can stimulate microbial activity, the breakdown of organic matter and hence the loss of carbon from soil (Dorrepaal et al. 2009). Extreme weather events, such as drought and heavy rainfall, can have substantial effects on microbial physiology and the composition of soil microbial communities, with consequences for ecosystem-level carbon dynamics (Schimel, Balser & Wallenstein 2007). However, there is now increasing evidence that some of the most significant effects of climate change on ecosystem carbon and nutrient dynamics are mediated via plants and their interactions with soil organisms (De Vries et al. 2012). Moreover, multi-trophic interactions between plant and soil communities might become decoupled by climate change as plants, herbivores and predators may respond differently to climate changes, and this will almost certainly affect ecosystem functioning (Wardle et al. 2011).

The numerous routes by which climate change can impact plant–soil interactions and ecosystem functioning can operate at a variety of spatial and temporal scales. At the individual plant level, and over intra- and inter-annual timescales, changes in temperature, water availability and rising atmospheric carbon dioxide (CO₂) concentration, all impact plant photosynthesis and the transfer of photosynthetic carbon to soil, with cascading effects on soil organisms and ecosystem functioning. An example of this is a study on pine forest showing that elevated atmospheric CO₂ concentrations increased the flux of C to roots, which in turn stimulated microbial activity and the mineralization of organic matter, and hence C loss from soil (Drake et al. 2011). However, this process also stimulated the release and turnover of N and its uptake by the trees, which triggered a positive feedback that sustained enhanced rates of tree production and hence C gain under elevated CO₂. Changes in temperature and precipitation regimes and elevated CO₂ can also change the productivity and composition of plant communities over decadal timescales, which in turn impacts soil organisms and C cycling by altering the amount or quality of organic matter entering the soil.

Shifts in vegetation composition due to altered rainfall patterns, as found in tropical rainforest (Engelbrecht et al. 2007) and African savanna (Good & Caylor 2011), not only influence photosynthesis (Ward et al. 2009) and hence the exudation of photosynthate in the soil, but also the amount and quality of organic matter entering the soil as plant litter (roots and shoots). Moreover, such shifts can modify the soil physical, chemical and biological environment by changing root architecture and rooting depth (Jackson et al. 1996). These effects may influence composition of soil microbial communities and functioning. Finally, over hundreds of years, climate-mediated expansion of plant species from lower latitude and altitude, and its impact on organic matter supply to soil and soil physical conditions could potentially have consequences for soil food webs and the biogeochemical cycles at local and, potentially, global scales (Chapin et al. 2009; Ostle et al. 2009).

The current climate-mediated range expansion of many species (Walther et al. 2002) is expected to disrupt current communities when species differ in dispersal capacity: plants disperse faster than many soil organisms (Berg et al. 2010). Plant species with rapidly shifting ranges can potentially outrun their natural soil-borne enemies, symbionts and decomposer organisms (Fig. 1d). For example, range shifting of Tragopogon dubius in Europe has resulted in release from growth-reducing soil biota (Van Grunsven et al. 2010). Thus, range shifting can expose plants to soils with a more positive balance between enemy and mutualist effects, which has also been shown for invasive exotic plant species from other continents. Two phylogenetically controlled experiments showed that range-shifting plant species can indeed have neutral soil feedback comparable in magnitude to cross-continental exotics and different from the negative direct soil feedbacks of natives (Engelkes et al. 2008). On the other hand, the range-shifting species had little effect on ecosystem processes such as litter decomposition (Meisner et al. 2012), so that their plant–soil feedbacks benefits were more due to altered pathogen/symbiont balances than to enhanced nutrient availability. These studies have all focused on the effects of range expansion, whereas little is known about legacy effects in soils and the resulting priority effects in plant–soil feedback interactions of the remaining plant species following range contractions.

Biodiversity and ecosystem functioning

Until recently, there has been little work on plant–soil feedbacks in relation to plant species diversity influences on productivity. Recent work suggests that effects of plant species diversity on productivity may be due to combined community level responses to soil-borne pathogens and nutrition aspects (de Kroon et al. 2012). In 2011, two studies by Maron et al. (2011) and Schnitzer et al. (2011) showed that the proportionally low productivity in plant monocultures, compared to that in plant mixtures, could be increased by (partial) soil sterilization. Their studies suggested that pathogenic soil fungi may have been causing the negative effects on productivity in monocultures, thus explaining the typical positive relationship between plant species diversity and productivity.

These results provide a new perspective on plant diversity-functioning studies as they suggest that the phenomenon of overyielding in mixed plant communities may not be caused by resource complementarity alone, but also by dilution of negative plant–soil feedbacks effects in mixed plant communities (Fig 1f). Kulmatiski, Beard & Heavilin (2012) placed these findings in the context of a biomass-explicit, multi-species plant–soil feedbacks model to demonstrate that across a full range of plant–soil feedbacks effects, negative feedbacks should result in overyielding while positive feedbacks should result in underyielding. This and related work (Kulmatiski, Heavilin & Beard 2011; Hendriks et al. 2013) has highlighted the need for combined experimental studies and model comparisons between plant-soil community feedback effects (Box 1) and individual feedback effects.

Role in above-ground–below-ground multitrophic interactions

Soil organisms in the rhizosphere, such as root-feeding insects, root pathogens or mycorrhizal fungi, as well as soil organisms that do not directly interact with the plant such as decomposers or predators, can affect the performance of above-ground herbivorous insects and their natural enemies (Bardgett & Wardle 2010). Interactions between plant roots and soil organisms can result in altered growth and nutritional quality of above-ground plant parts. Many studies have shown, for example, that root feeding by insects or nematodes, or infections with soil pathogenic or mycorrhizal fungi can lead to changes in the concentration of primary or secondary plant compounds in leaf tissues. Such changes in the nutritional quality of above-ground plant parts can subsequently affect the performance of foliar herbivores and the natural enemies of these herbivores, such as parasitoids or predators (Pineda et al. 2010; Soler et al. 2012).

Both above-ground and below-ground herbivory can lead to changes in the composition of the soil biota (Bardgett & Wardle 2010). Above-ground grazing, for example, typically leads to short-term increases in C excretion from roots and this affects the functioning and composition of the soil microbial community (Hamilton & Frank 2001). Moreover, herbivory can lead to changes in the concentration of primary and secondary plant compounds in root tissues and this can affect direct interactions between roots and soil organisms, as well as the decomposition of root tissues, and hence affect the composition of soil communities (Van Dam et al. 2010). Via these influences on the soil community, herbivory can affect plant–soil feedbacks responses (Fig. 1g). Even though very few studies have examined such effects so far, the available evidence suggests that multi-trophic interactions including foliar herbivores or fungal endophytes can indeed affect plant–soil feedbacks responses (Matthews & Clay 2001; Mikola et al. 2005; Sørensen, Mikola & Kytöviita 2008). It is also possible that plant–soil feedbacks can influence the nutritional quality of plants, thereby modifying above-ground or below-ground multitrophic interactions.

While most research has focused on the joint effects of a shared host plant, above-ground–below-ground interactions can also occur when above-ground and below-ground organisms do not share the same plant. Kostenko et al. (2012) showed that herbivory, by affecting the soil community, can alter the composition of secondary compounds in plants that grow later in the soil, as well as the performance of above-ground herbivores and parasitoids on those plants. Clearly, multitrophic interactions between plants and herbivores above-ground and below-ground can influence plant–soil feedback interactions. Moreover, plant–soil feedbacks can affect the interactions between plants and consumers. Therefore, future studies should place plant–soil feedbacks in a multitrophic context and examine in more detail how top-down effects become bottom-up and vice versa (Moore et al. 2003).

Evolutionary perspectives, foundation species and community genetics

Plant–soil feedbacks in natural systems are an emerging, but little studied, area of research linking ecological and evolutionary processes. Whether directly or indirectly, plant functional traits, such as the C to N content of leaves, can create conditions that affect soil biota (i.e. their composition or activity) and decomposition processes that soil biota at least partly regulate (i.e. nutrient depolymerization or mineralization). Such niche construction, in the context of strong environmental gradients in soil, can create selective gradients that may affect the evolution of both plants and soil biota. The data so far demonstrate that genetically controlled variation in functional plant traits can influence soil communities and the processes they mediate in soil (i.e. create an ‘extended phenotype’; Whitham et al. 2003, 2006; Schweitzer et al. 2008). The literature based upon above-ground systems has commonly found that herbivores and tri-trophic interactions serve as important selection pressures for plants (Fritz & Simms 1992; Hunter & Price 1992). However, the soil biota is also emerging as an important factor in mediating the evolutionary dynamics of plant functional traits (Lau & Lennon 2011). Recent studies show the importance of within-species genetic variance on plant–soil feedbacks that is mediated by aspects of the soil biota (Pregitzer et al. 2010; Felker-Quinn, Bailey & Schweitzer 2011; Lankau et al. 2011). These data are some of the first in natural systems to show that genetically based functional variation in plant traits can influence soil communities and that soil biota shaped by functional plant traits can then act as agents of selection for above-ground traits through feedbacks. As such, they implicate the soil biota as an important regulator of both below-ground processes as well as genetic variation in these systems. Plant–soil feedbacks has the potential to impact selection and evolutionary processes, from plant trait divergence to speciation, but examination of this possibility has only just begun.