1.Ideas about mechanisms controlling plant species richness are founded on empirical studies of above-ground vegetation. In many ecosystems, however, the majority of vegetation (e.g. 50–90% in temperate grasslands) occurs below-ground as roots, rhizomes and shoot bases. Whether the richness patterns described for above-ground vegetation also hold for the large below-ground component is still unknown.
2.Here, we provide a concise overview how the use of DNA-based techniques might alter our perception of richness patterns in plant communities. We focus mainly on temperate grasslands, but new patterns should also arise in other community types (except epiphyte-rich ecosystems).
3.We hypothesise that DNA-based measurements of below-ground plant richness will reveal that richness is greater below- than above-ground because many perennial plants persist below-ground even in the temporary absence of above-ground shoots, and because the roots and rhizomes of plant individuals occupy larger areas than do shoots. Consequently, the species–area relationship may show steeper slopes for below-ground than above-ground richness. Further, above-ground richness may not be a constant proportion of below-ground richness, so the ratio of below/above-ground richness may vary along environmental gradients of productivity, disturbance and heterogeneity.
4.We also hypothesize that the often-observed decrease in above-ground richness with increasing productivity may not occur for below-ground richness, partly for the reasons noted above, and partly because of differences between above- and below-ground resources. Light is supplied largely in one dimension and does not persist in the environment for later uptake, whereas nutrients and water are supplied in three dimensions and can persist in the soil. Together, these differences should allow more niche differentiation and greater species richness below-ground.
5.Current DNA-based methods that allow measurement of below-ground richness in the field are likely to reveal patterns different from those well-documented for above-ground richness and may also produce new insights about plant community structure and function.
Biodiversity is a primary focus of both theoretical and experimental ecology (Huston 1994; Hubbell 2001; Pereira et al. 2010; Wilson et al. 2012). Until very recently, plant species diversity and coexistence theories were entirely based on above-ground data. In contrast, many widespread ecosystems of temperate, arid and polar regions such as grassland, steppe, desert and tundra have >50% of plant production or biomass below-ground (Jackson, Mooney & Schulze 1997; Steinaker & Wilson 2005; Mokany, Raison & Prokushkin 2006; Poorter et al. 2012). The measurement of species richness of this large below-ground component has been hindered by the difficulty of assigning roots and rhizomes to species. DNA-based methods now allow below-ground richness to be studied in the field (Frank et al. 2010; Jones et al. 2011; Kesanakurti et al. 2011; Hiiesalu et al. 2012), and the further development of these methods will allow much broader, cheaper and more accurate exploration of below-ground richness (Mommer et al. 2011). As the number and scale of studies increases, we expect major differences between below- and above-ground plant richness to emerge in most vegetation types, but especially in those dominated by below-ground allocation.
We hypothesize that among the most important differences likely to emerge is that below-ground richness generally exceeds that above-ground, because of greater dispersion of plant parts below- than above-ground in both time and space. Most perennial plants have persistent below-ground storage organs and meristems that allow short- or long-term dormancy for up to decades in the absence of above-ground biomass (Klimesova & Klimes 2007; Reintal et al. 2010). Below-ground richness should also exceed above-ground richness because roots and rhizomes tend to be laterally more wide-spread than above-ground plant parts (Coupland & Johnson 1965; Kummerow, Krause & Jow 1977; Kutschera & Lichtenegger 1992; Gibbens & Lenz 2001; Schenk & Jackson 2002; Casper, Schenk & Jackson 2003).
In this paper, we discuss new theoretical insights about below-ground species richness likely to be gleaned from the application of DNA-based techniques to previously described patterns of above-ground species richness. Most of our examples are from grasslands because their above-ground richness has been well-studied (Grime 1979; Tilman 1982; van der Maarel 1993; Tilman 1999; Wilson et al. 2012), but below-ground species richness may be relatively important in other vegetation types as well. We examine richness at the plant neighbourhood-scale because it is frequently reported, and it is the distance within which plants interact (Cahill 2003; Milbau et al. 2005; Weigelt et al. 2007). For example, herbaceous species richness in grasslands is often studied at scales 0·1–1 m2 (Mittelbach et al. 2001). In addition, we also provide a short overview of current DNA-based methods for measuring below-ground richness (see Appendix S1, Supplementary Information), including next generation sequencing – a novel and powerful method for identifying roots and rhizomes from multispecies samples collected from the field. We also discuss some methodological issues that need to be considered: the choice of DNA marker regions, the treatment of closely related species indistinguishable by DNA analyses, and comparable sampling protocols for the below- and above-ground realms.
Relative magnitudes of below- and above-ground species richness
In all vegetation types besides these dominated by epiphytes, below-ground richness cannot be less than above-ground richness within identical sample sizes because all shoots occurring above-ground are rooted in the soil (Fig. 1). Below-ground richness can be considered as the sum of traditionally measured above-ground richness and additional species revealed solely by using DNA-based methods (Hiiesalu et al. 2012). Thus, below-ground richness describes the number of species that coexist within the plant neighbourhood scale. Plants rooted outside a sample area but that lean into the above-ground sample are mostly unimportant in this regard because they influence above-ground species richness only at the scale of few centimetres (Kilburn 1966). Other types of vegetation, such as rainforests with epiphytes, will have different relationships between below- and above-ground richness. Next, we explore why below-ground richness should exceed above-ground richness. We discuss the effect of the greater dispersion of below-ground plant parts in both time and space, and the greater diversity of resources below-ground than above-ground.
Greater Dispersion of Below-Ground Parts in Time
Below-ground parts are more dispersed in time than above-ground organs because roots and below-ground meristems can survive during unfavourable seasons and years in the absence of above-ground biomass (Eissenstat & Yanai 1997; Wells & Eissenstat 2001). Seasonally missing shoots could be counted through repeated sampling over the course of the growing season. Shoots may also be missing in some years but present in others (Shefferson 2009; Reintal et al. 2010), resulting in >30% variation in above-ground richness among years in natural vegetation (Pärtel & Zobel 1995; Wilson & Tilman 2002; Wilson 2007). Consequently, the cumulative number of species observed above-ground over several consecutive years increases even if the average above-ground plant species richness among years is constant (van der Maarel & Sykes 1993). We hypothesize that repeated inventories over many years can produce measures of above-ground richness closer to below-ground richness, but repeated above-ground inventories cannot separate real changes in species composition from the temporary absence of species that are dormant below-ground. Another consequence of below-ground dormancy is that seasonal and stochastic fluctuations in below-ground richness should be much smaller than that above-ground.
Greater Dispersion of Below-Ground Parts in Space
Below-ground organs are also more widely dispersed in space than above-ground organs (Coupland & Johnson 1965; Kutschera & Lichtenegger 1992), partly due to differences in how below- and above-ground biomass function in their respective environments, soil and air. Because roots and rhizomes are physically supported by soil, they require relatively little structural tissue for strength (Barthelemy & Caraglio 2007). Apart from a few stout roots that are needed to anchor plants, the roots that take up resources and interact with other plants are very small, <2 mm in diameter (Jackson, Mooney & Schulze 1997). Therefore, a gram of root or rhizome tissue is able to occupy a much larger environmental volume than the same amount of shoot tissue (Kummerow, Krause & Jow 1977; Casper, Schenk & Jackson 2003), resulting in relatively greater exploration of the below-ground environment.
Greater Diversity of Below-Ground Resources
Below-ground resources may be more diverse than above-ground resources, which should also promote greater below- than above-ground richness, for several reasons. (i) There are more kinds of below-ground resources (water, macronutrients and micronutrients, contrasting ionic and organic forms) compared with above-ground resources (mainly light, as CO2 is typically not limiting). (ii) Soil resources diffuse and can be acquired from all directions, causing roots to grow in three dimensions, whereas light only comes from above and provides limited opportunities for spatial niche diversification (Reich, Wright & Lusk 2007). (iii) Below-ground nutrients have a broader range of temporal availabilities because they can either be taken up immediately or stored in the soil with turnover times ranging from seconds for ions to decades for organic matter (Parton, Stewart & Cole 1988). Light, in contrast, is not stored in the environment but must be used immediately. (iv) In contrast to the atmosphere, soil is also highly diverse in symbionts (e.g. mycorrhizal fungi), which can be crucial for acquiring resources. Indeed, complex soil microbial interactions might determine the whole nature of plant communities (Bever et al. 2010). In sum, if environmental diversity begets plant species richness (Hutchings, John & Stewart 2000), one might hypothesize that the relatively heterogeneous nature of soil in space, time, storage ability, texture and symbioses directly promotes plant richness below-ground, whilst its influence on above-ground plant richness can only be indirect via root–shoot linkages.
Evidence that below-ground richness exceeds above-ground richness comes from a diverse European grassland, in which species richness at plant neighbourhood scales (1000–8000 cm3) was up to two times greater below-ground than above-ground (Hiiesalu et al. 2012). At spatial scales up to 50 m2, the greatest above-ground plant species richness in the world occurs in temperate grasslands, for example, 25 species have been recorded from 10 × 10 cm area in an Estonian wooded meadow (Kull & Zobel 1991; Wilson et al. 2012). We hypothesise that at such scales, below-ground richness is even greater. Greater below-ground richness also occurred in a Canadian old-field, in which a third of the 29 taxa identified from 1285 root fragments were not found above-ground (Kesanakurti et al. 2011). Taxa not found above-ground were generally rare at the site and were hypothesised to be dormant or had very extensive root systems.
Overall, one of the most important and general results from measuring below-ground richness may be that neighbourhood-scale soil samples contain more species than can be detected by examining shoots (Fig. 1). It is possible that above-ground plant richness constitutes some constant proportion of below-ground plant richness (Fig. 1, dotted line), but it seems more probable that nonlinear relationships exist, causing the ratio of below/above-ground richness to vary in space and time. For example, the relationship might be saturating if increases in below-ground richness do not also occur above-ground (Fig. 1, dashed line). Indeed, Hiiesalu et al. (2012) found that the increase in below-ground richness within a 1000 cm3 sample in grassland was initially associated with an increase in above-ground richness, but average above-ground richness reached an asymptote at about seven species whereas below-ground richness exceeded 10 species.
The species–area relationship below- and above-ground
One of the most commonly observed relationships in ecology is that species richness increases with increasing sample size. Williamson (2003, p. 904) reviewed information on plant species–area relationships and stated that ‘I know of no study attempting to record underground parts, which is not surprising, although they are important parts of plants.’ DNA-based methods will provide new information about species–area relationship.
In very small sample areas (near zero), both below- and above-ground richness are identical (zero or a single species). At the largest spatial scales, below-ground richness may include temporarily dormant species that are not detected above-ground. Consequently, we predict that the rate of increase in richness with sample area should be greater below-ground compared with above-ground (Fig. 2a). Hiiesalu et al. (2012) explored grassland plant richness in scales from 1000 to 8000 cm3 (10 cm above and below soil surface) and always found greater below-ground richness and that the difference increased with increasing scale. Interestingly, they also found that below-ground richness exceeded that of above-ground at the community scale when cumulative species richness was compared over several samples scattered over a 2 ha area. This can be explained by species dormancy at the time of above-ground sampling because species found only below-ground were known to be present on the study site during previous years.
Lack of correspondence in species–area curves for below- and above-ground also suggests that to include the same number of species, different sample sizes are needed below- and above-ground. This difference in scale would be useful to understand how plants are distributed and function below- and above-ground. For example, data from Hiiesalu et al. (2012) suggest that, in a temperate grassland, a sampling unit of 1000 cm3 below-ground contains as many species (on average 8–9) as an above-ground sample three times larger.
We hypothesize that neighbourhood-scale below-ground richness increases linearly with the size of the species pool (all species in the surrounding potentially able to arrive and live in an area, Eriksson 1993; Pärtel, Szava-Kovats & Zobel 2011). In contrast, above-ground richness may show a curvilinear saturation pattern stemming from our inability to detect species that are present only below-ground (Fig. 2b). This possibility is supported by an analysis in which the cumulative above-ground richness over several years was well correlated with a theoretical species–area relationship, but average above-ground richness was below the cumulative regression line at small scales, probably because the above-ground component was an incomplete sample of plant species actually present (Fridley et al. 2006).
Plant richness below- and above-ground along ecological gradients
The unimodal relationship between above-ground plant richness and habitat productivity is a common pattern in temperate ecosystems, especially in herbaceous vegetation (Mittelbach et al. 2001; Pärtel, Laanisto & Zobel 2007). How might below-ground plant richness vary with spatial variation in productivity? At very low productivity, the species pool is typically relatively small (Zobel & Pärtel 2008), so both below- and above-ground richness should be relatively low. At increasing productivity, where competition starts reducing above-ground richness (Foster et al. 2004), we hypothesize below-ground richness to be less affected for the same reasons that generally cause below-ground richness to exceed that above-ground: below-ground dormancy, the ability of roots and mycorrhizal symbionts to exploit large overlapping soil volumes in three dimensions, and the environmental storage of resources (Fig. 2c). In relatively productive habitats, asymmetric light competition (Weiner 1990) is the main cause of competitive exclusion (Zobel 2001; Hautier, Niklaus & Hector 2009). In contrast, the diffusion of nutrients in soil can be slow and soil nutrients might be depleted only very close to the root surface (Craine 2005), so that soil resource competition is less likely to cause competitive exclusion. A recent study indeed found support for biotic assembly rules above-ground, but below-ground plant communities were structured more by abiotic processes (Price et al. 2012). In total, competitive exclusion at high productivity should be less common below-ground than above-ground.
Long-term maintenance of relatively high below-ground richness in productive habitats requires that all species are able to photosynthesize, with the rare exception of non-photosynthetic myco-heterotrophic plants that obtain carbon from other plants (Shefferson 2009). Productivity often varies among years in natural ecosystems, as does disturbance. Thus, high below-ground richness can provide a chance for species to persist in the ecosystem when they are excluded temporarily above-ground. Evidence that plants can persist by exploiting occasional opportunities for photosynthesis comes from the Carousel Model, which describes how average above-ground neighbourhood-scale richness can be constant among years at the same time that species composition varies, such that a study area can be inhabited by more species than are encountered in a one-time sample (van der Maarel & Sykes 1993). There is also evidence that below-ground biomass remains alive for some time in the absence of above-ground photosynthetic biomass in grasslands (Richards 1984; Thorne & Frank 2009) and Populus forests (Kosola et al. 2001), with below-ground meristems remaining dormant and hidden for years or even decades (Shefferson 2009; Reintal et al. 2010). Recent studies have shown that dormancy is a relatively efficient form of phenotypic plasticity (Jäkäläniemi et al. 2011), leading to increased lifetime reproductive success (Gremer, Crone & Lesica 2012). Consequently, at high productivity, the rapid decline in above-ground plant richness attributed to light competition might not occur below-ground for some time (Fig. 2c). Such a delay in the reduction in species richness provides a buffer period during which restoration of degraded sites could be successful.
Support for different patterns in species richness below- and above-ground along productivity gradients comes from a northern European grassland, which showed a negative relationship between above-ground richness and soil fertility, in accordance with several previous observations. Below-ground richness, however, was not related to soil fertility, because the number of species recorded only below-ground actually increased with increasing soil fertility (Hiiesalu et al. 2012). Further studies are needed to explore the generality of this pattern.
It has been noted that the unimodal productivity–diversity relationship might be a sampling artefact (Oksanen 1996) arising from the fact that individuals in productive habitats are relatively large and each of them occupy a relatively large volume of above-ground space. Therefore, sampling a fixed area will encounter fewer individuals and apparently fewer species. Below-ground vegetation, dominated by intermingled fine roots even in relatively productive habitats (Pärtel & Wilson 2002), should not suffer from this artefact.
Next, we consider how the above-described differences in the behaviour of below- and above-ground richness may require rethinking of classical models that relate richness to resource competition predicated on observations of above-ground richness (Grime 1979; Tilman 1982). For example, if below-ground richness does not decline at high productivity for some time (Fig. 2c), the role of competition in suppressing richness in productive habitats will be further called into question (Brooker et al. 2005; Wilson 2007).
Disturbance, defined as any mechanism that removes biomass (Grime 1979), is another major abiotic driver of diversity (Connell & Slatyer 1977; Grime 1979; Huston 1979). In undisturbed environments, as in productive habitats, we expect that below-ground richness will exceed that above-ground.
At intermediate disturbance, richness should also be greater below- than above-ground for the reasons outlined earlier. Further, below-ground biomass is protected from above-ground disturbance by the soil, so that the bud bank in the soil is an important source of revegetation and species richness after fire, drought and heavy grazing (Klimesova & Klimes 2007). In addition, owing to the 40% longer growing season of roots compared with shoots (Steinaker & Wilson 2008), and the relatively high turnover rates of roots (Pärtel & Wilson 2002), plants may be able to replace lost roots in the same season that the disturbance occurred, whilst the recovery of foliage might not occur until the following growing season. This should also promote greater species richness below-ground than above-ground.
At very high disturbance intensities, only a subset of disturbance-tolerant species should persist, resulting in a decline in richness both below- and above-ground. In summary, we hypothesize that below-ground richness exceeds that above-ground in less disturbed habitats, but both below- and above-ground richness should be low in the most disturbed habitats (Fig. 2d).
Environmental heterogeneity has been widely posited to control species richness (Tilman 1982; Hutchings, John & Stewart 2000). Recent overviews, however, have found little empirical evidence for positive heterogeneity–richness relationships in plant communities (Lundholm 2009), especially at neighbourhood scales where competition and heterogeneity should be most relevant (Tamme et al. 2010). This gap between expectation and data might be due to the measurement of only the above-ground fraction of plant richness. Below-ground richness might be more strongly related to heterogeneity than is above-ground richness, because the soil environment supports higher amounts of heterogeneity than the atmosphere. Both soil and air have one-dimensional gradients of distance from the surface and its concomitant effects on resource availability, but, as noted above, soil has additional heterogeneity imposed by three-dimensional variation in particle size, organic matter content, root and hyphal placement, root depletion zones and animal activity.
Alternatively, there might be negative relationships between heterogeneity and richness if more heterogeneous soils are dominated by few relatively large plants that forage in both nutrient-rich and nutrient-poor patches, averaging over patchy environments. Plants that dominate heterogeneous soils often have a well-developed ability to locate and use nutrient-rich patches through clonal below-ground growth (Eilts et al. 2011). These species can also occupy nutrient-poor soil patches by transporting resources from nutrient-rich patches. Such species might outcompete smaller plants that cannot access rich patches (the heterogeneity-size asymmetry hypothesis, Grime 1994; Rajaniemi 2003). Consequently, heterogeneous soil conditions might result in the exclusion of smaller species. Consequently, new insights about the relationship between heterogeneity and richness may emerge as more information becomes available about below-ground richness.
Beyond below-ground richness
Plant ecologists will learn much from new methods that measure below-ground richness, an aspect of community composition that has been previously hidden. Currently, our ideas and discussions are based largely on only species that appear above soil surface, the ‘tip of the iceberg' for many vegetation types (Jackson, Mooney & Schulze 1997; Mokany, Raison & Prokushkin 2006). There are only a few pioneering studies on below-ground richness in temperate grasslands (Frank et al. 2010; Kesanakurti et al. 2011; Hiiesalu et al. 2012), and other ecosystems are still totally unstudied in this regard.
DNA-based identification of below-ground plant parts will allow exploration of other topics beyond richness, such as the distribution of species along environmental gradients, the identities of coexisting roots and rhizomes, and below-ground assembly rules of plant communities (Price et al. 2012). A promising emerging field is ‘plant behavioural ecology’, which studies whether and why plants direct roots either towards or away from other species or even other genotypes of the same species (de Kroon 2007; Semchenko, John & Hutchings 2007; Cahill et al. 2010; Cahill & McNickle 2011). Molecular methods might help to determine how far actually plants can send their roots and rhizomes, for how long they can be dormant, and which species co-exist. Similarly, evolutionary ecology can benefit from new insights and methods. For example, the question of why there are so many small plants (Aarssen, Schamp & Pither 2006) could be studied by examining plant size and competitive ability below-ground.
The same DNA-based methods can also be used to study root-inhabiting biota (bacteria, fungi, invertebrates) using the same samples for plant identification but with the application of appropriate taxon-specific primers. This would allow us to go beyond plant community ecology to other trophic levels (Wardle et al. 2004). It would also provide insights into plant–microbe competition for nutrients, which is interesting because in some cases plant–microbe competition may be more important than plant-plant competition (Reynolds et al. 2003).
A complete picture of richness relationships is also vital for vegetation management and diversity conservation. Eutrophication, anthropogenic disturbance and habitat change are major threats to biodiversity. Below-ground richness might be useful for estimating restoration potential. Plant community restoration will be more efficient in areas where below-ground richness is still high.
In summary, we expect that improving DNA-based techniques will allow us to explore neighbourhood-scale plant richness below-ground, where much or most plant biomass occurs. This has the potential to revise many plant diversity patterns and coexistence theories, which were developed based on above-ground vegetation only.
We thank Marina Semchenko and journal reviewers for comments. This research was supported by the European Union through the European Regional Development Fund (Center of Excellence FIBIR), Estonian Science Foundation (Grants 8323, 7738), and the Natural Sciences and Engineering Research Council of Canada.