Do subordinate species punch above their weight? Evidence from above- and below-ground



Variation in the relative abundance of species is a ubiquitous feature of ecological communities. Understanding the link between the relative abundance of a species and its contribution to ecosystem function is key to predicting ecosystem stability over time or during perturbations. According to the ‘mass ratio hypothesis’ (Grime, 1998), proportional inputs to primary production act as immediate controls on ecosystem function and sustainability. While dominant species are considered more important in ecosystems because of the large amount of biomass they produce, an increasing number of recent studies have shown that subordinate species may have a larger influence on ecosystem functioning than their relative abundance suggests. In particular, subordinate species can have significant impacts on soil microbial communities (Peltzer et al., 2009; Holdaway et al., 2011; Mariotte et al., 2013d). Growing recognition of how belowground processes affect biodiversity, ecosystem functioning and services (De Vries et al., 2013; Grigulis et al., 2013; Van der Putten et al., 2013) likely explains the recent interest in less abundant species. Nevertheless, studies of subordinate species remain scarce, largely because of the influence of the ‘mass ratio hypothesis’, which likely overestimates the role of dominant species, as well as a lack of adequate understanding of the distinguishing features of subordinate species. In addition to differences in relative abundance, other characteristics may differ between dominant and subordinate species including functional groups, traits, resource acquisition strategies, and spatial growth. While these differences remain poorly understood, their implications could be far reaching if subordinate species are capable of buffering ecosystem functions against perturbations, and under climate change.

This paper aims to reconcile theories about subordinate species in order to provide a framework for future studies of biodiversity effects on ecosystem functioning. I synthesize the current state of knowledge about subordinate species and give a clear definition for these species, I provide evidence of their functional role in ecosystems, and I show how the importance of this group may increase with global climate change; a topic which has not been addressed in the literature. I address each of these issues by including an above- and below-ground perspective, which is often missing from studies of functional diversity. While evidence is drawn mainly from experiments in grassland communities, the basic principles can be applied more widely.

Reconciling theories of subordinate species

Many studies have classified species in plant communities to demonstrate the importance of functional groups (Wardle et al., 2003; MacLaren & Turkington, 2010), functional traits (Lavorel et al., 2011), relative abundance (Whittaker, 1965; Grime, 1998), or keystone species (Lyons & Schwartz, 2001; Boeken & Shachak, 2006). These different components of biodiversity can have different effects on ecosystem function (Hooper et al., 2005) but are not necessarily independent. In his attempt to define the functional role of species, Whittaker (1965) recognized that the simplest way to classify components of biodiversity is to order species according to their relative abundance or productivity. The categorization of dominant, subordinate and transient species (DST classification), suggested by Whittaker (1965), was incorporated into the ‘mass ratio hypothesis’ (Grime, 1998). Grime suggested that dominant species would directly affect ecosystem properties due to the large amount of biomass produced, while subordinate species would only act as a filter, influencing regeneration by dominants following major perturbations in grasslands (Grime, 1998; Walker et al., 1999; Boeken & Shachak, 2006) or increasing diversity of climbing plants in forests (Garbin et al., 2012, 2013). Grime's classification was similar to the core-satellite species hypothesis (CSS classification; Hanksi, 1991). These classifications were compared by Gibson et al. (1999), who showed no difference between dominant and core species, between intermediates and subordinates and between satellites and transient species. Similarly, dominant species have also been referred as competitors (Grime, 1973), foundation or matrix species (Gibson et al., 2012) and subordinates as fugitive (Platt & Weis, 1985), interstitial (Keddy et al., 1994), minor (Walker et al., 1999), redundant (Rastetter & Shaver, 1996) or low abundant species.

Dominant and subordinate species intrinsically exist, as demonstrated by Olff & Bakker (1998), through a statistical test on field data. Indeed, when considering a habitat as a homogeneous area in term of vegetation type, and divided into a grid with small-scale plots, subordinate species are defined as species found in most of the plots but which never attain dominance. This means that subordinates are interspersed within the community and can occur at high frequency in plant communities but yield low relative cover compared with dominant species. The selection of both species-groups requires using replicated plots in homogenous vegetation, in order to include the maximum species diversity of the studied community. By including frequency of occurrence and relative cover, both easily measurable in the field, a generalization can be attempted to distinguish dominant and subordinate species in various ecosystems, summarized by a frequency–abundance curve in Fig. 1. Dominant and subordinate species are both frequent in plant communities and should occur in most of the plots by contrast to transients, which generally fail to regenerate and persist in the vegetation. The distinction between dominant and subordinate species is then based upon cumulative relative abundance (i.e. mean relative species abundance in a site, ranked in ascending order and cumulated for each species). As suggested by Grime (1998), subordinate species would represent 10% of the plant community. Based on experiment carried out in grasslands (Grime, 1998; Mariotte et al., 2012, 2013a,b,c,d) and shrublands (Kichenin et al., 2013), I suggest that species with a cumulative relative abundance below 2% should be considered as transients, comprised between 2% and 12% as subordinates (i.e. 10%) and > 12% as dominants (Fig. 1). This species selection has been set at arbitrary values and principally tested in grasslands. Nevertheless, this gives a first attempt to generalize the DST classification and provides basis for further research in a broad range of ecosystems.

Figure 1.

Frequency–abundance curve used for the selection of competitive hierarchical groups in herbaceous vegetation. Dominant (1, gray area) and subordinate (2, dark gray area) species are always present in the community and are distinguished from transient species (3), which do not persist in the vegetation, by a frequency > 50%. The distinction between species groups is then based upon cumulative relative abundance (%) and has been set at an arbitrary value of 12 to 100% for dominant, 2 to 12% for subordinate (i.e. 10% total abundance) and under 2% for transient species. (Modified, with permission, from Mariotte et al., 2013a).

While relative abundance has been the main factor differentiating species-groups, more recent studies suggest that dominant and subordinate may also diverge in several other characteristics. For example, dominant and subordinate species vary in their morphology and functional traits (Grime, 1998; Peltzer et al., 2009; Doherty et al., 2011; Mariotte et al., 2013d); dominant species are generally taller with large leaves (i.e. high canopy height, specific leaf area (SLA), leaf area, leaf N) allowing better light capture, while subordinate species are smaller in stature with small leaves (i.e. opposite traits; Table 1). Traits of dominant species reflect a strategy for rapid acquisition of resources, while those of subordinate species are associated with resource conservation (Grime et al., 1997; Diaz et al., 2004; Mariotte et al., 2013d). Above-ground, the differences between the species-groups result in a trade-off between biomass production of dominants and nutrient retention of subordinates (Lavorel et al., 2011). Below-ground, traits of dominant species such as high litter quality promote bacterial-dominated communities with fast microbial activities while those of subordinates (i.e. low litter quality) favor fungal-dominated communities with slower activities (De Vries et al., 2012a; Grigulis et al., 2013). Trait differences between dominant and subordinate species arise principally from grasslands but there are also few examples for forest ecosystems. For example, dominant woody species in forest communities are associated with two above-ground plant traits, high height and high shade tolerance (Koide, 2001), while subordinate woody species had relatively higher root nutrient content (nitrogen (N) and phosphorus (P)), thicker root diameter and more root hairs than dominants (Holdaway et al., 2011). Similarly, the dominance of climbing species in forest has been associated to shade-tolerance with high photosynthetic rate and low dark respiration (Gianoli et al., 2012) reflecting a strategy of maximizing exploitation of resource availability. The classification of dominant and subordinate is then intimately related to the plant economics spectrum research discussed earlier, except that studies on plant functional traits focus principally on dominant species (e.g. species making up 80% of cumulated biomass; Garnier et al., 2007; Lavorel et al., 2011) by analogy to the ‘mass ratio hypothesis’. Moreover, while measuring plant functional traits (leaves and roots) is constraining, destructive or somehow expensive, determining dominant and subordinate species based on frequency and relative cover in the field (see earlier), would be a simplified and more rapid method to estimate functional diversity and resource capture strategy of all species in the plant community.

Table 1.  Characteristics of subordinate and dominant species in grassland ecosystems
  1. SLA, specific leaf area.

Previous denominationFugitive, intermediaries, interstitial, minor, rare, redundant or low abundant speciesCompetitor, foundation, matrix or core species
Competitive abilitiesLowHigh
Relative biomass proportionLow but numerous individualsHigh but few in number
Deterministic processes

Niche differentiation

High functional dissimilarity

Habitat filtering

High functional similarity

Plant traitsLow stature, SLA, leaf area, leaf N, High root N, PHigh canopy height, SLA, leaf N, Low root N, P
Resource strategyConservationAcquisition
Above-ground trade-off

Nutrient retention

Low-quality litter

Biomass production

High-quality litter

Below-ground trade-off

Fungal-dominated communities

Slow microbial activity

Bacterial-dominated communities

Rapid microbial activity

Spatial growthAggregatedRandom

Differences in resource capture strategies between dominant and subordinate species are also associated with spatial niche differentiation and complementarity for resource use (Von Felten et al., 2009). Moreover, due to fast growth and rapid acquisition of resources, dominant species can displace subordinates from occupied patches but subordinates cannot displace dominants (Tilman, 1994; Amarasekare, 2003). Using a multidimensional trait-based approach, Maire et al. (2012) showed that within competitive plant communities, dominant species are mainly affected by habitat filtering (i.e. ecological filters selecting species with suitable traits for a given habitat), while subordinate species are stabilized by niche differentiation. These findings suggest that co-dominant species should be functionally similar and occupy the same ecological niches while subordinate species should be functionally dissimilar and occupy multiple niches dependent on specific abilities for soil nutrient preemption (Werger et al., 2002; Dassler et al., 2008) or different phenology (Catorci et al., 2012). This implies also that systems with multiple niches would promote communities composed essentially of subordinate species, while systems with low niche availability would promote dominance of few dominant species. When habitat filtering is weak relative to niche differentiation, subordinate species, favored by size-asymmetry (Hodge et al., 1996; Latenzi et al., 2012), usually grow in patches between or under the canopy of randomly growing dominants (Liancourt et al., 2009; Lamošová et al., 2010). This aggregated pattern seems to reduce exclusion of these less competitive species, at short-term (Wassmuth et al., 2009; Lamošová et al., 2010) and long-term (Porensky et al., 2012), promoting subordinate's persistence in plant community. Intraspecific aggregation also prevents dominant species from moving into patches colonized by subordinate species, or at least slow down subordinate's displacement by better competitors (Racz & Karsai, 2006).

By connecting recent studies and theories with existing literature on the DST classification, two patterns emerge. First, differences in morphology, functional traits and resource acquisition strategies that are correlated with relative abundance are likely to produce sets of characteristics that are predictably associated with dominant and subordinate species, and these characteristics may be generalizable across a broad range of ecosystem types. However, it should be noted that, because plant traits, competitive abilities and niche differentiation are influenced by environmental factors (Koide, 2001; Wellstein et al., 2013), a species can be simultaneously subordinate in a site and dominant in another depending on specific conditions. Second, recent evidence supports a much stronger functional role for subordinate species than Grime (1998) originally envisioned, suggesting that relative abundance alone cannot be used to predict ecosystem function for all species groups.

Functional role of subordinate species: a new framework

While communities are generally dominated by few dominant species, the range and number of subordinate species is highly variable and considerably influences plant diversity (i.e. species richness). The number of subordinate species strongly varies with disturbance (e.g. resources, burning, drought), and follows a humped-back model with a higher proportion of subordinates at intermediate levels of disturbance (Fig. 2; Grime, 1973; Pierce et al., 2007; Mariotte et al., 2013a). Other factors may then modify the amplitude of this curve, such as selective grazing (De Deyn et al., 2003; Van der Putten, 2005; McCain et al., 2010), trampling (Kohler et al., 2006; Mariotte et al., 2012) or mycorrhizas (see mutualism–parasitism continuum; Van der Heijden et al., 1998; Urcelay & Diaz, 2003; Mariotte et al., 2013b), depending on their relative influence on dominant and subordinate species. By conditioning the persistence of subordinate species, disturbance factors restrict the range of situations in which these species can be functionally important relative to dominants (e.g. species-rich communities with intermediate levels of resources).

Figure 2.

Relationships between species diversity and disturbance as highlighted by Grime (1987). The humped-back model includes the relative proportion of the three components of plant biodiversity: dominant (1), subordinate (2) and transient (3) species along the disturbance gradient (e.g. resources including water, nutrients, light, etc.), according to Grime (1973, 1987) and Mariotte et al. (2013a).

Based on specific plant traits, subordinate plant species may have larger impacts on ecosystem functioning than expected, especially below-ground. Indeed, fungal-dominated communities favored by subordinate species are known to reduce N leaching (De Vries et al., 2011) and to maintain productivity over the long-term. Using a 3-yr removal experiment, Mariotte et al. (2013d) demonstrated that subordinate species were associated with distinct bacterial and arbuscular mycorrhizal fungal communities, and improved plant productivity through positive plant–soil feedbacks (see also Mikola et al., 2002; Van der Putten, 2005). Similarly, Peltzer et al. (2009) highlighted the disproportionate influence of nonnative subordinate species on soil biota compared with dominant species. As illustrated by Grime (1998), the ‘mass ratio hypothesis’ is restricted in application to the role of plants, and the impacts of other trophic elements (bacteria, pathogens, symbionts) on ecosystem processes are less predictably dependent on their abundance (see also Urcelay et al., 2009). Therefore, the influence of subordinates on below-ground communities may be disproportionate to their relative abundance, and the effect on ecosystem function may be large even when the proportion of the total microbial biomass involved is low. The specific impacts of subordinate species on soil microbial communities have important implications for soil processes and nutrient cycling (Grigulis et al., 2013) and recent research from De Deyn et al. (2009, 2011) in grasslands showed that a single subordinate species (Trifolium pratense) improved soil C and N storage, whereas a second subordinate (Achillea millefolium) promoted nutrient sequestration in plant tissue. Together, these findings slightly modify the ‘mass ratio hypothesis’ (Grime, 1998) by showing that, not only dominant species, but also low-biomass subordinates, may greatly influence ecosystem processes and functioning.

The functional importance of subordinate species has been mainly investigated through the lens of compensatory dynamics and ecosystem stability (Adler & Bradford, 2002; Suding et al., 2006). If interspecific competition maintains dominance and functional redundancy is high, as expected in species-rich plant communities, better resistant subordinate species should be able to compensate for any loss of dominant species by release from competition (Adler & Bradford, 2002). The increasing interest in climate change has given a new importance to this process through the insurance hypothesis (Yachi & Loreau, 1999). Recent findings suggest a significant role of subordinate species in the resistance of plant communities to climate change. Results of a mesocosm experiment (Kardol et al., 2010) and a field experiment in mountain grassland (Mariotte et al., 2013c) showed that subordinate species increased their biomass production during drought and enhanced community stability. Both studies suggested that dominant species responded most strongly to the direct impacts of drought, whereas subordinate species were more resistant to drought, and responded to reduced competition with the dominant species. Interestingly, the fungal-based soil food webs associated with subordinate species have also been shown to be more resistant to drought than the bacterial-based food webs found with dominant species (De Vries et al., 2012b). This finding suggests that plant–soil feedbacks and litter promote more resistant fungal communities (e.g. mycorrhiza), which then improve the resistance of subordinates against perturbations. The resistance of subordinate species to climate change is not limited to drought; similar responses have also been observed under warming in sub-arctic dwarf shrub community (Richardson et al., 2002), and under elevated carbon dioxide (CO2) (Navas et al., 1997; Stöcklin & Körner, 1999; Maestre et al., 2005). In both cases, subordinate species improved biomass production and N uptake, suggesting that subordinates can increase their biomass relative to dominant species when resources (e.g. water, CO2) increase in initially low resource systems, or when resources decrease in initially high resource systems (Fig. 3). Therefore, while subordinate species are expected to have greater effects on ecosystem functioning at intermediate resource levels where they are most abundant, these species may become more important at low and high resource levels under climatic perturbations (Fig. 3) and may compensate for the loss of less resistant species to improve stability. However, findings on subordinate species are still scarce and while the abundance of subordinate species seems to be related to resources, less is known about the functional importance of subordinates in relation to soil microbial communities along the resource gradient. This is also not entirely clear if the higher resistance of plant communities to climate change is related to higher abundance of subordinates or only to the presence of one or few subordinate species. Determining the mechanisms of resistance and resilience to climate change is an important contemporary theme of research at broad scale and the relative abundance of subordinate species in plant communities might explain, at least partially, why some communities are more resistant than others. Future experiments utilizing the new framework presented here (Fig. 3) could easily test the functional role of subordinate species under present and projected climate in a large range of ecosystems and environmental conditions to draw more general patterns for the functional role of these species.

Figure 3.

Conceptual model representing the dominance–rank curve (dominant (D) vs subordinate (S)) in function of available resource (e.g. water, CO2, etc.) and changes induced by a shift (increase or decrease) in resources. The dotted box summarizes cases where subordinate species are expected to be functionally important (i.e. subordinate insurance hypothesis) either due to an increase in abundance or biomass, depending on the initial resource state and the direction of resource shifts. Gray boxes highlight cases forecasted under climate change, such as an increase in a resource that is typically scarce in a particular ecosystem (e.g. an increase in spring rainfall in Mediterranean areas) or a reduction in a resource that is typically abundant (e.g. a decrease in summer precipitation in temperate ecosystems).

Future directions

The recognition of ecosystem-level effects of subordinate species is recent, probably because their effects were missed in experiments with randomly assembled communities (Bardgett & Wardle, 2010), which generally neglect the relative abundance of species and functional traits in the community. By synthesizing existing knowledge on this species-group, this paper addresses these issues and highlights the ‘subordinate insurance hypothesis’, suggesting that subordinate species may assist dominant species or compensate for their loss on ecosystem functions. Without refuting the ‘mass ratio hypothesis’, this synthesis shows that subordinate species also matter in ecosystems and emphasizes the importance of below-ground processes, which remain poorly understood. For example, the specific root-associated microbial communities and below-ground traits of subordinates have not been well studied. Another aspect to be considered is the functional effect of a single subordinate (i.e. species identity) vs the effects of several subordinate species (i.e. species-group). By using the new framework presented here, future research should assess these mechanisms to understand whether and to what extent subordinate species may buffer or stabilize ecosystem functions under climate change. This shall definitely include (1) whether higher community resistance is related to higher number of subordinates at intermediate resource level or whether one single or few subordinate species at high and low resource levels may also compensate for the loss of dominant species, (2) whether the role of subordinate species is similar under different environmental conditions and (3) whether traits of subordinate species change under climate perturbations to fulfill the same functions than dominant species. The classification of dominant, subordinate and transient species is well adapted to a range of ecosystems (e.g. forests, grasslands, wetlands, etc.) and given the key role of subordinate species where they have been studied, future challenges include determining their importance at a broader scale and for multiple ecosystem functions, in order to better understand patterns of functional diversity.


The author is grateful to Katharine Suding, Paul Kardol, Emily Farrer, Erica Spotswood, Lauren Hallett and Alexandre Buttler for their assistance in the writing of this paper and to the Swiss National Science Foundation, which supported the research (no. 31003A 114139 and PBELP3 146538).