Plant dominance in a subalpine montane meadow: biotic vs. abiotic controls of subordinate diversity within and across sites

Study site

Our study sites were located at the Rocky Mountain Biological Laboratory (RMBL), Gothic Colorado (latitude 38°53′N, longitude 107°02′W, elevation 2,920 m above sea level) (Saleska, Harte & Torn, 1999). Annual precipitation averages 750 mm, 80% of which was snow (snowmelt typically ending in May) (Saleska, Harte & Torn, 1999; Harte & Shaw, 1995). Mean daily-average summer air temperature is ∼10 °C. Mean snowfall at RMBL is 11,140 cm with a 24-year trend towards lower snowfall overtime (Inouye et al., 2000) with field summer seasons ranging from 0.69 m and 0.47 m (water equivalent), respectively (Harte & Shaw, 1995). Soil texture is a well-drained Cryoboroll, which is a deep rocky outcrop that is non-calcareous and formed on a glacial till (Saleska, Harte & Torn, 1999). Below a sparse litter layer (due to snowpack), the soil is uniform in color and texture down to about 50 cm. Organic content averages ∼10% at a soil depth of 5 cm below the litter layer and drops to ∼6% at 50 cm (Harte & Shaw, 1995). Soils at the experimental and observational sites averaged a pH of 5.7–6.3 (Saleska, Harte & Torn, 1999).

Experimental design

Plant community measurements

To examine how subordinate diversity varied along a plant dominance gradient (Observational Study) and how dominant species directly affected subordinate diversity (Experimental Study), we measured species-specific foliar cover, species richness (the number of species), Shannon’s diversity and evenness in each observational and experimental plot twice in each growing season (Observational Study = one growing season, Experimental Study = three growing seasons). To estimate species-specific foliar cover, we used a modified Braun-Blanquet scale that included six categories: <1%, 1–5%, 5–25%, 25–50%, 50–75%, 75–100%. The median of each cover class category was assigned to each plant species in each plot and used as an estimate of species-specific abundance. Shannon’s diversity (H′) was calculated as: H′ =  − sum(pi∗(ln∗pi)) and evenness was calculated as J′ = H′∕S, S is species’ richness.

Above and below ground functional trait measurements

To determine above- and belowground variability across the three sites (varying in dominance of Festuca and Potentilla), two 30-meter transects were established at each site (outside of our plant sampling plots). Every six meters, a 1 m × 1 m plot was placed (totaling 5, 1-m² sampling plots per transect and N = 10 per site). At each plot for Festuca and Potentilla, percent species-specific foliar cover was recorded (see methods above) and leaves and roots were harvested according to the methods by Cornelissen et al. (2003). We quantified specific leaf area (SLA) by harvesting three relatively young, but fully expanded leaves from each individual. Leaves were then scanned and we used ImageJ to estimate leaf area (cm²). Leaves were then oven dried for approximately 48 hours at 65 °C and weighted. We then divided area by mass to obtain SLA (cm² g⁻¹). We sampled absorptive roots from a single individual of each species (Festuca and Potentilla) per plot in each transect to estimate specific root length (SRL, cm g⁻¹). For Potentilla we dug up the entire plant and root systems with a spade and then bagged the entire mass for later analysis. For Festuca we used a soil core (one cm in diameter ×10 cm in length) to sample roots from the plant species, by angling the soil core into the base of the plant, to ensure only roots of that species were extracted. Ten fine root pieces from each core were separated and used for analysis. Using ImageJ again, we scanned and interpreted the data using a Plugin called ‘IJ Rhizo v0beta’. We then oven dried roots for approximately 48 hours at 65 °C and weighted.

Microclimate measurements

To determine how resources and conditions varied along a dominance gradient (Observational study), as well as being impacted by dominant species (Experimental study), we tracked light and soil nutrient availability (resources) as well as soil temperature (conditions). We measured photosynthetic active radiation (PAR, hereafter light availability) once during the peak of the growing season (July) in each of the experimental and observational plots. Soil temperature and soil nitrogen availability were measured overtime by being deployed, ibuttons and resin bags, in early June and retrieved from plots in late July. To estimate light availability below the vegetation canopy, we used a line-integrating ceptometer (Decagon Accupar; Decagon Devices, Pullman, WA) with all light availability measurements made on clear days between 11 am and 2 pm. Specifically, we placed the line-integrating ceptometer approximately 2 centimeters from the ground. To determine soil temperature, we used ibuttons (MAXIM) that recorded surface soil temperature every minute. To assess the availability of NO3-N and NH4-N in the soil solution, we placed mixed-bed ion-exchange resin bags in nylon stockings (H-OH form, #R231-500; Fisher Scientific International Inc., Pittsburgh, PA, USA) at five-cm soil depth at two locations in each of the experimental and observational plots (Hart et al., 1994). Resins were then air-dried, and two g of resins from each plot were extracted with 2 M KCl. Pool sizes of NO3_ and NH4 + were analyzed on a Lachat AE Flow Injection Autoanalyzer (Lachat Quikchem 8000; Hach Corporation, Loveland, OH). All values expressed in this article are based on air-dried resins.

Statistical analyses

To determine how subordinate species diversity, as well as above- (SLA) and below-ground (SRL) plant functional traits, varied along a plant dominance gradient (Observational study), we ran a series of one-way analyses of variance (ANOVAs) with ‘site’ as our main factor (e.g., Potentilla dominated, Festuca dominated, Co-dominated). To determine the direct role of dominant species on subordinate species diversity we performed one-way ANOVAs with ‘plant removal’ (control, Potentilla removed, Festuca removed) as our main fixed effect (Experimental study). All the ANOVA analyses were conducted using Jump 11 (JMP).

To determine (1) how compositional similarity of subordinate species varied along a plant dominance gradient and (2) how dominant species affected compositional similarity of subordinate species, we generated a Bray-Curtis similarity matrix from the log transformed plant composition (log x+1). We then performed a permutational multivariate analysis of variance (PERMANOVA; Anderson, 2001) on the Bray Curtis similarity matrix. A pseudo F-ratio is calculated within the PERMANOVA framework comparing the variability in species composition both within treatments and among treatments based on the observed variability in species composition vs. the variability in species composition using a generated null distribution (Anderson, Ellingsen & McArdle, 2006). PERMDISP (permutational multivariate analysis of dispersion) analysis, on the other hand, is a measure of ‘dispersion’ of community composition in multivariate space (Anderson, Ellingsen & McArdle, 2006). We used PRIMER version 1.0.3 (Plymouth Marine Laboratory, UK) for these analyses. We performed a series of principal coordinate analyses (PCoA) to illustrate species compositional similarity and dissimilarity in a two-dimensional multivariate space,. Finally, we used a similarity percentage analysis (SIMPER) to determine the relative contribution of plant species driving compositional dissimilarities both in the observational as in the experimental study.

Community structure and compositional similarity across a dominance gradient (Observational study)

Plant community structure differed across a dominance gradient. Co-dominated sites, generally, had greater total cover, richness, evenness and diversity than Potentilla or Festuca dominated sites (Table 1, Fig. 1A). For example, total cover was 26% greater while evenness, richness, and diversity were 6%, 24%, and 15% greater respectively when both Festuca and Potentilla co-dominated than when either species became dominant. Plant species composition, similar to community diversity, differed across a dominance gradient (Table 1, Fig. 1B). While all sites differed from one another in compositional similarity, co-dominated sites differed the most in compositional similarity to either Potentilla or Festuca dominated sites.

Shifts in above-and belowground functional traits across a dominance gradient (Observational study)

Both above- and belowground plant functional traits varied along a plant dominance gradient, but plant identity dictated such variation. For example, the average area of Festuca leaves (F = 9.24, P = 0.001), but not SLA (P > 0.05), was 40% greater in Festuca dominated (9.79 ± 0.54) and co-dominated sites (11.15 ± 0.75) than in Potentilla dominated sites (6.86 ± 0.84) where Festuca is subordinate. On the other hand, Festuca SRL was 38% greater when it became subordinate (e.g., Festuca 2712.88 ± 351.58 in Potentilla dominated site) than when it dominated local communities (1990.39 ± 214.44 in Festuca dominated site) (F = 9.37, P = 0.001).

Similarly, Potentilla differed marginally in aboveground functional traits, while differing strongly in belowground functional traits across a plant dominance gradient. For the aboveground functional traits, the average area of Potentilla leaves was greater in Festuca dominated than co-dominated or Potentilla dominated sites (F = 2.83, P = 0.07). Potentilla dominated site leaf area was on average 143.9 cm² ± 19.7 while in co-dominated site and Festuca dominated site leaf area was 171.3 cm² ± 6.9 and 170.2 cm² ± 20.35, respectively. Specific leaf area, on the other hand, did not differ (F = 0.84, P = 0.44) across dominance gradient for Potentilla (Potentilla dominated site: 103.4 cm² ± 10.8; Co-dominated site: 102.5 cm² ± 10.21; Festuca dominated site: 131.72 cm² ± 8.42). Belowground functional trait (SRL) for Potentilla was 30% greater when (F = 3.77, P = 0.35) Potentilla was a co-dominant than when it was a dominant (Potentilla dominated site) or subordinate (Festuca dominated site).

Community structure, compositional similarity (experimental study)

We found that neither dominant species, Potentilla or Festuca, affected plant richness, evenness and diversity (Table 2, Figs. 2A–2D, Appendix Fig. S1). Similarly, compositional similarity was not impacted by the removal of neither dominant species (Table 2, Figs. 2E–2F, Appendix Fig. S2).

Microclimate across a plant dominance gradient (Observational & Experimental study)

We found light availability and temperature, but not soil N availability, to vary along a plant dominance gradient. Light availability was 25% greater in Festuca dominated sites than co-dominated or Potentilla dominated communities. Further, co-Dominated sites and Festuca dominated sites had the largest minimum and maximum temperature difference (60.20 °C and 60.11 °C, respectively). Potentilla dominated sites had a lower temperature difference of 53.28 °C, which coincides with having the lowest light availability measurements within these plots. Finally, we found no effects of dominant species on resources (light availability and soil N) or conditions (soil temperature) (Table 3).

Community diversity and compositional similarity across a dominance gradient: observation vs. experiment

Co-dominance by Potentilla and Festuca was associated with greater subordinate species’ abundance and overall diversity than when either Potentilla or Festuca were dominant in a plot, along a dominance gradient. Dominant species have been shown to strongly impact subordinate species’ abundance and biodiversity by disproportionately utilizing resources or conditions that would otherwise be available for subordinate species, especially in favorable (Wardle et al., 1999; Wilsey & Polley, 2002; Diaz et al., 2003) rather than unfavorable environments (Smith et al., 2004). Under favorable conditions, dominant species can have antagonistic effects on subordinate counterparts given their higher competitive abilities with higher resource availability. Over–yielding, dominant individuals may modify resources and conditions for subordinates drastically. Similar to our documented patterns, Suding & Goldberg (2001) found dominant species to reduce subordinate biodiversity by monopolizing resources and therefore exhibiting greater resource uptake rates, reducing subordinate species’ abundance. Plant dominance may not only lower biodiversity at the plot level, but overall diversity among subordinate assemblages. Belowground biotic interactions are also critical in determining dominance patterns and interactions between dominant and subordinate species. For instance, fungi and earthworms have been shown to strongly influence the individual growth but also the composition of subordinate species in ecological communities (Mariotte et al., 2015; Mariotte et al., 2016; Mariotte, Canarini & Dijkstra, 2017). Further, lower among assemblage diversity or spatial–temporal homogenization of subordinate species, can lead to a potentially long-term biodiversity deficit due to a lower regional species-pool which won’t be resupplying local subordinate assemblages with more plant species propagules (Huston, 1999; Witman, Etter & Smith, 2004). In other words, co-dominance patterns could promote short- (richness, evenness, diversity) and long-term (compositional similarity) biodiversity patterns in montane meadows.

However, when we directly tested for the effects of co-dominant species to influence subordinate biodiversity we found that removing either Festuca or Potentilla did not affect subordinate diversity. Similar to our findings, Smith & Knapp (2003) also found that dominant species did not affect subordinate species diversity. Smith and Knapp found that after removing dominant C₄ grasses the subordinate assemblage in the grassland ecosystem did not compensate for the loss of dominant species. Instead, they found that subordinate productivity was unaffected by even a 50% reduction in density. In a field experiment conducted by Souza, Weltzin & Sanders (2011), diversity of the subordinate community was found to be on average 20% greater in plots with the removal of a dominant forb species, Solidago altissima. Similarly, in Verbesina removal plots, diversity on average was 30% greater than in plots where Verbesina was present. Even though the removal of dominant species affected diversity and evenness, there were no effects on composition of these plots, because richness did not change (Souza, Weltzin & Sanders, 2011). On the other hand, dominant species may attain high abundance by being good ‘stress-tolerators’ rather than a great competitor relative to other species (Read et al., 2017) that have low abundance and classified as subordinate or transient.

Whittaker (1965) suggests that a closer look should be taken to differentiate between subordinate and transient species, as there is a keen distinction that separates them. Where subordinates consistently co-occur with specific dominants in larger abundance than the dominants, though smaller in build, transient species lack consistency of association with dominants and infrequent occurrence temporally and spatially. Transient species have been found to make a small contribution to biomass, though most are species that occur as dominants or subordinates in other communities, often nearby. Though through our observations and experiments, we found that dominant species did not affect subordinate diversity, as the majority of the plant species in our montane meadows were transient and very few were actually subordinate species (see Appendix Table S1).

Changes in biodiversity, along the plant dominance gradient, translated into divergence in subordinate species’ compositional similarity. Co-dominated subordinate communities exhibited greater equitability of subordinate forbs than either Festuca or Potentilla dominated communities that exhibited two main subdominant forbs making up subordinate species assemblages. Co-dominated communities had a greater proportion of perennial forb species that differed in identity from perennial forbs in sites domintated by either Potentilla or Festuca. For example, co-dominated communities had a greater abundance of Erigeron speciosa, Artemesia ludiciviana, and Fragaria virginiana than Potentilla or Festuca dominated sites (which had greater proportion of perennial forbs such as Helianthella quinquenervis and Thalictrum fendleri) which is a clear shift in composition.

Similar to our documented patterns, dominant species have been found to alter compositional similarity of subordinate assemblages (Grime, 1998). However, when we directly tested for the effects of co-dominant species to influence subordinate species composition we found that removing either Festuca or Potentilla did not affect subordinate compositional similarity. Similar to our findings, Souza, Weltzin & Sanders (2011) also found that dominant species did not affect subordinate species composition in an old-field ecosystem. For instance, when Souza, Weltzin & Sanders (2011) removed either C₃ perennial forb: Solidago or Verbesina species, compositional similarity of subordinate species did not converge or diverge relative to dominant species removal treatments. Further, other plant removal studies were only able to detect strong effects of dominant species on subordinate species composition over longer time frames (e.g., 8 years or longer) (Schmitz, 2003; Munson & Lauenroth, 2009).

Shifts in above-and belowground functional traits across a dominance gradient

Plant functional traits varied along the plant dominance gradient, but the documented patterns did not support our original prediction that functional traits would be associated with dominance patterns. Specific root length, plant allocation towards greater investment on root area than mass increasing surface area to volume ratio that promotes greater resource uptake, increased for both Festuca and Potentilla when they became more subdominant than dominant. Such shift in belowground traits for both Festuca and Potentilla likely resulted from greater resource competition when they are subdominant than dominant. Such belowground strategy differs from other studies that have found subdominant species to generally have root traits associated with resource conservation rather than rapid acquisition (Mariotte et al., 2013a; Mariotte, 2014). Perhaps montane plant communities with narrower growing season windows relative to other systems, foster greater plasticity in belowground traits that promote persistence of subdominant species. Surprisingly aboveground functional traits, such as SLA did not shift when Festuca and Potentilla became subdominant. Greater total leaf area production (e.g., greater foliar cover) in dominated sites promoted dominance regardless of changes in leaf function. There are many different factors that can contribute to a lack of correlation in diversity and above- and belowground functional traits, such as abiotic constraints (Hooper et al., 2000): species or groups of plants could be responding to different abiotic constraints, such as soil nutrients and water availability (Hooper et al., 2000). Though above- and belowground functional traits do not directly associate with species relative abundance, functional traits of dominant plant species influence ecosystem resilience and resistance. Generally, communities dominated by slow growth plants tend to have low resilience and high resistance, while the opposite is true for communities dominated by fast growing plants (Aerts, 1995; Leps, Osbornovakosinova & M, 1982; Macgillivray et al., 1995). However, a recent study performed in a montane meadow nearby (with greater dominance of Festuca than our plant removal plots) found fast compensatory responses of functional traits in subordinate species in removal relative to control plots (Read et al., 2017).

Biotic and Abiotic filters determining species’ relative abundances

Biotic and abiotic filters can determine the distribution and relative abundances of species across space and time. Abiotic filters, such as environmental factors like climate, can dictate the distribution and relative abundance of species across biomes (Whittaker, 1965; Grime, 1979; Huston, 1999; Pavoine et al., 2011). Biotic filters, such as species interactions as in the form of predation or competition, can influence the relative abundance of species in local assemblages (Mouquet & Loreau, 2003). Subordinate diversity and composition in our system are likely shaped by differences in environmental factors. Similarly, sedge dominated plots varied in relative abundance due to soil nutrient as an abiotic factor in montane meadows studied by Theodose & Bowman (1997). In these montane dry meadow and wet meadow sites, Theodose and Bowman observed changes in community composition and diversity following additions of nitrogen and phosphorous fertilizers over a five-year study. In the dry meadow, Theodose and Bowman found species diversity increased significantly with fertilization, in the form of Nitrogen and Phosphorus, over the course of the study. This increase of diversity seems to have been due to an increase in the relative abundance of rare species, while the dominant species declined. In juxtaposition, the wet-meadow species diversity decreased in response to fertilization over the course of the study. This comparison allows for the comparison of the effects of fertilization on diversity between communities that differ in resource availability (Theodose & Bowman, 1997).