Co-occurring bunchgrasses are associated with different plant species in dry pine savannas

Introduction

Savannas are predominantly herbaceous ecosystems with high understory plant diversity that is widely maintained by some degree of fire (Strömberg 2011). Restoring fire and sustaining plant species richness are complementary conservation goals in these systems, but approaches can differ. Grasses are often targeted to restore fire in savannas because they are the dominant fuel for frequent, low-intensity fires (Mulligan et al. 2002; Butterfield et al. 2017). In this case, one or a few flammable grass species might be targeted (Aschenbach et al. 2010; Chang et al. 2018), which should be enough to restore the functional role of fire. On the other hand, similar grass species might have different associations with other plants, such that restoration goals to maximize biodiversity might be achieved by including several functional players. For example, other species groups like cushion plants (e.g. Laretia spp. and Azorella spp.) are functionally similar yet can support distinct plant community compositions (Cavieres et al. 2008; Hupp et al. 2017). Though there are many mechanisms for associations among plants, bunchgrasses are often structurally dominant in savannas and are strong candidates as foundation species (Ellison et al. 2005; Angelini et al. 2011), such that they could directly or indirectly support habitat for a variety of other species. Thus, targeting multiple, functionally similar species during restoration could benefit ecosystem function and biodiversity conservation.

Restoring prescribed fire is a common conservation target in pine savannas of the southeastern United States. In fact, fire suppression is one of the major causes of habitat degradation in these ecosystems, which have been dramatically reduced in area (i.e. presettlement to present day; van Lear et al. 2005; Hanberry et al. 2018). Pine savannas have an exceptionally diverse understory dominated by fine fuels such as grasses and pine needles that sustain feedbacks with fire (Fill et al. 2015; Noss et al. 2015; Robertson et al. 2019). Wiregrass (Aristida beyrichiana Trin. & Rupr.) is an endemic, perennial bunchgrass usually preferred as a candidate for understory restoration where the goal is to reinstate frequent fires (Trusty & Ober 2009). This species is widespread in pine savannas along edaphic gradients (Wells & Shunk 1931), important for vertebrate diversity (Means 2006), able to establish in degraded sites (Walker & Silletti 2006), resilient to climate extremes once established (Young et al. 2021), and a key driver of the grass-fire feedback (Fill et al. 2016). Wiregrass-dominated understories also support many endemic and rare species (Hardin & White 1989; Walker 1993), and wiregrass itself might facilitate pine seedling establishment (Miller et al. 2019).

Although targeting wiregrass for restoration in pine savannas is a common approach, including other flammable but less-targeted bunchgrass species might promote biodiversity as well as fire because they could have associations with different plant species. Pineywoods dropseed (Sporobolus junceus [P. Beauv.] Kunth) is locally co-dominant with wiregrass in dry pine savannas (Gates & Tanner 1988) and can only be distinguished from Aristida beyrichiana in the vegetative state by the lack of a minute tuft of hair where the leaf blade meets the sheath (Hall 1989; Canfield & Tanner 1997). Both species are important groundcover fuels and support functional fire spread through the understory. We tested the hypothesis that in a reference system, wiregrass and pineywoods dropseed are two structural dominants that associate with different plant species, thereby providing a guide to whether planting similar species will help meet multiple restoration goals of re-instating fire feedbacks and promoting plant species diversity. In this study, we investigated whether wiregrass and pineywoods dropseed are (1) associated with different species richness of co-occurring plant species and (2) have different compositions of understory plant species under or adjacent to their canopies.

Methods

Sampling

During the fall of 2019, we selected 30 focal bunchgrass individuals (i.e. 15 wiregrass and 15 pineywoods dropseed) at each of 4 sites (120 total bunchgrass individuals). All sites burned in 2019, except one that had last burned in 2017. Bunchgrass individuals were selected as we encountered them to span a range of sizes (wiregrass: 9–1779 cm²; pineywoods dropseed: 35–1,059 cm²; see Fill et al. 2021 for methods used to measure basal area), and thus selection was not truly random. We sampled plant community composition by recording the number of stems of each understory plant species present. Because our sampling was not standardized with time since fire, we did not record species percent cover in each plot. Each plot was divided into two quadrats (0.33 × 0.33 m) that were positioned to sample the plant community (1) under the canopy of draping bunchgrass leaves and (2) adjacent to the focal bunchgrass just outside of the bunchgrass canopy (Fig. 1A). To avoid multiple bunchgrass individuals affecting the sampled area, the plot frame, which included the under and adjacent quadrats, was situated on each focal bunchgrass such that the adjacent quadrat did not have any bunchgrass individuals other than the focal plant within 1 m.

Results

Species richness did not differ between bunchgrass species or by proximity to bunchgrasses. We found that species richness was approximately 1.2 per quadrat (0.33 × 0.33 m) whether occupied by wiregrass or pineywoods dropseed (estimate ± SE = 0.108 ± 0.157, p = 0.464; Fig. 1B), under or adjacent to bunchgrasses (estimate ± SE = −0.279 ± 0.172, p = 0.104; Fig. 1B). Some quadrats had no plants other than the focal bunchgrass, whereas others had as many as six. There was also no interaction between bunchgrass species and proximity (estimate ± SE = −0.102 ± 0.241, p = 0.673; Table S1).

Rarefaction curves confirmed similar species richness levels under and adjacent to both species. Estimated species richness under wiregrass, under pineywoods dropseed, adjacent to wiregrass, and adjacent to pineywoods dropseed (with 95% confidence intervals) was 28.0 (21.0, 35.0), 20.4 (16.2, 24.6), 20.0 (16.5, 23.6), and 16.3 (13.4, 19.1). Richness tended to be lower adjacent compared to under each bunchgrass species.

Although there were no differences in species richness, our multivariate analysis showed that the occurrence of different plant species varied between wiregrass and pineywoods dropseed (Dev = 109.5, p = 0.005), while proximity to the focal bunchgrass did not affect species occurrence (Dev = 35.47, p = 0.39). Several species were found in the proximity of only one or the other bunchgrass species (Fig. 2). However, according to our model, the observed difference in species occurrence was primarily determined by a few species: Andropogon capillipes (p_unadj. = 0.025), Bulbostylis barbata (p_unadj. = 0.005), Carex sp. (p_unad. = 0.005), Q. laevis (p_unad. = 0.005), Smilax auriculata (p_unadj. = 0.04), and Vaccinium myrsinites (p_unadj. = 0.045). Of these species, A. capillipes was only present near wiregrass, while the other species were more likely to occur near pineywoods dropseed (Fig. 2). Similarly, species abundance also differed between wiregrass and pineywoods dropseed (Dev = 105.09, p = 0.005) irrespective of proximity to the focal bunchgrass (Dev = 31.06, p = 0.585). Apart from S. auriculata, differences in species abundance were driven by the same species that were driving differences in species occurrence: A. capillipes (p_unadj. = 0.04), B. barbata (p_unadj. = 0.05), Carex sp. (p_unadj. = 0.02), Q. laevis (p_unadj. = 0.01), and V. myrsinites (p_unadj. = 0.025). As in the previous analysis, A. capillipes was the only species that increased in abundance near wiregrass, while the other species were more abundant near pineywoods dropseed.

Discussion

Restoring two morphologically similar bunchgrass species may promote understory plant diversity more than restoring an individual species. Although species richness was similar in quadrats associated with wiregrass and pineywoods dropseed, the occurrence and abundances of some understory plant species differed between quadrats associated with each bunchgrass species. Thus, including more than one bunchgrass species in restoration planning could enhance species diversity in xeric pine savannas while also promoting the grass-fire feedback. Actual restoration outcomes may differ from the dynamics observed in our reference system, however, documentation of experimental and practical projects involving various bunchgrass species is valuable. For example, Aristida stricta, the widespread counterpart to A. beyrichiana, is restricted to the northern range of pine savannas and likely fulfills a similar ecological and restoration role as A. beyrichiana in the southern range. Similarly, congeners of pineywoods dropseed such as Sporobolus pinetorum that are similar in growth form and habitat could affect patterns of understory plant species diversity. Functional redundancy has been shown to promote ecosystem stability and resilience (Biggs et al. 2020), implying that restoring multiple, functionally similar species might help meet restoration goals (Laughlin et al. 2017).

Various ecological mechanisms might structure associations between plant species, including direct competition or facilitation. Although competition is generally considered one of the primary drivers of plant community dynamics (Grime 1977; Connell 1983), competitive interactions among understory plant species in pine savannas have been shown to be weak (Kirkman et al. 2001; Roth et al. 2008). For example, Myers and Harms (2009) found limited evidence that grass competition reduces understory plant species richness. On the other hand, facilitative interactions play an essential role in shaping plant communities in savannas (Loudermilk et al. 2016; Miller et al. 2019; Peláez et al. 2019), some of which could be related to unique microclimatic conditions generated by the architecture of different grass species (Vinton & Burke 1995; Iacona et al. 2012) or to varying degrees of flammability (Simpson et al. 2016). For example, those species that are more resistant to differences in grass flammability would be more likely to persist and maintain an association with that grass species. The interplay between these mechanisms is likely complex because of the scales on which microclimatic conditions and fire are operating and could also be affected by overstory fuel inputs, which can alter flammability and microclimatic conditions (Mitchell et al. 2009).

Although we examined wiregrass and pineywoods dropseed associations in natural pine savannas and not in restoration sites, we expect that including multiple species should increase biodiversity in either context (Bullock et al. 2007). However, careful consideration of conservation and restoration goals should drive decision-making (Aschenbach et al. 2010; Bullock et al. 2011). For example, information on direct wiregrass–pineywoods dropseed interactions is currently lacking, such that it is possible only a few species might dominate in multispecies plantings. Moreover, grass species may flower at different times in pine savannas, so collecting seeds from multiple grass species for restoration sowing might be impractical or cost-prohibitive in some instances (Armstrong 2021). Conservation goals that include both fuel-fire feedbacks and increasing biodiversity will likely require that multiple species are reintroduced.