The contribution of pathogenic soil microbes to ring formation in an iconic Australian arid grass, Triodia basedowii (Poaceae)

Keywords: arid zone, Australia, culm, fire, hummock grassland, obligate seeder, seedling emergence, semi-arid zone, spinifex, soil pathogens, Triodia basedowii, clonal plants.

Ring formation is a curiosity observed among many species of clonal plants (Bonanomi et al. 2014; Watt 1947). Growth in ring-forming clonal plants is characterised by outward radial expansion and establishment of individual ramets followed by progressive die-back of inner, older roots and shoots (Bonanomi et al. 2005, 2014; Sheffer et al. 2007; Cartenì et al. 2012). Ring-forming species are often found in grassland communities (Bonanomi et al. 2014) and semi-arid or arid regions (Sheffer et al. 2007; Ravi et al. 2008) including Australian hummock grasslands (Beadle 1981; Specht 1981). In this study, we investigated ring formation in the Australian grass Triodia basedowii E.Pritz. We asked whether soil microbes might contribute to the distinctive ring growth form of T. basedowii which is commonly seen in Australia’s arid regions.

The hummock grasslands of Australia’s arid and semi-arid interior are dominated by species in the genus Triodia R.Br. (spinifex) and cover at least 1.3 × 10⁶ km² or 18% of the continent (Griffin 1984; National Vegetation Information System 2007). Although Triodia species are a grass, their ‘hummock’ growth form, consisting of a stiff tangle of culms (hollow stems) and pungent leaves (Burbidge 1945), is functionally similar to that of a shrub (Rice and Westoby 1999; Nicholas et al. 2011), where large individuals can form domes up to 2 m high and more than 2 m in diameter (Burbidge 1945; Lazarides 1997). Triodia hummocks form habitat and provide food for small mammals, lizards and birds (Murray and Dickman 1994; Daly et al. 2007; Brown et al. 2009) and accumulated biomass fuels cyclical wildfires initiating landscape regeneration (Nicholas et al. 2009, 2011; Gamage et al. 2012). Triodia species are valued by Australian indigenous people, providing resins for use in traditional tool making and medicine as well as food and fibre (Gamage et al. 2012), and pastoralists use some species for grazing (Lazarides 1997). As a result of their geographic extent, Triodia grasses are of great ecological, cultural and economic importance.

Ring formation occurs in several Triodia species (Burbidge 1945; Lazarides 1997). Of the known ring formers, Triodia basedowii is widespread across the interior of Australia, growing among sand dunes from 19 to 30°S (Lazarides 1997; Grigg et al. 2008). T. basedowii is killed by fire and is therefore an obligate seeder, relying on an accumulated soil seed bank for regeneration (Casson and Fox 1987; Westoby et al. 1988; Rice and Westoby 1999). Hummocks comprise a collection of individually rooted culms linked by stolons, where the death of roots in older culms is thought to initiate die-back within the centre of ring (Burbidge 1945). Westoby et al. (1988) found that recruitment in T. basedowii was concentrated along the outer edges of burnt hummocks despite similar densities of live seeds under hummocks and around their edges. This outward bias in seedling recruitment could explain ring-shaped hummocks in second-generation plants, but does not explain death and collapse within older hummocks (Burbidge 1946). Thus, although outward radial growth with central die-back is an established phenomenon in Triodia species (Burbidge 1945; Beard 1984), little is known of what causes the death of older culms.

The fact that many ring-forming species occur in arid ecosystems has led some researchers to suggest that water availability might drive ring-formation in some species (Getzin et al. 2016; Yizhaq et al. 2019; Herooty et al. 2020). Although models and observational data are consistent with this idea in some ecosystems, other researchers have noted that ring-formation also occurs in several ecosystems in which water limitation is highly unlikely to be important, including salt-marshes and alluvial grasslands (Cartenì et al. 2012). Further, although the presence of plants can affect soil moisture, the soil in the centre of Asphodelus ramosus rings in Israel have a comparable water content to the soil in the surrounding matrix (Yizhaq et al. 2019). Thus, in this system, water availability does not seem to provide a good explanation for ring-forming plants expanding into the matrix while dying back inside rings. Getzin et al. (2019) also found that the texture and compaction (both factors that affect moisture absorption into soil) of soil inside Triodia rings were not significantly different to that found in bare soil areas outside the rings. Finally, the proposed mechanisms through which water availability might cause ring-formation are not consistent across sites or species (Getzin et al. 2016; Herooty et al. 2020). Experimental work on the role of water availability in ring-formation would be a worthwhile next step, but this is not the goal of the present study.

Previous studies have rejected the idea that ants or termites might be responsible for the die-back in the centre of Triodia hummocks (Getzin et al. 2016, 2019).

Another possibility is that dieback in the centre of ring-forming plants could be caused by depletion of soil nutrients. Rice et al. (1994) found higher levels of soil nitrogen concentrated within T. basedowii, T. pungens and Plectrachne schinzii hummocks compared to outside, whereas available soil phosphorous was lower inside hummocks compared to outside. However, the addition of fertiliser or ash within the boundaries of hummocks did not result in new growth over 4 months following summer rain (Westoby et al. 1988). Thus, the existing evidence suggests that ring formation is not caused by nutrient limitation. We therefore did not consider soil chemistry further in the present study.

Evidence is accumulating that plant–soil feedbacks might contribute to ring-formation in some species. For example, seedlings of Scirpus holoschoenus accumulated twice as much biomass when grown in soil from outside S. holoschoenus tussocks than in soil from inside S. holoschoenus tussocks (Bonanomi et al. 2005). Interestingly, the effects were species specific, with the same soils having much less effect on seedlings of other species (Bonanomi et al. 2005). Models based on negative plant–soil feedbacks have successfully recreated the real-world patterns seen in ring-forming vegetation (Cartenì et al. 2012).

One potential mechanism for ring formation that has not been studied in Triodia is the interaction of the plant with the soil microbial community. Previous studies show that plants can accumulate pathogens in their root zone over time (Bever 1994; Van der Putten et al. 2013; Smith-Ramesh and Reynolds 2017). These pathogens can then weaken plant growth and dominance, facilitating the coexistence of plant species within communities (Klironomos 2002; Bonanomi et al. 2012; Reinhart 2012; Yang et al. 2015; Chung and Rudgers 2016). In addition, soil pathogens can reduce interspecific competitive ability (Petermann et al. 2008; Hendriks et al. 2013) and shape succession of species within a community over time (Oremus and Otten 1981; Van der Putten et al. 1993; Frouz et al. 2016). For example, seedlings grown in soil that has been exposed to the roots of conspecifics have reduced biomass compared to seedlings grown in sterilised conspecific soil or soil exposed to other species (Packer and Clay 2003; Callaway et al. 2004; Gundale et al. 2014). Plants may temporarily escape their pathogens by colonising new soil through clonal growth (van der Stoel et al. 2002; Van der Putten 2003) or by dispersing offspring away from the parent plant (Packer and Clay 2000). In this way, the outward expansion of ring-forming plants may be an escape response from accumulated soil pathogens, with die-back in the centre of the plant being due to older ramets being overwhelmed by soil pathogens. Consistent with this, a study of Bouteloua gracilis in New Mexico, USA showed that roots from the inner edge of grass rings had 1.4 times higher fungal colonisation in the field than did roots from outside the rings (Carlton et al. 2018). While live soil had an overall negative effect on plant growth, glasshouse studies did not show greater negative effects of soil from inside the rings (Carlton et al. 2018). Here, we aimed to test the idea that pathogenic soil microbes might contribute to ring formation in Triodia. Specifically, we tested the hypothesis that microbes in the soil inside naturally formed T. basedowii rings negatively affect the emergence, survival and growth of T. basedowii seedlings.

We focused on Triodia basedowii growing naturally on flat, interdune soil in the Northern Territory, Australia. Our study site was located at Deep Well Station, 40 km south of Alice Springs (Fig. 1; details in Table S1 of the Supplementary material). Mean temperatures at this site range from 21.5–36.3°C in January to 4.0–19.8°C in July (Bureau of Meteorology 2017a). The mean annual rainfall, although highly variable year to year (Bureau of Meteorology 2017b), is 282 mm and falls mainly over the warmer months (Bureau of Meteorology 2017a). Vegetation at the site consisted of an open woodland (Specht 1981) of scattered desert oak trees (Allocasuarina decaisneana) up to 15 m tall, with an Acacia shrub layer up to 3 m tall, and a ground layer dominated by ~20% cover of 30 cm high T. basedowii (Table S2 of the Supplementary material). The soil was a red, fine sandy loam with a clay content of 10–20% (Macdonald and Isbell 2009), and dry soil Munsell colour: 2.5YR 4/6. Surface soil samples from other Triodia grasslands in the southern Northern Territory have soil N of 0.015 to 0.022%, soil P of 0.007–0.015%, pH of 5.8–7 and soluble salts of 0.003–0.015% (Winkworth 1967; Rice et al. 1994). Previous significant rainfall occurred during December 2016 (125 mm) and January 2017 (40 mm) (Bureau of Meteorology 2017c). Soil and vegetation data were collected during 27–28 February 2017. Seeds for the glasshouse experiment were collected at Deep Well Station in May 2012 (B. Wright, pers. comm., 9 February 2017).

Fig. 1.  Vegetation profile at the study site. The arrangement of Triodia basedowii hummocks in relation to larger shrubs (~2 m tall) and trees is shown (image: N. Ross).

We laid out three 100-m transects with random starting locations and directions. At each 10-m interval along the transects we used a random number generator to select whether we sampled to the left or right of the transect. We then located the nearest T. basedowii that had an internal ring with a diameter greater than 30 cm. We chose a 30-cm diameter to focus our sampling on well-developed rings with substantial central die-back. This size of ring also allowed us to take soil that was unambiguously from the centre of the hummock. The sampled T. basedowii had internal diameters from 0.3 to 1 m, with a mean internal area of 0.265 m² (n = 30 plants, s.e. = 0.03). If more than one T. basedowii was encountered at the same distance away from the transect, each individual was given a number and then one was chosen using a random number generator. All sampled T. basedowii were at least 2 m (but usually more than 20 m) from established trees.

We then used a random number generator to select a compass direction from the centre of each hummock and collected soil at two points along these radii: inside the hummock, at 10 cm away from the inner edge of the living hummock, and outside the hummock, at 10 cm away from the edge of the plant. Leaf litter and organic matter were removed from the surface of the ground before taking soil samples (following Kardol et al. 2006; Knappová et al. 2016). We used cylindrical cores (10 cm in diameter and 10 cm deep) to extract soil before double-bagging it in polyethylene press-seal bags. Soil was shipped to UNSW Sydney at 4°C to minimise changes in soil biology during transport (Kardol et al. 2006). All soil collection tools were disinfected with 70% ethanol between samples to minimise cross-contamination (Abreu et al. 2013; Cortois et al. 2016).

We passed soil samples through a 2-mm sieve and removed large roots and stones. Each soil sample was then homogenised (by mixing with a spoon) and divided in half (by weight, visually ensuring equal representation of different soil textures in each fraction) to form live and sterile treatments. Each treatment was then divided into bulk soil (95%) and inoculum (5%) components and bagged in polyethylene press-seal bags.

We used 50 kGy gamma irradiation at the Australian Nuclear Science and Technology Organisation in Sydney to sterilise our soil samples. Gamma irradiation causes fewer changes in soil properties than other methods such as autoclaving (Newman et al. 1977; McNamara et al. 2003); however, all methods of soil sterilisation can cause a flush of nutrients that might affect plant growth (McNamara et al. 2003). We therefore sterilised the bulk soil components for both live and sterile treatments to reduce confounding effects. The inoculum component for the sterile treatment was also sterilised. We then recombined bulk and inoculum soil components to produce the live (95% sterile, 5% live) and sterilised (100% sterile) soil treatments for each hummock position (i.e. inside and outside the T. basedowii rings). The fact that all plants were grown in at least 95% sterilised soil minimises differences between treatments that might have arisen from the effects of sterilisation.

We placed 290 g of live or sterile soil into new 50 × 50 × 120-mm forestry pots, lined with a square of new paper towel to prevent soil leakage. Pots were saturated with water (~50 mL) shortly before seeds were sown. All equipment that came into contact with soil was disinfected with 70% ethanol between samples to avoid cross-contamination.

We extracted the seeds from the florets by rubbing them against a 1.2-mm sieve and then winnowing them (W. Lewandrowski, pers. comm., 7 March 2017). The seeds weighed an average of 0.92 mg, and were ~1 mm wide and 2 mm long. We then soaked the seeds in smoke-water solution for 24 h to release physiological dormancy before rinsing and drying them (Erickson et al. 2016; W Lewandrowski, pers. comm., 7 March 2017). Twenty-five seeds were sown into each pot (3000 in total), 1–2 mm below the surface of moist soil.

Pots were individually covered with polyethylene wrap to maintain surface soil moisture, and were placed randomly on mesh benches in the UNSW glasshouse with positions randomised weekly to minimise differences between treatments that might be caused by position effects in the glasshouse. Every 3 days, the wrap was lifted, the soil was sprayed with water and the pots were surveyed for the presence of live plants. The amount of water provided was comparable among all pots.

Seedling emergence plateaued after 18 days, after which the polyethylene wrap was removed. One randomly selected seedling from each pot was replanted in the centre of its pot. Plants were then watered at least once daily to maintain soil moisture. The growth phase of the experiment ran for a further 101 days, a duration sufficiently long to develop soil microbe related growth effects (Wagg et al. 2014) while maintaining unique microbe communities (Newman et al. 1977). At the completion of the growth phase, we carefully removed the plants from the soil, separated the roots from the base of the grass and discarded any remaining seed coats. We dried the samples at 70°C for 48 h and then cooled them in bell jars with a desiccant to reduce moisture reabsorption. Root and shoot biomass samples were then weighed using a microbalance (Mettler Toledo XP26, 1-μg precision).

We recorded four traits: seedling emergence (as a percentage of seeds planted); seedling emergence time (days); seedling survival (the number of seedlings harvested at 101 days as a percentage of initial number of seedlings at the start of the growth phase); and total seedling biomass (accumulated at the end of growth phase). Data for seedling emergence, seedling emergence time and total seedling biomass were analysed using generalised linear mixed models (GLMM) with a random effect for hummock and fixed effects for position (inner or outer) and sterilisation (sterile or live). Seedling emergence and survival (both binary) were fitted with binomial models while seedling emergence time was fitted with a negative binomial model. Data for total seedling biomass were log-transformed to improve normality. Data were analysed in R (ver. 3.4.1, R Core Team, R Foundation for Statistical Computing: Vienna, Austria, see https://www.r-project.org/) using the lme4 (ver. 1.1-13, see https://cran.r-project.org/web/packages/lme4/index.html; Bates et al. 2015) and nlme (ver. 3.1-131, J. Pinheiro, D. Bates, S. DebRoy, D. Sarkar and R Core Team, see https://CRAN.R-project.org/package=nlme) packages.

The emergence of T. basedowii seedlings was simultaneously affected by soil source position and sterilisation treatment (P < 0.001, Fig. 2a). Consistent with our hypothesis, we found that seedling emergence was 41% higher in live soil from the outside of hummocks than in live soil from the inside of hummocks (P = 0.037, Fig. 2a), and that seedling emergence inside hummocks was 46% higher for seeds planted in sterilised soil than those planted in live soil (P = 0.015, Fig. 2a). Interestingly, seeds planted in soil from outside hummocks showed 39% higher emergence in live soil compared to sterilised soil (P = 0.034, Fig. 2a). Finally, seedling emergence was 44% higher in sterilised soil from within hummocks than in sterilised soil from outside hummocks, but this difference was marginally non-significant (P = 0.052, Fig. 2a). On average, seedlings emerged 1.9 days earlier in live soil than in sterilised soil (P = 0.007, Fig. S1 of the Supplementary material), with no significant differences in emergence times detected among soil positions (P > 0.3, Fig. S1).

Fig. 2.  Effects of soil inoculation treatments on Triodia basedowii seedlings measured as: (a) emergence as a percentage of seeds planted; (b) surviving seedlings harvested at 101 days as a percentage of initial number of seedlings at the start of the growth phase; (c) total biomass (log scale) accumulated at the end of growth phase with number of remaining replicates (n) marked for each treatment; (d) picture of a T. basedowii hummock showing the location of ‘inner’ and ‘outer’ samples (image: N. Ross). Different letters over bars indicate significantly different means, and represent comparisons of all four bars within each plot (P < 0.05).

More than half of the seedlings died during the experiment with mortality unrelated to sterilisation (P = 0.730) or soil position (P = 0.219; Fig. 2b). Initial losses of 13 to 27% among treatments were caused by dehydration; however, further steady losses followed during the growth phase of the experiment. Only nine seedlings remained to represent the ‘inner-live’ and ‘inner-sterilised’ treatments, six seedlings to represent the ‘outer-live’ treatment and seven to represent the ‘outer-sterilised’ treatment. This low sample size decreased our ability to detect differences between treatments in subsequent measures. No significant differences were found in the total biomass of T. basedowii seedlings grown in soil from inside or outside hummocks (P = 0.178, Fig. 2c). There were also no significant differences in biomass between seedlings grown in live v. sterile soil (P = 0.461, Fig. 2c).

Our study showed that the presence of soil microbes might contribute to the distinctive ring-shaped growth form of the common Australian spinifex grass, T. basedowii. This idea is supported by two key findings. First, consistent with our hypothesis, we found that unsterilised soil from inside hummocks negatively affects seedling recruitment. This was evident from the significantly lower emergence of T. basedowii seedlings in live soil from inside the rings compared to live soil from outside the rings. Second, and also consistent with our hypothesis, we found significantly greater seedling emergence for seeds planted in sterilised soil from inside hummocks than for seeds planted in live soil from inside hummocks (Fig. 2a), suggesting that the presence of soil microbes may restrict T. basedowii recruitment inside hummocks. Our findings are similar to those of Bonanomi et al. (2005), who found that Scirpus holoshoenus from an alluvial grassland in Italy showed substantially less growth in soil from inside tussocks than in soil from outside tussocks.

Most current work on ring-formation in plants in arid ecosystem tends to favour water-limitation as a mechanism (Getzin et al. 2016; Yizhaq et al. 2019; Herooty et al. 2020), whereas evidence for the importance of soil microbes tends to come from more mesic environments (e.g. Bonanomi et al. 2005, 2012). Our findings, and the findings of Carlton et al. (2018) suggest that soil microbes may also have a role in shaping community dynamics and vegetation structure in arid ecosystems. This contributes to a growing body of evidence suggesting that plant–soil feedbacks are important in a wide range of ecosystems worldwide (Piercey et al. 2021).

Soil microbes are often cited as beneficial for plant performance and essential to plant survival on land (e.g. mycorrhizal fungi and rhizobia; Cain et al. 2008; Reece et al. 2012; Evert and Eichhorn 2013). Indeed, recent work on dryland ecosystem restoration has found that in some species, including Triodia, seedling growth and establishment is enhanced with the addition of specific strains of cyanobacteria applied to the seed coats (Muñoz-Rojas et al. 2018; Chua et al. 2020). Consistent with this idea, we saw significantly higher emergence of T. basedowii seeds in live than sterile soil from outside the rings (Fig. 2a). Thus, the expansion of rings outward may allow plants to reach areas with more beneficial microbiota. Conversely, in soil from inside T. basedowii rings, the effects of pathogenic soil microbes seem to outweigh the effects of beneficial soil microbes. This is consistent with findings from other studies, which show that sterilisation often increases the germination success of seeds grown in soil previously occupied by other species (Mohler et al. 2012; Lou et al. 2016), and that seedlings of plants grown in soil exposed to the roots of their own species perform better in sterilised soil samples than in soil samples that contain live microbes (Kardol et al. 2007; Gundale et al. 2014; Carlton et al. 2018).

Our results suggest that die-back in the centre of T. basedowii hummocks might be explained by older culms succumbing to a build-up of pathogenic soil microbes through time. This has been demonstrated, for example, in the European coastal pioneer Ammophila arenaria (also a grass species adapted for growth in sand-dunes), where ramets establishing in new wind-deposited sand initially grow strongly before degenerating over time in the presence of root parasitic nematodes (Van der Putten et al. 1988; van der Stoel et al. 2002). In addition, the outward expansion of T. basedowii hummocks could be explained as an escape response from accumulated soil pathogens with clonal growth away from the pathogenic centre or dispersal of offspring into soil with fewer microbes. This idea is supported by other studies, for example, seedlings of the North American black cherry, Prunus serotina, are far less affected by damping-off fungus when grown in soil away from parent trees compared to soil from within the root zone of parents (Packer and Clay 2000). This combination of die-back and escape as a result of pathogenic soil microbes is consistent with the ring-shaped growth form observed in T. basedowii.

Our study points to three worthwhile directions for future research. First, now that we know that soil microbes can have substantial effects on the establishment of arid-zone plants, it would be well worth using molecular methods to identify the types of microbes involved. Second, our finding that seedlings in live soil treatments emerged 1.9 days earlier than seedlings in sterile soil treatments suggests that perhaps different types of microbes affect seedlings at different times or under different conditions of germination (Funk et al. 2014). Identifying the microbes that are most important at different times would help us to understand the forces shaping plant recruitment, growth, and survival. Third, our non-significant results for seedling survival and biomass are likely related to a small sample size following the loss of many seedling replicates due to non-treatment effects. It would be interesting to determine whether the non-significant trend for seedlings to perform better in sterilised soil than in live soil (Fig. 2c) is upheld in a larger experiment.

Considering the extensive distribution and ecological importance of Triodia species, we know remarkably little about this iconic group of plants. Our data suggest that soil microbes may play a role in the ring formation of the widespread T. basedowii. This new information helps us further understand the unique ecology of Australia’s arid grasslands, and adds to our growing recognition of the crucial function that soil microbes play in terrestrial ecosystem processes worldwide (van der Heijden et al. 2008).

The authors declare that they have no conflicts of interest.

This study was partially funded by the UNSW School of BEES Student Research Project Fund.