Do native grasses emerge and establish in areas rehabilitated using vetiver grass?

Species-rich grasslands provide important ecosystem services, and in South Africa, approximately 40% of these grasslands are degraded. Vetiver grass (from India) is often used during rehabilitation efforts to restore soil function without a thorough understanding of the potential negative ecological impacts. Hence, a study was initiated to investigate vetiver’s ecological impacts during grassland rehabilitation. Firstly, a field survey was conducted using a contiguous quadrat method to evaluate the extent of grass secondary succession in these rehabilitated sites. Secondly, the effect of vetiver competition and seed sowing method on the recruitment of two native grasses (Eragrostis curvula and Megathyrsus maximus) was examined using pot trials. The field survey results showed no evidence of grass secondary succession, but rather the abundance of bare ground around vetiver, and a marked increase in grass species richness with increasing distance from planted vetiver. Subsequently, in the pot trial, vetiver facilitated emergence in both native grasses, and soil surface sowing of indigenous grass seeds showed greater emergence than other sowing methods. However, vetiver inhibited native grass seedling establishment, even when root competition was excluded. This study suggests that areas rehabilitated using vetiver are unlikely to become productive grasslands with good grazing, because vetiver inhibits colonisation by native grasses.

Grasslands provide a wide range of ecosystem services and South Africa's grasslands are a remarkable and irreplaceable global biodiversity asset (Bengtsson et al. 2019;Binder et al. 2018;Buisson et al. 2021).In particular, ancient grasslands (also referred to as old-growth grasslands), which are dominant in South Africa, reflect grassland's coevolution with fire and herbivory (Buisson et al. 2021;Silveira et al. 2020;Bond 2016).The ecosystem services that these grasslands provide include water provision (strategic water sector) (Le Maitre et al. 2018), fodder for livestock (Tilman et al. 2006), carbon storage/climate regulation (Smith 2014), biodiversity maintenance (Bengtsson et al. 2019), habitat for wildlife (Maphisa et al. 2016), sources of nectar (Nicolson and Wright 2017;Kleijn et al. 2015), soil formation and protection (Abdalla et al. 2020), nutrient cycling (Daba et al. 2018), and aesthetic beauty provision (Bengtsson et al. 2019).Notably, these grasslands need disturbance (i.e.fire and/or herbivory) to maintain the dominance of the herbaceous layer and ultimately the provision of these ecosystem services (Buisson et al. 2021;Silveira et al. 2020;Bond 2016).However, many current land-use practices, that suppress or completely remove these disturbances (i.e.fire), are degrading grasslands through loss of diversity and loss of soil (soil erosion) (Dlamini et al. 2014;Bengtsson et al. 2019).This hinders the grassland's ability to provide these essential ecosystem services.
Grasslands cover about 30% of South Africa's land surface, but about 40% of this has been converted and degraded by agricultural activities and poor management, with nearly 60% of the remaining natural grasslands classified as threatened (Müller-Nedebock and Chaplot 2015; Bengtsson et al. 2019).Major degradation of these systems is often the result of soil loss from land-use practices that result in the loss of vegetation biomass both above and below ground (Buisson et al. 2021).These South African grasslands, particularly those in high rainfall regions are dominated by resprouter species, reliant on vegetative propagation, and maintained by fire regimes (Klaus et al. 2018;Jaksetic et al. 2018;Zaloumis and Bond 2016).These species are less reliant on soil seed banks for regeneration, even in their natural state (Buisson et al. 2021;Zaloumis and Bond 2016).Topsoil erosion often remove not only seedbanks but also underground rootstocks, making regeneration difficult (Klaus et al. 2018;Jaksetic et al. 2018;Zaloumis and Bond 2016).Restoration becomes very important because species-rich grasslands are considered more stable and resistant to global change drivers (i.e.climate change and atmospheric carbon dioxide level increases) than grasslands dominated by one or a few competitive species (Bengtsson et al. 2019;Tilman et al. 2006;Binder et al. 2018).In addition, high levels of diversity enhance the grassland's ability to provide Introduction Do native grasses emerge and establish in areas rehabilitated using vetiver grass?Lindokuhle X Dlamini 1, 2 * , Michelle J Tedder 1 and Kevin P Kirkman 1 ecosystem services (Binder et al. 2018), through which these grasslands support South Africa's economic activities, and thus their restoration is of utmost importance.Large-scale restoration in natural grasslands, especially those heavily degraded, is not yet feasible, despite their rich biodiversity, provision of ecosystem services, and direct support of human livelihoods (Knight and Overbeck 2021;Bell et al. 2019;Buisson et al. 2021).This is because restoration is often difficult and time-consuming (Zaloumis and Bond 2011).Among the many challenges making restoration difficult, reintroducing native seeds and promoting their recruitment in a degraded environment remains the key challenges (Bakker et al. 1996;James et al. 2011;Shackelford et al. 2013;Bucharova et al. 2017).Another challenge is the endorsement of tree planting programs by the UN decade of restoration (2020-2030 cohort), which often assumes that open systems like grasslands are secondary vegetation systems, arising from forest clearing (Parr et al. 2014;Bond 2016;Bastin et al. 2019;Silveira et al. 2020;Buisson et al. 2021).This endorsement threatens not only already degraded grasslands, but ancient intact grasslands (Silveira et al. 2020;Buisson et al. 2021).For this reason, grassland restoration must first distinguish between secondary grasslands and ancient grasslands and then develop an area-specific restoration program (Buisson et al. 2021;Zaloumis and Bond 2016).In a South African context, Zaloumis and Bond (2016) proposed effective characteristics to distinguish between ancient grasslands and secondary grasslands.Buisson et al. (2021) further outline the problems and knowledge gaps associated with grassland restoration and then proposed practical ecological solutions for restoration.Fynn et al. (2009) outlined grass species traits and characteristics relating to invasion and resistance to invasion in the context of restoration, while Demmer et al. (2019) proposed seed mixtures and seed priming methods to promote the restoration of speciesrich grasslands.These studies provide a foundation for the establishment of effective ecological restoration programs in South Africa.
The limitation outlined by most studies is that restoration programs are often limited by a lack of understanding of the social context (i.e.people's perception and acceptance of restoration practices), which is very important for establishing an effective restoration program (Buisson et al. 2021;Gudyniene et al. 2021).Restoration requires a solid link between science and practice on the ground (Buisson et al. 2021;Gudyniene et al. 2021;Knight and Overbeck 2021).Aligning restoration efforts in a South African context with what farmers and communities on the ground are already employing can offer great benefits and lead to successful restoration programs that are cost-effective and self-sustaining (Knight and Overbeck 2021).This requires understanding the ecosystem services lost due to grassland degradation (e.g.fodder for livestock) and adopting restoration efforts that restore these specific ecosystem services (Buisson et al. 2021;Zaloumis and Bond 2011).Vetiver [Chrysopogon zizanioides (L.) Roberty] is used mostly by farmers and land managers as a cost-effective, natural rehabilitation technique for trapping topsoil and restoring nutrient cycling, particularly in soil-eroded grasslands (Huang et al. 2019;Greenfield 2002).
Vetiver grass is used extensively for land rehabilitation.Restoration programs would benefit if heavily adopted programs were aligned with ecological restoration principles.This can be facilitated by understanding vetiver's role in grass secondary succession stages.These stages of grassland succession often follow a clear pattern from pioneer grasses (often annual grasses with effective seed dispersal strategies) to subclimax grasses (densely tufted species with competitive advantages over annuals) to climax grasses (strong perennial and densely tufted grasses, which vary greatly in their environmental requirements) (Tainton 1981(Tainton , 1999)).The potential for vetiver to alter this successional process is not known, especially in areas where reseeding of native species is not part of the restoration strategy.
Aligning restoration ecological principles and vetiver use requires addressing myths about vetiver grass potential and evaluating whether reseeding with native grass seeds is likely to be successful in areas already partially rehabilitated using vetiver.This would supplement understanding of the mechanisms that allow or inhibit vetiver coexistence with native grasses.Vetiver is a densely rooted and tufted perennial C 4 grass that has been extensively used for rehabilitation worldwide (Greenfield 2002).Irrespective of environmental and soil conditions, vetiver establishes faster than most other grasses, rapidly extending its dense roots down to 4 m deep into the soil, binding soil particles together (Vieritz et al. 2003).For example, Greenfield (2002) showed that vetiver grass could reach rooting depths of 3.6 m after 8 months of growth.The dominant cultivar produces non-viable seed, relying on vegetative reproduction, with its large rootstocks allowing it to survive for decades (Xia 2002).This cultivar was propagated because the original vetiver cultivars were becoming invasive (Council 1993;Truong and Creighton 1994).This cultivar is only planted vegetatively, with a recommended planting distance of 15 to 30 cm apart in rows (i.e.contour lines) that are 6 to 10 m apart, depending on the slope angle to reduce intra-species competition (Truong and Creighton 1994).The assumption is that contour lines of vetiver should trap eroding soils and restore soil function, allowing other species to colonize the 6 to 10 m spaces between vetiver contours.Furthermore, because vetiver is planted vegetatively, chances of invasion are considered to be very low, because only extreme events (e.g.floods) would facilitate secondary dispersal of vegetative rhizome fragments, thus enabling its spread (West et al. 2014;Richardson et al. 2000).Richardson et al. (2000) suggested that for a species to be classified as invasive it must have spread 100 m within 50 years.Other studies have shown that the invasiveness of vegetatively planted species like Miscanthus ogiformis (formerly Miscanthus giganteus) is possible only under extreme events (i.e.floods) in areas where there is no proper monitoring and follow-up control programmes in place (West et al. 2014).These studies suggested that monitoring is important even for species of low invasion potential, such as vetiver grass (West et al. 2014).
Notably, vetiver restores some ecosystem services (e.g.water provision and soil protection), but it does not restore grassland condition (fodder for livestock), ecosystem stability and ecosystem multifunctionality, which are all linked to species coexistence (Soliveres et al. 2016;Meyer et al. 2018).Furthermore, vetiver does not provide an alternative source of forage as it is relatively unpalatable (Council 1993).Its use is supported by anecdotal claims that it acts as a pioneer type species allowing recruitment of other grasses around it, but no attempts have been made to test these claims (Council 1993).Consequently, there are concerns that using vetiver for rehabilitation may result in monospecific, low diversity and low cover swards with limited contribution to ecosystem services.Vetiver alone should therefore not be the endpoint of grassland restoration, but should be the first step toward the restoration of grasslands ecosystem services.There are several steps or stages in the restoration process, all depending on the type of degradation (Buisson et al. 2021).For example, where degradation entails the total or partial loss of nutrient-rich topsoil, the first step would be restoring abiotic conditions (including restoring soil infiltration capacity or reducing surface runoff) (Buisson et al. 2021).Vetiver grass is able to do this very well (Council 1993).However, the second stage would be introducing native species through seeding or vegetatively using plugs (Buisson et al. 2021).Commercially sourced seed mixes are beneficial for this stage because they offer high germination potential compared to naturally sourced seeds (Demmer et al. 2019).The later stages of restoration are maximizing establishment, ecosystem functioning, and resilience, as well as developing an effective monitoring program to ensure there is species coexistence and succession (Demmer et al. 2019;Buisson et al. 2021).Understanding vetiver grass coexistence with native grasses and its contribution to the restoration of speciesrich grasslands is critical for aligning its use with effective grassland restoration principles.
Colonisation by native species is necessary and fundamental for the restoration of species-rich grasslands (Fynn et al. 2009).The ability of any species to allow the succession of other species is dependent on factors such as that species' competitive ability, other species' competitive ability, and seed production (Tedder et al. 2011;Fynn et al. 2009).Competitive dominant species in an ecosystem can provide strong resistance to the recruitment of small subdominant species, resulting in restoration failures (Fynn et al. 2009;Bakker et al. 2003;Flory and Clay 2010;Guido et al. 2019).Numerous studies have shown that established grass plants can either facilitate seed germination and seedling establishment by creating a favourable microclimate through shade and moisture retention (Rees and Brown 1991;Brooker et al. 2008;Tedder et al. 2011;Ren et al. 2008), or can be territorial and inhibit seed germination and seedling establishment through shading, direct root competition and allelopathy (Schenk and Jackson 2002;Guido et al. 2019;Dear et al. 1998).Dear et al. (1998) showed that established Phalaris grass swards (Phalaris aquatica) reduced the emergence and early growth of other grass species through competition for water and light.Species' competitive ability is linked to their physical traits (Nash Suding et al. 2003;Fynn et al. 2009).Fynn et al. (2009) showed that fast-growing broad-leafed species with high total leaf mass, many tillers and low specific leaf area, exert a greater competitive effect and resist invasion by other grasses.Species with such traits are not recommended for the restoration of species-rich grasslands.Considering that vetiver is a wide-leafed, tall, unpalatable grass species with many tillers, it is likely that it will resist invasion by native grasses and inhibit their establishment and growth, hence limiting ecosystem services that multispecies grasslands can provide.
Plants that inhibit the growth of other species in close proximity to them, but do not spread are described as territorial (Dias et al. 2016;Schenk et al. 1999).Territoriality in plants relates to both root spatial segregation and allelopathy (Dias et al. 2016;Schenk et al. 1999).Plants with spatially segregated roots can occupy vacant soil volumes and thus avoid direct root competition with neighbouring plants (Schenk et al. 1999).Allelopathy alters this segregated root behaviour by releasing allelochemicals that inhibit seed germination and establishment of other plants (Ghebrehiwot et al. 2014;Dias et al. 2016).Therefore, allelopathic plants are seen as territorial and invasive.Mao et al. (2004) showed that the oil produced by vetiver roots has allelopathic properties that inhibited weed species germination, emergence and establishment in their trial.Even though their focus was on using this oil as a pesticide (Mao et al. 2004), this allelopathic behaviour could also inhibit native grass emergence and establishment; however, no study has addressed this question.The focus of rangeland rehabilitation projects has shifted from soil protection alone to promoting species diversity to improve a range of ecosystem services (Bakker et al. 2003).Therefore, vetiver research should shift from the application of vetiver in soil and water conservation to questions surrounding its ecology and its impact on biodiversity and a wide range of ecosystem services.These questions include, but are not limited to, its effect on native grass germination, emergence and seedling recruitment.
South African grassland species are not heavily dependent on soil seed banks for their regeneration; therefore, restoration often requires the reintroduction of seeds or vegetative propagation sources to enable colonisation (Zaloumis and Bond 2011;Demmer et al. 2019).Several methods of species reintroduction (by seeds) have been developed to improve recruitment and establishment in various environmental conditions (Demmer et al. 2019;Tamura et al. 2017).Many of these methods are area-specific, often focusing on arid regions (Buisson et al. 2021;Demmer et al. 2019).Hydroseeding is one method that has been shown to produce favourable results especially in improving seed germination, seedling emergence and seedling establishment (Demmer et al. 2019;Buisson et al. 2021;Tamura et al. 2017).Hydroseeding is similar to broadcasting seeds on the surface because seeds mixed with a cellulose-based water-retention gel are laid on the soil surface, not buried (Demmer et al. 2019).Tamura et al. (2017) investigated which methods were best suited for restoring moderate and steep slopes in a Mediterranean climate, finding that hydroseeding on moderate slopes was the most cost-effective method, which resulted in a species-rich relatively stable grassland.However, other studies in arid regions where grasslands often rely on seed banks, suggest that seed burial offers better results, because grass seeds often remain dormant underground for extended periods, germinating when conditions are favourable (Klaus et al. 2018;Jaksetic et al. 2018;Gudyniene et al. 2021;Tamura et al. 2017).Seed germination generally increases with a decrease in seed burial depth (Harris 1996;Aldrete and Mexal 2005;Lodge and Harden 2009).Harris (1996) showed that seeds on the soil surface dry out quickly in areas where rainfall is a limiting factor, hence reducing the germination potential of these seeds.He found that slightly (< 2 cm) burying seeds increases the germination percentage significantly (Harris 1996).Similarly, Lodge and Harden (2009) found that tropical perennial grass species (e.g.Panicum coloratum, Digitaria eriantha and Chloris gayana) seeds sown at a 1 cm depth had greater seedling emergence than surface sown seeds (0 cm), while below 1 cm, emergence decreased with increasing depth.For an integrated vetiver restoration process, an effective native seed sowing method and the effect of vetiver grass on the emergence and establishment potential of native grasses need to be understood.
From a conceptual model by Buisson et al. (2021) outlining a practical restoration pathway, vetiver falls under the site preparation stage where restoration of abiotic conditions is necessary.This is followed by the reintroduction of species from seeds and maximizing species establishment (Buisson et al. 2021), which is what this study aims to investigate.It should be noted that the integration of ecologically sound restoration approaches with established and well-promoted vetiver application programs might help reduce the costs of restoration.This may also shift the narrative away from the notion that tree planting or similar intensive solutions, which shift grassland structure and dynamics, are appropriate restoration methods in grasslands.Due to limited knowledge on the restoration of ancient grasslands, a study that evaluates vetiver coexistence with native grasses becomes an important foundational study focused on shifting the application of vetiver from soil rehabilitation to the restoration of species-rich grassland.Coexistence between grass species is essential for ecosystem stability and resilience (Meyer et al. 2018).As vetiver is used extensively for rangeland rehabilitation it is important to understand its coexistence with native grasses, so its use could potentially be modified to facilitate restoration rather than rehabilitation.
This study had two subdivisions, a field survey and pot trials, with two aims.The first aim was to evaluate field sites rehabilitated using vetiver grass to determine which species grew in close proximity to vetiver grass and the extent of grass succession.Vetiver's growth traits and wide environmental tolerance make it a hardy long-lived grass.Therefore, it was hypothesized that vetiver will show no signs of invasiveness but exhibit territoriality, and thus slow native grass succession.It was expected that (1) few native grasses will recruit in close proximity to vetiver, and (2) vetiver will remain where it was planted, showing no signs of spreading.
The second aim was to determine the effect of vetiver competition on the recruitment of two native grasses using a pot trial with different seed sowing methods.Considering that vetiver is a robust, wide-leafed, tall, unpalatable grass species with dense roots and many tillers, which are traits associated with competitiveness; it was hypothesised that vetiver affects the recruitment of native grasses growing around it and reduces native grass seedling performance.It was expected that (1) vetiver grass tufts will inhibit seedling emergence and establishment of native grasses, (2) seed burial will increase seed germination, seedling emergence and establishment probability, and (3) vetiver will have denser roots than native grasses and thus excluding root interaction between vetiver and native grasses will reduce the competitive effect exerted by vetiver tufts on the establishment of native grass species.

Field survey
The field survey was conducted in the Okhombe Valley, located at the foot of the northern Drakensberg Mountains, Bergville, KwaZulu-Natal (KZN), South Africa.The Okhombe Valley is a communal rangeland that receives a mean annual rainfall of 800-1 000 mm, with about 82% of the rainfall received in summer (October to March) (Mansour et al. 2012).The mean altitude of this area ranges from 1 200 to 2 350 m (Temme 2008).It is a cool region with mean monthly temperatures ranging from 11.5 °C to 16 °C, but temperatures can go below 5 °C in winter (June to July), with frost and snow occurring almost every year (Temme 2008;Mansour et al. 2012).The area has mixed geology, ranging from mudstone and sandstone to amphibolite, basalt and tillite with increasing altitude (Mansour et al. 2012).The mixed geology, rainfall and topography give rise to mostly oxidic, well-drained soils with Griffin, Hutton (Oxisol) and Clovelly (Alfisol) on the slopes, and shallower soils, such as Mispah, on top of the catena (Mansour et al. 2012).
The grazing camps are situated on top of the ridges surrounding the settlements, which are located at the bottom of the valley (Figure 1).This presented a management problem, because livestock walk up to the grazing camps every morning and down to the settlements every night.This movement combined with continuous grazing created paths on the slopes, which then eroded during intense rainfall periods.As a result, these grassland soils became heavily eroded, necessitating the initiation of rehabilitation programmes, commencing in 1992.Vetiver grass was planted to rehabilitate and manage these degraded slopes (catchments), as part of a programme funded by the Department of Environmental Affairs and run by staff from the University of KwaZulu-Natal (UKZN).Vetiver tufts were planted 15 cm apart in rows 6 m apart (forming contour lines) to allow them to trap the soil and for other species to colonise between the rows.One other species, Paspalum notatum, which is an alien grass native to South America and North America was also concurrently established vegetatively in between vetiver rows for soil stabilization (Everson et al. 2007).Everson et al. (2007) reported that P. notatum was planted by community members within the 6 m vetiver rows, often a few months after vetiver.This grass has a different morphology to vetiver, as it is a perennial grass with fibrous, stolon-like, elongated rhizomes, which provides a solid soil cover.This programme trained local people in the basic causes and effects of soil erosion, the importance of rehabilitation, methods to rehabilitate eroded sites (e.g.how to plant vetiver and create stone packs) and formed the Okhombe Monitoring Group.This programme empowered the community and helped them to utilize simple techniques to rehabilitate degraded catchments (see Everson et al. 2007 for more detail).This became a community-driven, self-sustaining program that demonstrated how involving communities in rehabilitation programs could be beneficial for land management.
The area is (and was) a primary grassland, classified as upland moist grassland (Mucina and Rutherford 2006) and collaboration with the University of KwaZulu-Natal, as it is a catchment for the Tugela River, which is the largest river in KZN, and hence provides many ecosystem services.For this work, six sites where vetiver grass has been used to rehabilitate eroded areas were identified (Figure 1).These sites were rehabilitated between 1992 and 2015.

Data collection
The contiguous quadrat method was used for recording species composition and aerial cover data (noting bare soil).All species in each quadrat were identified and their aerial cover estimated.Starting at the edge of the planted vetiver grass row, a 0.5 m × 0.5 m quadrat was placed and then turned over six times to cover a distance of 3 m (the maximum distance between contour rows spaced 6 m apart).This was replicated four times per site.GPS coordinates were also recorded at all six sites (Figure 1).

Pot trials
The pot trials were conducted in a greenhouse at the NM Tainton Arboretum, University of KwaZulu-Natal, Pietermaritzburg, South Africa (29°37′47″ S, 30°24′07″ E).Two native grass species, namely Eragrostis curvula and Megathyrsus maximus (previously known as Panicum maximum) were used in this study.These two species are common in South African grasslands and are frequently used for restoration, but are different in their morphology and ecology.For example, E. curvula is a relatively shadeintolerant densely tufted pioneer grass, producing large numbers of small seeds with a high germination potential, while M. maximus is a shade-tolerant broad-leaved late seral grass, with a creeping rhizome, producing fewer, larger seeds with lower viability (Meredith 1955;Adkins et al. 2011;Fish et al. 2015).These two species can be used to predict the response of other South African species that are useful for restoration.

Experimental design
The study consisted of two trials, namely an (1) emergence and establishment trial, and (2) a root exclusion trial.In order to maximise germination and emergence potential, commercially produced seeds were used (McDonald's Seeds, Mkondeni, Pietermaritzburg, South Africa).

Emergence and establishment trial
Using 6 litre plastic pots filled with commercial potting soil, vetiver tufts of similar size and age were planted one month prior to the sowing of native grass seeds, to allow them to establish.Pots with no vetiver were used as a control and compared with pots containing one vetiver tuft (low competition), and two vetiver tufts planted 15 cm apart (intense competition).The distance of 15 cm apart simulates an accepted vetiver tuft spacing method used in rehabilitation (Dalton et al. 1996).
One hundred seeds of each native grass species were sown per pot.These seeds were laid on the soil surface (unburied), buried (1 cm deep) and spread on the soil surface mixed with a hydro-seeding gel containing moisture retaining polymers.For the hydro-seeding mixture, 2.5 g of fluid drilling gel and 100 seeds of either E. curvula or M. maximus were added to 500 ml of water to make a jelly-like mixture (Pill 1991).The pots were watered every second day.The experiment was a full factorial (2 native species x 3 vetiver competition level x 3 sowing methods), completely randomised design, with six replicates of each treatment combination and 18 combinations in total.
This trial ran for four months (November 2016 to February 2017) and the percentage germinated seeds, seedling survival (%), seedling height (cm), number of leaves per seedling and number of tillers per seedling recorded every second week.

Root exclusion trial
This trial focussed on whether the exclusion of vetiver root interaction would boost the performance of native grass seedlings.As this was a small supporting trial, it did not have a full root interaction treatment (full root interaction vs. none), because it would have given us similar results to the previous trial, and therefore, the control treatment comprised the native species growing alone.Megathyrsus maximus had a low germination and emergence success in the previous trial.It was therefore omitted from this follow-up trial, with E. curvula being the test species.The interaction between the roots of vetiver and E. curvula was prevented by dividing the pot into chambers by pasting 42-micron Nitex mesh [Meshcape Industries (Pty) Ltd, New Germany, South Africa] into the pots to create chambers using a strong waterproof epoxy (Epidermix 372).The Nitex mesh excludes direct physical root competition but allows any allelopathic chemicals, nutrients and water to pass through.
Five three-week-old seedlings of E. curvula were planted alone, with one vetiver tuft, and with two vetiver tufts, with the roots of these species carefully separated into individual chambers within the pot.The experiment had six replicates, laid out in a completely randomised arrangement.Pots were watered every second day and the water that infiltrated through the pots was collected in a drip tray under each pot and used to re-water the pots to ensure that any compounds released by the vetiver were not lost.
The trial lasted 4 months (March to June 2017) and seedling height, number of leaves per seedling and number of tillers per seedling were recorded every second week.The aboveground biomass of the E. curvula seedlings was harvested at the end of the trial, oven-dried for 48 hours at 60 °C, and then weighed.

Root density analysis
Previous studies have found that measuring root density has proven difficult because fine roots are easily washed out and lost during the root extraction process (Cahill 2002).For this reason, a different approach was used to quantify how much soil space was available for other species' roots to colonise using E. curvula and vetiver plants grown alone.This approach mimics the use of root imaging devices (e.g.rhizotron) which measure root density and spatial segregation (Cabal et al. 2021).Four replicates of representative pots containing one vetiver tuft, two vetiver tufts and mature E. curvula growing alone, were selected.The soil and root material were carefully removed from the pot, as a single unit, and cut in half, without disturbing the soil, to form a root profile.A photograph of the root profile was taken and analysed using NIS-Element BR software (Nikon Inc.).The NIS-Element BR software analysed the pixel colour of the roots and the soil and by setting the intensity of the colours as a constant (roots coloured blue and soil coloured red with the intensity of both colours standardised to 80%) the total area occupied by roots and soil was calculated.These data were then used to determine the percentage (%) space occupied by roots and soil space available for other roots to colonise.

Field survey
The quadrat data were used to calculate species richness and species abundance.The residuals for species richness were normally distributed; therefore, the relationship between species richness and distance from vetiver grass was analysed using linear regression in SPSS version 24 (IBM Inc. 2016).
Given the species turnover (spatial turnover) within the dataset, a unimodal method, namely correspondence analysis (CA), was used for the ordination of species composition, assessing which species are associated with vetiver grass in relation to years since rehabilitation and distance from planted vetiver grass rows.Distance from planted vetiver rows and years since rehabilitation were overlaid as environmental variables.The CA was performed using Canoco, with the ordination graph drawn using CanoDraw (Ter Braak and Smilauer 1997).

Pot trials data analysis
Generalized linear models (GLM) in SPSS version 24 (IBM Inc. 2016) were used to analyse the data, because the data violated the assumptions of the analysis of variance (ANOVA) even when transformed (Quinn and Keough 2002).A gamma distribution with a log link function was used to assess the effect of vetiver presence (full competition), sowing method and their interaction on the scale data, such as the mean number of leaves per seedling, mean seedling height (cm) and mean number of tillers per seedling in the emergence and establishment trial.
The count data (i.e.number of emerged seeds) were analysed using a Poisson distribution.To assess the effect of competition and sowing method on seedling survival a GLM with a binomial distribution and a logit link function was used, where survival (number of seedlings that survived) was an 'event' and the number of emerged seeds was a 'trial'.For the survival data, where the generalized linear models were not appropriate, because of a lack of variance (e.g. in M. maximus where only one seed emerged and one seedling survived), a single value was substituted, which made the test more conservative (Kiepiel and Johnson 2014).Values from a linear scale were back-transformed to obtain marginal means, which resulted in asymmetrical standard errors (Kiepiel and Johnson 2014).
For the root exclusion trial, a linear distribution with an identity link function was used to assess the effect of vetiver presence (competition) on the mean seedling height (cm), mean number of leaves per seedling, mean number of tillers per seedling and aboveground biomass (g).Furthermore, a GLM with linear distribution and an identity link function was used to assess whether there was a significant difference in soil space occupied by the roots of E. curvula, and one or two vetiver grass tufts (i.e.root density analysis).For all these analyses, where the model showed significant differences, sequential Šidák was used to adjust for multiple comparisons (α = 0.05).The tables with all these outputs have been added as supplementary material.Furthermore, in this study, likelihood ratio tests of full models are presented.The table with parameter estimates for the beta coefficient, odd ratios, lower and upper confidence intervals for each species, sawing method, vetiver competition and their interactions, for understanding the effect size, are also included in the supplementary material.All graphical representations were performed using SigmaPlot version 14 (Systat Software Inc. 2018).

Field survey
Overall, a total of ten grass species, one Helichrysum species and one sedge were found within 3 m of vetiver planted in rows (Table 1).Aristida bipartita, Sporobolus africanus, Eragrostis plana, Eragrostis curvula and Paspalum notatum were present in all sites, while Hyparrhenia hirta and Cymbopogon caesius were present in most sites except the site that was rehabilitated in 1992 (Table 1).Aristida congesta and Helichrysum species were present in sites rehabilitated in 1992, while Chloris gayana and sedges were found in sites rehabilitated in 2002 and 2011, respectively (Table 1).Surprisingly, there were two sites (2002 and 2011) that had vetiver grass present, where it was not originally planted (Table 1).Species richness significantly increased with increasing distance from planted vetiver grass rows (r 2 = 0.4606; p < 0.001; Figure 2; Table 2).However, an increase in variability in species richness moving away from the planted vetiver grass rows was evident (Figure 2).The relationship is not entirely linear, so it should asymptote at some point, but the spatial scale of data collection (0-3 m) could not show this trend.
The correspondence analysis revealed that distance from planted vetiver explained more variation than years since rehabilitation, in species distribution of these rehabilitated sites (Figure 3).The secondary axis accounted for more variation in species and environment relations than the primary axis (CA2; Figure 2).Distance from vetiver was more strongly associated with the secondary axis than time since rehabilitation (CA2; Figure 2).The main variation shown by the correspondence analysis was along the primary axis and neither time nor distance from planted vetiver explains it (CA1; Figure 2).Sporobolus africanus, Chloris gayana, Paspalum notatum and sedges were dominant, or growing within 0.5 m of vetiver planted in rows (Figure 2).There was also evidence of vetiver recruiting within the same range (Figure 2).However, areas around vetiver rows were dominated by bare soil.Aristida bipartita, Melinis nerviglumis, Cymbopogon caesius, Hyparrhenia hirta, Eragrostis curvula and Eragrostis plana started appearing when moving away from vetiver planted in rows.
Aristida congesta was the only species that dominated areas furthest from vetiver planted in rows (Figure 2).

Effect of vetiver competition
Overall, both native grasses showed improved emergence in the presence of vetiver tufts compared to the control (p < 0.001), but there was no significant difference in percentage emergence between one vetiver tuft and two vetiver tufts (p = 0.602, Figure 3a).Individually, the two native grass species showed a similar trend in percentage emergence, although E. curvula emergence was substantially greater than M. maximus (p < 0.001), with E. curvula having a maximum percentage emergence of 45% and a minimum of 30%, compared to 6% and 1% maximum and minimum emergence by M. maximus, respectively (Figure 3b).

Sowing method
Seeds sown on the soil surface germinated (emerged) better (21.16% ± 3.31, p = 0.003) than those that were buried (12.89% ± 2.21) and those mixed with water retention gel (17.60% ± 3.32), but there was no significant difference between seeds buried and those mixed with a water retention gel (p = 0.976).

Main effects
Eragrostis curvula and M. maximus did not differ significantly in terms of their seedling survival (p = 0.227) and overall, vetiver competition, but not the sowing method (p = 0.772) affected seedling survival (p = 0.004) (Table 3).However, E. curvula and M. maximus did not differ in their response to both vetiver competition (p = 0.103) and the sowing method (p = 0.292) (Table 3).The sowing method also did not affect seedling height (p = 0.238), number of tillers per seedling (p = 0.372) and number of leaves per seedling of both native species (p = 0.843).Furthermore, the interaction between competition and the sowing method did not affect the seedling survival of these species (p = 0.240) (Table 3).

Effect of vetiver tuft competition
Overall, seedlings survived better when growing alone (i.e.control) (78.24% ± 3.68) than when growing with two vetiver tufts (72.65% ± 3.30; p = 0.003), but there was no significant difference in seedling survival between the control and one vetiver tuft (76.23% ± 2.92; p = 0.111) and between one vetiver tuft and two vetiver tufts (p = 0.075).

Growth traits
Eragrostis curvula and M. maximus did not differ in terms of seedling height (p = 0.180) or number of tillers per seedling (p = 0.191), but differed significantly in the number of leaves produced per seedling (p = 0.04; Table 4), with E. curvula producing more leaves per seedling (10.34 ± 0.896) than M. maximus (6.15 ± 0.897).Vetiver competition affected seedling height (p < 0.001), number of tillers per seedling (p < 0.001) and number of leaves per seedling overall (p < 0.001) (Table 4).Full competition exerted by one and two vetiver tufts negatively influenced native grass growth (Figure 4).Overall, native grass seedlings grew taller (Figure 4a), produced more tillers (Figure 4b) and had more leaves per seedling (Figure 4c) when growing alone (control) compared to when growing with either one or two vetiver tufts.However, there was no significant difference in seedling height (Figure 4a), number of tillers (Figure 4b)  , and their successional and ecological status.Two sites were rehabilitated in 2002, while the other sites were rehabilitated in different years over 24 years and number of leaves produced per seedling (Figure 4c) when experiencing competition from one and two vetiver grass tufts.
Eragrostis curvula and M. maximus responded differently to vetiver competition in terms of seedling height (p = 0.007), but not number of tillers per seedling (p = 0.071) and number of leaves per seedling (p = 0.539).Even though M. maximus was significantly taller than E. curvula in the control (p < 0.001), these species did not differ significantly in their seedling height when growing with one (p = 0.972), or two vetiver tufts (p = 0.828) (Figure 5).The sowing method and the interaction between vetiver competition and the sowing method had no effect on any of the variables measured (see Table 3).

Field survey
The most consequential results were the marked increase in grass species richness with an increase in distance away from the planted vetiver, the abundance of bare ground around the planted vetiver and the presence of the recruiting vetiver grass away from where vetiver was planted for rehabilitation.Even though the spatial scale for this study was small (0-3 m) and limited by the 6 m vetiver row spacing, it was adequate to address the objectives of the survey.Native grasses appear to fail to recruit in close proximity to vetiver, with only a few species managing (Aristida bipartita and Chloris gayana, both subclimax) to successfully recruit half a meter to a meter away from planted vetiver.These subclimax species occurred relatively close to the planted vetiver compared to the climax and pioneer species.There was no clear relationship between    successional stages and years since rehabilitation.There was a relatively progressive successional gradient moving away from planted vetiver, i.e. turnover of subclimax to climax species accompanied by increased species richness.This turnover was not consistent across all sites.The surveyed sites were in a communal rangeland with continuous grazing, with rehabilitated areas originally a heavily eroded slope, caused by poor livestock management (Everson et al. 2007;Mansour et al. 2012).The abundance of bare ground around vetiver likely indicates vetiver's negative effect on the recruitments of other species, possibly its competitiveness, which the pot trial investigates further.

Source of variation
The grass community in these sites, therefore, reflects both the effect of selective grazing and overgrazing, and the extent of secondary succession as influenced by the planting of vetiver (Mansour et al. 2012).These rehabilitated rangelands can be classified as unpalatable recovering grassland, as they were dominated by more climax species than subclimax species.Further recovery to a species-rich grassland with palatable species is unlikely in the short term, especially if there is no intervention.Findings from this study are relatively similar to those reported by studies of plant succession in old fields or abandoned agricultural fields (Bonet 2004;Bonet and Pausas 2004).For example, Bonet (2004) in a vegetation change study examining a 60-year-old abandoned agricultural field showed that there was a nonlinear relationship between years since abandonment and plant successional status, and rather there was a clear coexistence of different plant functional groups.Similarly, Zaloumis et al. (2016) investigated 4 to 40-yearold secondary grasslands in South Africa recovering from afforestation with Pinus species and showed similar findings.They showed that there was no successional trend of increasing plant and forb species richness compared to the natural grasslands (Zaloumis and Bond 2016).Interestingly, they also showed no turnover in species composition with most species occurring across all sites differing in their recovery years (4-40 years).This is similar to our findings showing no huge species turnover as you move away from planted vetiver even though species diversity increased slightly.This shows that grassland succession takes longer than the period of the current study.
The ecological and successional status of these plant communities may be related to specific species' responses to defoliation (i.e.grazing) or the decreaser-increaser concept described by Foran et al. (1978) and modified by Tainton et al. (1980).The species dominating in all sites were mostly increaser II (i.e.Aristida bipartita, Eragrostis plana and Eragrostis curvula) and increaser III (i.e.Sporobolus africanus) species and one invasive (i.e.Paspalum notatum) species (Van Oudtshoorn 2012).Increaser II species are subclimax and pioneer grass species that dominate in the overgrazed veld (Foran et al. 1978), while increaser III species are climax grass species that dominate in selectively grazed veld (Tainton et al. 1980).Therefore, the species found may indicate some effect of grazing on the succession status and community structure of these rehabilitating sites.Some other species' reproductive traits may explain the dominance of increasers, for example, generally, most of these increaser species reproduce more readily by seeds, while more palatable decreaser species mainly reproduce vegetatively, thus spreading slower (Van Oudtshoorn 2012).However, Paspalum notatum was planted concurrently with vetiver (Everson et al. 2007).It is a palatable alien grass that increases in abundance when grazed, which may explain its dominance (Van Oudtshoorn 2012).Even this creeping strong competitive grass could not cover the bare ground around vetiver (Van Oudtshoorn 2012), suggesting that either the competitive ability of vetiver and/or allelopathy from oil produced in the roots is inhibiting the recruitment of other grasses near vetiver (Van den Berg et al. 2003).This, therefore, challenges the two claims made about vetiver's ability and coexistence with other grasses.These claims are 1) vetiver does not compete with adjacent crops, and 2) vetiver does not spread because it produces non-viable seeds and has no stolon or rhizomes (Vieritz et al. 2003).For these claims to hold, other grasses should coexist well with vetiver, and there should be no signs of vetiver recruiting outside of the planted areas.Unfortunately, the presence of bare ground around the vetiver plants and the evidence of vetiver recruiting, where it was not planted, strongly challenges the validity of these claims.
This study shows that native grasses and the planted alien grass (Paspalum notatum) did not grow in close proximity to vetiver, supporting the notion that it may develop permanent monotypic patches.This, therefore, supports the first hypothesis that vetiver will show no signs of invasiveness, but exhibit territoriality and thus slow native grasses succession.Even though vetiver was seen recruiting outside where it was planted, this does not qualify as an invasion, because it has not spread 100 m in the last 50 years (Richardson et al. 2000).The recruitment of vetiver elsewhere could be caused by extreme events like floods resulting in water dispersing vegetative roots or fragments of rhizomes (West et al. 2014).This form of dispersal is rare and if proper monitoring and control programmes are in place, it is unlikely that vetiver will be invasive.Furthermore, this study supports the two predictions 1) fewer native grasses will recruit in close proximity to vetiver grass planted in rows, and 2) vetiver will remain where it was planted, showing no signs of spreading, although this should be monitored.

Pot trials
In this section of the study, the effect of established vetiver on seedling emergence, seedling survival, and seedling establishment of native grasses was investigated, with the overall purpose being to test the claim that vetiver acts as a pioneer species enabling recruitment of native grasses and thus facilitating multi-species restoration.In addition, we investigated the basic effect of biotic factors on native grass recruitment.Our findings showed that vetiver tufts facilitated seedling emergence in native grasses.Most studies on seed germination or emergence have focused on abiotic factors e.g.soil moisture, temperature and light (Knipe 1968;Lindig-Cisneros and Zedler 2001;Kolb et al. 2016), with few studies testing biotic factors e.g.competition/facilitation and allelopathy from mature established plants (Aguiar et al. 1992).The ability of vetiver to facilitate seedling emergence can be partly explained using nurse plant theory, even though this phenomenon is often studied in alien invasion studies.In this study, nurse plant theory could explain vetiver facilitation of seed emergence.Nurse plants theory suggests that plants can create a favourable environment through shade and moisture retention, resulting in an increased emergence and seedling establishment around them (Fowler 1986b;Ren et al. 2008).Many well-studied abiotic factors that affect seedling emergence can be altered by established neighbouring plants e.g.available light through shading (Aguiar et al. 1992;Buisson et al. 2021).The alteration can produce either a favourable or an unfavourable environment for seedling emergence.Even though we did not measure these factors (soil moisture, temperature and light), it is fair to assume that vetiver tufts altered these factors, because emergence in native grasses was enhanced.Megathyrsus maximus is a shade-tolerant grass; hence its seeds could be adapted to germinate in relatively shady, moist areas (Fish et al. 2015;Van Oudtshoorn 2016).This might explain why the presence of a neighbouring plant boosted seedling emergence.However, because E. curvula is a pioneer species that produces small abundant seeds with greater viability compared to M. maximus, which is a late seral species producing large fewer seeds with reduced viability (Meredith 1955), it emerged better than M. maximus in all treatments.This also suggests that E. curvula has a wider microclimate tolerance for seedling emergence compared to M. maximus.Furthermore, seed size affects seed germination, with most studies reporting an increase in germination rate with a decrease in seed size, which ultimately affects emergence (Silvertown 1981;Gross 1984;Aldrete and Mexal 2005).Our findings are in line with this notion as the seeds of E. curvula are smaller than those of M. maximus.
The seed sowing method also affects seedling emergence making it an important factor in restoration studies (Demmer et al. 2019).Our study had a known initial number of seeds and showed that surface sowed seeds emerged better than buried seeds or seeds mixed with a water retention gel.These results differ from those reported by Demmer et al. (2019) and Gudyniene et al. (2021); however both these studies show that either method is useful for species reintroduction in grasslands.Several studies have shown that burying seeds reduces percentage emergence by increasing seed mortality (Maun and Lapierre 1986;Harris 1996;Aldrete and Mexal 2005).However, this is dependent on sowing depth, with an increase in sowing depth resulting in a decrease in seed germination rate and thus emergence.For example, Maun and Lapierre (1986) showed that the rate of emergence of germinated seeds and the total emergence of dune species (Elymus canadensis, Cakile edentula and Corispermum hyssopifolium) decreased with an increase in seed burial depth in sandy soils.Microsites that favour seedling emergence are those that do not allow seed desiccation (Fenner 1978).However, our current study does not support the prediction that seed burial will increase seedling and establishment probability.However, the findings of this study are important, especially because restoration can be costly, such findings show that often simple seed sowing approaches can yield needed results (Demmer et al. 2019;Buisson et al. 2021;Zaloumis and Bond 2016).
Even though vetiver facilitated seedling emergence, it also reduced the establishment of those emerged seedlings resulting in slow seedling growth and increased seedling mortality.This supports our hypothesis that vetiver affects the recruitment of native grasses around it and reduces the performance of native grasses growing in its proximity.However, the prediction that vetiver grass tufts will inhibit seedling emergence and establishment of native grasses, does not completely hold.Vetiver only affects the establishment but not the emergence of native grasses.Furthermore, our findings challenge the claim that vetiver acts as a pioneer species promoting the establishment of native grasses.Numerous studies have demonstrated that established surrounding swards have a negative effect on seedling establishment of either colonising species or of individuals of the same species (Fenner 1978;Snaydon and Howe 1986;Fowler 1986a).This is because the seedling stage is critical in plant development and requires enough above and belowground resources, hence seedlings tend to be more sensitive to competition than mature plants.For example, in a field trial, Snaydon and Howe (1986) studied the effect of the shoot, root and full competition exerted by established ryegrass (Lolium perenne) on the establishment of Poa annua, P. trivialis and Festuca rubra.They showed that full competition and root competition affected the seedling establishment of these grasses by reducing the dry weight of seedlings by about sevenfold (Snaydon and Howe 1986).Their findings suggested that seedlings are more severely affected by belowground competition than aboveground competition as shoot competition had little effect on seedling growth (Snaydon and Howe 1986).Vetiver has an extensive investment in belowground material, which may explain why seedling growth was substantially reduced by the presence of vetiver.
This study showed similar results to those of Fowler (1986a), Fenner (1978), Wesson and Wareing (1969) and Dear et al. (1998).Fowler (1986a) used a field trial to investigate the effect of established grasses on seed germination, seedling establishment and survival of other grasses.He showed that established grasses had no negative effect on seed germination, but rather had a slight positive effect, despite substantially reducing the number of tillers of emerged seedlings on Bouteloua rigidiseta (Fowler 1986a).Using artificial swards, Fenner (1978) showed that seedlings of ruderal species could not establish successfully in these swards (Fenner 1978).Wesson and Wareing (1969) focused on the effect of grass swards on buried seeds of weed species e.g.Veronica persica and Sinapis arvensis.They showed that seeds of these plants emerged well within established grass swards, but failed to establish (Wesson and Wareing 1969).Lastly, Dear et al. (1998) showed that a combination of competition for water, soil nitrates and light reduced the early growth of clover seedlings in Phalaris swards (Phalaris aquatica).This provides further evidence that the microclimate requirements for seedling emergence can differ from those of seedling establishment.
Seeds require specific abiotic conditions to trigger germination, which then leads to emergence, while seedlings are influenced by both abiotic and biotic factors with competition for resources being the key determinant of seedling survival (Aguiar et al. 1992).However, the trade-off between emergence and reduction in seedling establishment means that competition is the key driving factor.Furthermore, restoration approaches can use this trade-off by applying disturbances (e.g.fire or herbivory or simulated herbivory-mowing) after emergence to potentially reduce the competitive advance exacted by vetiver on seedling growth (Lenz and Facelli 2005).The competitive effect can be altered by disturbance because disturbance could shift resource allocation from competition to growth and survival.This is because seedlings fail to recruit when they have no fully developed shoots and roots that can help them compete effectively for both aboveground and belowground resources with the surrounding sward (i.e.vetiver) (Fenner 1978).Therefore, reducing the vetiver competitive effect during the critical stage of seedling growth may offer a benefit to restoration success.
The factors responsible for hindering the growth and establishment of native grasses were then investigated in more detail.Vetiver root density was compared to that of E. curvula.It was found that E. curvula had more available soil space around its roots (less dense roots) compared to vetiver, suggesting that there is limited space for the establishment of native grasses around vetiver.Reduced space could result in direct root competition between recruiting seedlings and vetiver roots (Schenk et al. 1999).Numerous studies have shown that when there is enough available soil space, roots tend to avoid direct root competition by foraging in unoccupied soil resource patches (Schenk and Jackson 2002;Bliss et al. 2002;Hutchings and John 2003).However, changes in available soil space force direct root competition between mature grasses and seedlings (Schenk et al. 1999), with seedlings experiencing substantial negative effects because of their sensitivity to competition.
Above and below ground competition interacts simultaneously in a natural ecosystem, but can exert different pressures on recruiting species (Cahill 1999).For this reason, an additional trial that excluded root competition was conducted.This was to determine if the exclusion of root competition reduced the negative effect exerted by vetiver on native grass seedlings.Seedlings still suffered a substantial decrease in growth even though root competition was excluded, which means shade, allelopathy, or both were responsible for reducing native grass seedling growth.Surprisingly, there was also a difference between one and two vetiver tufts, with two tufts exerting a more negative effect than one tuft despite direct root interaction being prevented.This is contrary to what was observed in the trial with no root exclusion (i.e.no difference between one and two tufts).The existence of strong root intraspecific competition between vetiver tufts could explain these findings.Root competition between vetiver tufts could be reducing the interspecific competition exerted by two vetiver tufts to the same level as that of one tuft.Strong intraspecific competition has been suggested to reduce the magnitude of interspecific competition when the density of the competing species increases, allowing coexistence with other species between those two tufts (Wedin and Tilman 1993;Tilman 1994).These findings do not fully support the prediction that vetiver has much denser roots and thus root interaction exclusion will reduce the competitive effect exacted by vetiver tufts on the establishment of Eragrostis curvula.Indeed, vetiver has much denser roots, but root interaction alone does not effectively reduce vetiver's competitive effect on native grasslands.
Vetiver oil has been reported to have allelopathic compounds that negatively affect the seedling establishment of other plants (Mao et al. 2004).For example, Mao et al. (2004) investigated the effect of vetiver oil on seed germination and seedling growth of six weed species.They showed that vetiver oil inhibited seed germination and seedling establishment in five out of six weed species providing evidence for the existence of allelopathic behaviour (Mao et al. 2004).The amount of oil, and thus the level of the allelopathic effect produced by the roots can also be affected by root intraspecific competition because it is closely related to metabolism in roots which is affected by both abiotic and biotic factors (Massardo et al. 2006).Oil production is also influenced by soil resource availability (Adams et al. 2003); hence intraspecific competition could result in a trade-off in resource allocation to either growth or oil production.Therefore, two vetiver tufts can produce a relatively similar amount of oil to that of one vetiver tuft as a result of intraspecific competition between the two vetiver tufts, which might explain why there was no difference in the performance of native grasses between one and two vetivers in the presence of root competition.However, this argument does not completely exclude competition for light (i.e.shading effect) as a contributing factor, it only suggests that allelopathic effects appear to be the major contributing factor to the result found in the root exclusion trial.A trial separating allelopathy interference, root competition and shading effect is needed to understand which factor is responsible for the suppression of growth in native grasses.

Conclusion
The field survey and pot trial both suggest that vetiver develops into permanent monotypic patches which are likely to impede the recruitment and growth of native grasses.Vetiver does not exhibit signs of invasiveness, but rather signs of a territorial grass, which is evidenced by its longevity and the unoccupied bare ground around it.Surface sowing and the presence of vetiver are beneficial only for seedling emergence, but not for seedling establishment, as vetiver reduced the growth and survival of native grasses.The reduction could be caused by one or a combination of these factors: 1) Direct root competition between vetiver and seedlings; 2) vetiver shade, reducing the photosynthetic activity in native grass seedlings; and 3) an allelopathic effect caused by vetiver oil.Results of this study supported our hypothesis, but generate more questions about the mechanism vetiver uses to suppress native grass performance.Even though vetiver might allow the emergence and recruitment of a few seedlings of native grasses, rehabilitation using vetiver is unlikely to allow long-term succession by native grasses.This trade-off between emergence and establishment must be explored further.Currently, it is unlikely that areas rehabilitated using vetiver may passively change to species-rich grasslands that support grazing as an ecosystem resource.Vetiver should only be used under clearly defined goals, in heavily degraded areas where maintaining grass cover is better than restoring species diversity.

Study limitation and future studies
The threat of climate change, alien invasive plants, as well as the recent tree planting threats on ancient grasslands (Bastin et al. 2019), all show the increasing need for a solid grassland restoration approach that will be area-specific and restore the specific ecosystem services (Richardson et al. 2000;Buisson et al. 2021).Complete restoration (restoring grasslands to their original natural state) is close to impossible (Buisson et al. 2021).Therefore, it is important to clearly define the restoration goals, so that the approach helps achieve those goals (Buisson et al. 2021).In this study, the goal was to understand if vetiver application could be improved from mere soil protection to restoration of species-rich grasslands that can at least allow grazing as a primary ecosystem resource.There is no doubt that vetiver is useful for environmental protection and rehabilitation.However, species-rich grasslands would provide other ecosystem services indirectly if restoration were successful (Buisson et al. 2021;Demmer et al. 2019).This study however had a few limitations.Firstly, it only evaluated rehabilitated sites without comparing this to close-by natural grasslands.Even though the comparison was going to strengthen the study, adding a natural site was going to raise other questions and complicate the study.For example, without the baseline grassland condition, arguing that adjacent grassland grazed continuously represents the natural state/benchmark sites and intended restoration outcome was going to be difficult.Therefore, evaluating only the rehabilitation site was sufficient for this study.While pot trials are useful mostly for the manipulation of species responses, these pots have limited space and that could over-emphasize competition between these species.Therefore, further studies should consider conducting a field experiment that aims to validate the results obtained in this study.However, pot trials are best suited for understanding competitive interactions and the mechanisms driving these, allowing the study to eliminate other confounding factors (Kawaletz et al. 2014).This meant that only a selected number of native grasses could be used to limit the complexity that might arise, because of intra-species competition as well as inter-species competition (Kawaletz et al. 2014).From this study, the following are the questions/ areas of research that still need addressing, which will increase the spatial and temporal scale of the current study: • A study that separates the effect of shade, root interaction and allelopathy (possible in a pot trial).This will give a clear indication of whether it is allelopathy, root competition, shade, or all these factors combined that inhibit native grass seedling survival and growth.It would be good also to increase the number of native species, from two to at least four (possibly two broad-leaved and two fine-leaved species) to strengthen the recommendations and conclusions.• A study that evaluates the competitive effect exerted by the broad-leaved and fine-leaved native grasses on vetiver grass under low and high soil nutrient conditions, to understand whether some native grasses can suppress vetiver grass.This is because it could be that competition is symmetrical, rather than asymmetrical.This could be both a field trial and/or a pot trial.• There are limited studies and only anecdotal evidence on vetiver grass's response to disturbances, such as fire, frost and grazing.Future studies, therefore, should also try to address these gaps, by studying vetiver's response to the effect of these disturbance factors.An example of a question that still needs addressing is, can vetiver alter fire behaviour when planted in fire climax grasslands as it stays green even during the dormant season?• Grazing also is dependent upon species palatability and digestibility among other factors; however, it is not fully known whether vetiver is digestible and palatable enough to be eaten by livestock and wild grazers.If it is, then at what growth stage is it most palatable and digestible?Future studies should also try to answer these questions.• Furthermore, studies should also combine all information known about vetiver and try to develop a practical guide that points out ways that will guarantee that vetiver is succeeded/suppressed by native grasses.For example, combining possible disturbances that might be useful to suppress vetiver and allow secondary successions of native grasses.In addition, outline if any specific planting or sowing method can encourage species coexistence.A list of possible species that may be useful for competing directly with vetiver or coexist well with vetiver, possibly broad-leaved species, could also be developed.

Figure 1 :
Figure 1: The Okhombe Valley in KwaZulu-Natal showing the location, the six survey sites and the different forms of land use in the region(Mansour et al. 2012)

Figure 3 :
Figure 3: Main effect of vetiver on the mean percentage emergence (± SE) of Eragrostis curvula and Megathyrsus maximus overall (a), and their effect on each species (b).Different letters represent a significant difference between treatments (p < 0.05).

Figure 4 :Figure 5 :
Figure 4: Effect of vetiver (full competition) on the mean seedling height (± SE) (a), mean number of tillers per seedling (± SE) (b), and mean number of leaves per seedling (± SE) (c) of both Eragrostis curvula and Megathyrsus maximus overall.Different letters represent a significant difference between treatments (p < 0.05).

Figure 6 :
Figure 6: Effect of vetiver with root interaction excluded on the mean aboveground biomass (± SE) (a), mean seedling height (± SE) (b), mean number of tillers per seedling (± SE) (c), and mean number of leaves per seedling (± SE) (d) of Eragrostis curvula.Different letters represent a significant difference between treatments (p < 0.05).

Table 1 :
Species found in areas rehabilitated using vetiver grass in different years

Table 2 :
Analysis of variance for a linear regression of species richness against distance from planted vetiver grass rows (m) in all sites rehabilitated using vetiver grass in the Okhombe Valley, Bergville, KwaZulu-Natal

Table 3 :
The effect of the presence of vetiver (competition), sowing method, and their interaction using a generalized linear model on seedling emergence (Poisson distribution, AIC = 595.83,deviance = 1.788) and seedling survival (%) (binary distribution, events by trials, AIC = 518.46,deviance = 3.278) for two native grass species (i.e.Eragrostis curvula and Megathyrsus maximus).Significant p-values (p < 0.05) are in bold.This trial lasted four months (November

Table 4 :
The effect of the presence of vetiver (competition), sowing method, and their interaction on seedling height (cm), number of leaves per seedling, and number of tillers per seedling for two native grass species (Eragrostis curvula and Megathyrsus maximus) using a generalized linear model with a gamma distribution and a log link function.Significant p-values (p < 0.05) are in bold.This trial lasted four months(November 2016 to February 2017)