Control of African lovegrass by flupropanate in a flora conservation context

Introduction

Weed invasions have substantial impacts on biodiversity (Pyšek et al. 2012) and are regarded as a significant threat to many threatened Australian plants (Burgman et al. 2007). Effective weed control methods are thus needed to achieve flora conservation objectives. The application of herbicides is one method widely used to manage weed impacts in conservation contexts, and in some situations, programs involving herbicide applications have achieved impressive reductions in weed populations and have occasionally led to the restoration of native flora (Mason and French 2007; Gooden et al. 2009; Kaiser-Bunbury et al. 2015). However, significant adverse impacts of herbicides on off-target species have been recorded (Kaiser-Bunbury et al. 2015), including on threatened flora (Matarczyk et al. 2002). Furthermore, successful control of one weed species has often led to replacement by different weed species, an undesirable biodiversity conservation outcome (Mason and French 2007; Pearson et al. 2016) (noting that such outcomes may meet some objectives of species-led weed management; Timmins and Owen 2001). Objectives for asset- or site-led weed management for biodiversity conservation (Timmins and Owen 2001) include weed control methods that have lesser impacts on co-occurring native species than ongoing weed invasion and, over time, an improvement in metrics measuring native species occurrence in addition to metrics of decreased target weed abundance.

African lovegrass (Eragrostis curvula (Schrad.) Nees), a long-lived tussock grass that occurs naturally in southern Africa, has become a significant weed globally (Roberts et al. 2021) and is considered among the highest impact grass environmental weeds in Australia (Van Klinken and Friedel 2017). African lovegrass is competitively superior to native grasses under a range of environmental conditions and higher African lovegrass biomass is associated with lower richness of co-occurring species (Firn et al. 2010, 2018). African lovegrass has been introduced widely through seed contamination and deliberately as a pasture species and for soil stabilisation but has spread further to become a weed in both agricultural and environmental settings (Firn 2009; Van Klinken and Friedel 2017). As a perennial grass using the C4 photosynthetic pathway, African lovegrass grows actively through the warmer months of the year. African lovegrass has the capacity to resprout strongly from the base of tussocks following fire and other disturbances, and is spread primarily via dispersal of seed by machinery, wind and animals.

The most common method for control of African lovegrass in conservation settings is application of the non-selective herbicide glyphosate, although mechanical removal can be effective for low-density infestations (Firn 2009; Blakely et al. 2022). Preventing spread is also an important African lovegrass management strategy (Firn 2009). African lovegrass control in conservation settings is challenging as there are risks of off-target impacts in the use of broad-spectrum herbicides. As an alternative to glyphosate, flupropanate (sodium 2,2,3,3-tetra-fluoropropionate; marketed under several product names) has been demonstrated to be effective for African lovegrass control in some eastern Australian agricultural and environmental situations (Campbell and Nicol 1998). While flupropanate may have potentially less uniform impact on co-occurring native vegetation than glyphosate (Campbell et al. 2002), off-target impacts on vegetation ranging from tree seedlings to desirable pasture species have been recorded (McLaren et al. 2008; Lusk et al. 2017; Blakely et al. 2022). Flupropanate has a low contact activity and is therefore mainly absorbed into the soil and taken up by plant roots after rain post-application. Flupropanate effectiveness is considered to be dependent on both temperature (acting faster with warmer conditions) and rainfall (requiring wet conditions for activation) (Lusk et al. 2017). Some residual soil effects are typical but may be short-lived (Bourdôt et al. 2017).

The aims of this study were to build on previous research by testing the efficacy of flupropanate to control African lovegrass in a different climatic and land use context, specifically to:

Materials and methods

Study ecosystem and site

The Swan Coastal Plain, that forms part of the Southwest Australian Floristic Region biodiversity hotspot, supports an exceptional concentration of threatened flora, particularly on the relatively older (late Pliocene–middle Pleistocene) Bassendean sands and Guildford sediments (Gosper et al. 2021a, 2022) on the eastern side of the plain. Levels of weed invasion are relatively high on the comparatively fertile Guildford sediments (Gosper et al. 2021b), with African lovegrass particularly problematic on the heavier soils typical of this formation. In a weed prioritisation for the Department of Biodiversity, Conservation and Attractions’ (DBCA) Swan Region, African lovegrass was ranked with high ecological impact and rapid invasiveness (https://www.dbca.wa.gov.au/management/threat-management/weeds). African lovegrass invasion impacts a range of threatened flora on the Swan Coastal Plain, including Grevillea curviloba (Western Australian conservation status of endangered; Fig. 1), Darwinia foetida (endangered), Synaphea sp. Fairbridge Farm (D. Papenfus 696) (critically endangered), Synaphea sp. Pinjarra (R. Davis 6578) (critically endangered) and Synaphea sp. Pinjarra Plain (A. S. George 17182) (endangered), and threatened ecological communities such as Corymbia calophylla – Kingia australis woodlands on heavy soils (critically endangered) and Corymbia calophylla – Xanthorrhoea preissii woodlands and shrublands (endangered).

Fig. 1.

(a) African lovegrass (Eragrostis curvula) invasion impacting a population of the endangered Grevillea curviloba; (b) habitat of roadside G. curviloba plants that were regularly slashed and that were tested for tolerance to flupropanate. Photos: Carl Gosper, Julia Cullity.

An area near Bullsbrook, north-east of Perth in south-western Australia, was chosen for the study, having an extensive African lovegrass infestation adjoining and in similar habitat (soils, remnant vegetation) to key threatened flora in which African lovegrass control is required. The site, in a narrow (less than 100 m wide), linear rail easement vested in the Perth Transport Authority and the City of Swan, had been historically cleared of large trees but retained scattered mature native shrubs over a mixed ground layer of dense African lovegrass, invasive perennial veldt grass (Ehrharta calycina), invasive annual grasses, and native and invasive herbs, geophytes and graminoids.

Results

Application of flupropanate resulted in a significant reduction in African lovegrass cover (Fig. 2), demonstrated by a significant year × treatment interaction. The higher application rate resulted in a greater decline in African lovegrass cover in the first year post-treatment and although a second application further reduced African lovegrass cover at both herbicide rates, the difference between application rates was maintained.

Fig. 2.

Effects of treatment with flupropanate (control – untreated; low rate – 1.5 mL L⁻¹; high rate – 3.0 mL L⁻¹) on cover of live African lovegrass (Eragrostis curvula). ANOVA results (in text box) are based on log₁₀ transformed cover values (the number of point intercepts out of 200 with the species per plot) that were converted to proportional cover for graphical presentation. ***P < 0.001, **P < 0.01, *P < 0.05.

No effects of flupropanate treatment were detected for any of the invasive annual, invasive perennial groundcover, native perennial groundcover or native shrub functional groups (Table 1). For invasive annuals, there was a significant effect of Year, with greater cover in the two post-treatment years. While there were insufficient intercept data and a lack of replication across the sample plots for robust statistical analysis at the species level for native flora, there was a suggestion of an adverse impact to flupropanate in Desmocladus virgatus (Fig. 3). There was both an increase in dead intercepts and a decline in live intercepts in the first year post-treatment in all flupropanate-treated plots where this species was recorded, but this pattern of change in cover was not apparent in the control plot with D. virgatus. Live cover of D. virgatus recovered somewhat following the second year of treatment.

Fig. 3.

Effects of treatment with flupropanate (control – untreated; low rate – 1.5 mL L⁻¹; high rate – 3.0 mL L⁻¹) on cover of Desmocladus virgatus, showing intercepts with live D. virgatus plants (green lines and symbols) and dead D. virgatus plants (cyan lines and symbols). D. virgatus was present in 4 of the 9 plots (two in the herbicide low treatment, with these data showing means ± s.e.) and occurred so infrequently that robust statistical analysis was not possible.

Flupropanate application had a significant effect on plant community composition when all species were included, as evidenced by a significant treatment × year interaction in PERMANOVA (Fig. 4a). In ordination, two of the three control plots remained highly similar in composition across all years, while all flupropanate-treated plots showed a consistent directional shift in location in multivariate space over time associated with lower African lovegrass cover. In contrast, using the same data except for excluding cover of the target weed African lovegrass, there were no significant effects of flupropanate application on community composition (Fig. 4b). Plots did vary in composition (significant site nested in treatment effect) and over time, and these were reflected in the associations of the perennial native shrubs Acacia saligna and Xanthorrhoea sp., and invasive annuals Bromus diandrus and Avena barbata with the location of sites in multivariate space respectively.

Fig. 4.

Effects of treatment with flupropanate ( control – untreated; low rate – 1.5 mL L⁻¹; high rate – 3.0 mL L⁻¹) on community composition based on plant cover (a) including all species; and (b) excluding cover of African lovegrass (Eragrostis curvula). Ordinations are non-metric multidimensional scaling with (a) stress 0.15, vector showing African lovegrass; (b) stress 0.13, vectors showing species with a Pearson’s correlation >0.8. PERMANOVA results (in text boxes) give pseudo-F values for effects of Treatment (Tr), sample Year (Yr), Site nested in Treatment (S(Tr)) and the interactions of Tr × Yr. ***P < 0.001, **P < 0.01, *P < 0.05.

Intentional exposure of three G. curviloba individuals to flupropanate did not result in any mortality over the period of sampling nor any obvious change in plant condition. Similarly, the three control (no herbicide) individuals also survived.

Discussion

African lovegrass was effectively controlled by flupropanate, as has been found previously (Campbell and Nicol 1998), indicating that the strongly Mediterranean climate of the study area did not affect the herbicide’s effectiveness. Similarly to Bourdôt et al. (2017) with Nassella trichotoma, higher levels of weed control were achieved at the higher (3.0 mL L⁻¹) flupropanate application rate. Our results indicate that flupropanate is likely to be effective in controlling African lovegrass in conservation contexts in south-western Australia, specifically where infestations co-occur with the endangered G. curviloba. Deliberate exposure of G. curviloba to flupropanate did not suggest off-target impacts, noting that the sample size of individuals tested was limited.

Flupropanate resistance in African lovegrass has been recorded in NSW (Powells 2022), emphasising the value of an integrated African lovegrass control program, as neither treatment rate in this study resulted in complete local removal. In situations where biodiversity assets are threatened by African lovegrass, we recommend rotation of herbicides of different mechanisms of action and consideration of physical removal treatments to reduce the potential for development of flupropanate herbicide resistance (Norsworthy et al. 2012). The off-target impacts of broad-spectrum herbicides are likely to be less significant, and physical removal more feasible and with less soil disturbance, after an initial flupropanate treatment to dramatically reduce the cover of African lovegrass (Fig. 2). As with all weed management programs, further considerations in managing African lovegrass with flupropanate include the potential for African lovegrass recolonisation following treatment in the context of the existing seed bank, propagule pressure and dispersal pathways, or replacement with other weeds.

As co-occurring flora species were in general either too infrequently encountered or did not occur in most plots, robust statistical analysis of the effects of flupropanate at the species level was not possible, therefore species were aggregated into functional groups and considered at the community level. There were no significant effects of flupropanate application on non-target groups or at the community level (with African lovegrass excluded), noting these findings refer to the effects of incidental exposure of non-target species to herbicide rather than targeted exposure. The greater cover of invasive annuals in post-treatment years is potentially explained by annual weeds increasing in abundance following African lovegrass control, as has been found elsewhere following invasive perennial grass control with flupropanate (Lusk et al. 2017). However, the increased invasive annual cover in post-treatment years was also found in control plots (no year × treatment interaction), suggesting that variable seasonal conditions may be a more parsimonious explanation.

There was, however, an indication of increased mortality in D. virgatus with flupropanate application, although more robust testing is required for confirmation. D. virgatus (Restionaceae) is a rhizomatous, tufted perennial herb and as a low-growing plant would readily have been exposed to herbicide applied to co-occurring African lovegrass. As some off-target effects of flupropanate were likely in this study and have been shown elsewhere (McLaren et al. 2008; Lusk et al. 2017), robust testing of conservation-listed flora’s susceptibility to the herbicide should precede any application in the habitat of these species.

DBCA is further testing the effectiveness of flupropanate for control of other weed species of the Swan Coastal Plain and monitoring for off-target impacts on native flora in intact occurrences of two critically endangered threatened ecological communities, Banksia attenuata and/or Eucalyptus marginata woodlands on the eastern side of the Swan Coastal Plain and Corymbia calophylla – Kingia australis woodlands on heavy soils. The investigation of the trajectory of vegetation composition after weed control by flupropanate in native vegetation in good or better condition will be valuable, as in pasture settings replacement of the target weed by other weeds and bare ground has been the (undesirable) outcome (Lusk et al. 2017). In this study, insufficient time had elapsed after treatment to discern any patterns of vegetation change, with the dead thatch of African lovegrass remaining the dominant cover on treated plots. Additional research to improve the management of African lovegrass using flupropanate could include testing different integrated management approaches (e.g. altering the order and frequency of rotations of different herbicides and physical removal) and, as flupropanate enters the soil, the effects on soil biota (Brace et al. 2025).

Supplementary material

Supplementary material is available online.