Igniting river health? Testing the effectiveness of low intensity burns to improve riparian vegetation quality in modified ecosystems

Introduction

Land clearing has led to widespread ecological and geomorphic degradation of the world’s rivers (Brown et al. 2017; Wohl 2019). In coastal catchments of New South Wales (NSW), over 200 years of intense anthropogenic disturbance since European colonisation has led to the loss of geomorphic and vegetation condition of riparian systems (Richardson et al. 2007; Fryirs et al. 2013; Catford and Jansson 2014). However, since the end of the 20th century, the geomorphic and vegetative condition of rivers in coastal catchments of NSW has been improving, likely due to a mix of changing climate promoting increased plant growth, improved land management and active river rehabilitation (Fryirs et al. 2018; Cohen et al. 2022). In particular, riparian and instream vegetation cover has significantly increased since the 1980s in a phase referred to as ‘regreening’ (Cohen et al. 2022; Zhang and Fryirs 2023).

The reciprocal relationship between vegetation and geomorphic processes drives river recovery (Corenblit et al. 2014; Fryirs et al. 2018). In recovering rivers, vegetative recovery is the noticeable increase in woody vegetation cover in the riparian zone (Mabbott and Fryirs 2022; Zhang and Fryirs 2023). Geomorphic recovery is defined as an improvement in the physical structure and function of a river (Fryirs and Brierley 2016; Fryirs et al. 2018). The vegetation facilitating geomorphic recovery may be comprised of either native or exotic species, or both. However, preliminary research shows that a large proportion of returning vegetation in coastal NSW is dominated by exotic weeds, potentially leading to poor biodiversity outcomes in riparian corridors (O’Donnell et al. 2015, 2016; Duxbury et al. 2024). In some riparian systems, invasive exotic species incursions are so widespread and established that cost and time restraints hinder their effective rehabilitation. There is a need for large-scale, cost-effective techniques for managing the quality of riparian vegetation to improve biodiversity outcomes (McDonald and Williams 2009).

Fire presents a largely untested method for controlling invasive exotic species in riparian zones (Mandle et al. 2011; Bixby et al. 2015; Brancatelli et al. 2024). It has potential to be cost-effective and to work at-scale. Indigenous people in Australia have used fire for millennia to ‘clean up Country’ for cultural and spiritual reasons and to manage natural resources (Yibarbuk et al. 2001; Gott 2005). Low intensity burns are often perceived as the most beneficial for managing Australian native vegetation and are thought to cause mortality of exotic species propagules, and trigger germination of fire-adapted native species (Leavesley 2020). Low intensity burns are characterised by low flame height (<1.5 m), slow rate of spread (<100 m/h), and intensities less than 500 kW/m (Australasian Fire and Emergency Service Authorities Council Limited (AFAC) 2016). Low intensity burning may be used independently or in combination with other weed management techniques, such as herbicide application to target invasive species (Radford et al. 2008; Mabbott and Fryirs 2018). However, some native species and ecosystems do benefit from more intense fires and need hotter burns to cause germination in pyrogenic species (Bradstock and Auld 1995; Le Breton et al. 2020).

Burns conducted at low intensities are sometimes referred to as ‘cool burns’, though this term is poorly defined (Leavesley 2020). Cultural burning is conducted by Indigenous practitioners and is broadly defined as ‘burning practices developed by Aboriginal people to enhance the health of the land and its people’ (Firesticks Alliance Indigenous Corporation 2019). For example, a description of Cultural burning given by a Thungghutti practitioner is, ‘a burn done by Indigenous people aiming to bring back native vegetation and animals and keep them safe’ (Thungghutti Natural Regeneration Team, pers. comm. 2024). Hereafter, we will refer to low intensity burns as those that burn at cool temperatures with slow rates of spread and low flame heights. We will refer to burns conducted by Indigenous practitioners in the context of Indigenous values systems as Cultural burns.

Many ecosystems on recovering rivers are heavily modified and have no historic analogues (Fryirs et al. 2018). These systems have undergone anthropogenic impacts that cause irreversible changes to biotic and abiotic conditions, creating ‘novel’ ecosystems where species exist in combinations and abundances that have not previously occurred (Hobbs et al. 2006). As a result, burn regimes that were applied in perceived ‘natural’ or historic vegetation communities may no longer be appropriate for use in novel vegetation communities (Brooks et al. 2004; Enright et al. 2015). In addition, climate change is causing conditions to change so that riparian zones that had not previously burned are now susceptible to unprecedented bushfires because very dry years are causing the usually wet riparian biomass to dry and create large fuel loads (Pettit and Naiman 2007; Fryirs et al. 2022). For example, nearly 30% of riparian stream length in coastal catchments of NSW burned at high to extreme severity during the 2019–2020 Black Summer bushfires (Fryirs et al. 2022). Low intensity prescribed burning may now be necessary in highly modified riparian ecosystems to manage fuel loads, in addition to being used as a management method in novel riparian systems where environmental weeds are a significant issue.

Research into the use of fire to control woody exotic species in terrestrial shrubland, grassland and forest ecosystems has yielded promising results (Brancatelli et al. 2024). However, results on its effectiveness at specific sites are mixed (Mandle et al. 2011; Brancatelli et al. 2024). Few studies explore prescribed burning in highly modified ecosystems, or in riparian ecosystems (examples include Radford et al. 2008; Hankins 2013; Douglas et al. 2015; Mabbott and Fryirs 2018). While certain riparian systems act as refugia for fire sensitive species, other riparian species are adapted to frequent disturbance and may respond well to low intensity burns (Franklin et al. 2009; Petty and Douglas 2009; Bixby et al. 2015; Douglas et al. 2015; Le Breton et al. 2020). Additionally, Indigenous people in Australia and globally apply fire to certain riparian zones guided by seasonal cues for access, for spiritual reasons, to ‘care for Country’, and other reasons (Hankins 2013; Steffensen 2020; Thungghutti Natural Regeneration Team, pers. comm. 2024).

Existing research in riparian systems in northern Queensland and California has found that fire is effective in reducing target woody weed cover (Radford et al. 2008) and increasing the presence of native species (Hankins 2013). Riparian burning is a recommended management method for reducing the presence of the invasive riparian woody weed rubber vine (Cryptostegia grandiflora) in tropical savannas in eastern Australia (Bebawi and Campbell 2000, 2002; Radford et al. 2008). Radford et al. (2008) found burning riparian zones to control rubber vine did not have any irreversible effects on native riparian vegetation nor lead to ecological degradation. In the western United States of America, Hankins (2013) found that fire positively impacted native species richness in riparian vegetation due to reduced exotic species cover in the short term. They found that repeated burning every 2–5 years was necessary to maintain native species dominance (Hankins 2013).

Three common woody weeds in riparian zones in eastern Australia are lantana (Lantana camara), blackberry (Rubus anglocandicans), and madeira vine (Anredera cordifolia). These species pose significant biodiversity threats, with lantana and blackberry being identified as Weeds of National Significance (Department of Primary Industries (DPI) 1999). We could not find any studies on the use of fire to control blackberry, lantana, or madeira vine within a riparian context, though some studies exist on the use of fire to control them in other terrestrial environments (e.g. Ainsworth and Mahr 2004, 2006; Raizada and Raghubanshi 2010; Yeates and Schooler 2011; Ross et al. 2023). Hankins (2013) found that fire reduced the cover and exposed root crowns of a closely related blackberry species, Himalayan blackberry (Rubus discolor). There are no published studies on the use of fire to control madeira vine. Even results from studies in other terrestrial settings are mixed. Both lantana and blackberry can resprout post-fire, and short-term studies have found that these species recover after fire, or fire may trigger lantana germination (Gentle and Duggin 1997; Ainsworth and Mahr 2004, 2006; Raizada and Raghubanshi 2010). However, studies that track responses of competition between lantana and blackberry and native species after annual fire treatement show promising results (Ainsworth and Mahr 2004, 2006; Osunkoya et al. 2013). Further studies are needed to understand if and how fire can be used as an integrated weed management technique in riparian systems.

Our study was embedded within a broader program conducted by the Thungghutti Local Aboriginal Land Council and North Coast Local Land Services, local government authority aiming to rehabilitate a river reach of cultural significance to the local Thunggutti Traditional Custodians. Aims of this program were to heal part of the landscape that had been harmed by colonial land management practices, to recreate a meeting space that could be visited by local Thungghutti community members, and to provide employment opportunities for young people in the Thungghutti community. The experiment in our study focussed on determining the effectiveness of using a combination of techniques. Namely, treating target woody weeds with herbicide, which is considered business-as-usual in Australia, followed by burning the dried treated vegetation to clear debris, to suppress weeds and trigger the germination of native species to enhance biodiversity value. The aim of applying a Cultural burn was to conduct burning at a relatively low intensity during the winter months. This type of burn was deemed to be culturally suitable for the riparian vegetation community at the study site, which comprised a combination of rainforest-adapted and sclerophyll species. The broader Thungghutti project provided an opportunity to not only undertake an assessment of vegetation responses to the burn, but also to gather foundation data on the characteristics of low intensity and Cultural burns in these systems.

The aims of our study were to (1) quantify low intensity burn temperatures of this riparian system, and (2) assess the effectiveness of the combination of herbicide treatment followed by low intensity burning on native riparian vegetation recovery, and on three key woody weed species: lantana (Lantana camara), blackberry (Rubus anglocandicans) and madeira vine (Anredera cordifolia). We focussed on characterising soil surface and 10-cm depth temperatures before, during and after the burn. We expected that:

Regional setting

This study was conducted on a 175 m long reach of Nulla Nulla Creek near Bellbrook (Wupu Manhatinum), New South Wales (NSW), in the Macleay River Catchment (Thunghutti Burrai) (Fig. 1). It is approximately 50 km west of Kempsey, NSW, and 470 km north of Sydney. The treatment reach is on the traditional lands of the Thungghutti people of the Macleay River Valley. Nulla Nulla Creek is a tributary of the Macleay River, and the junction of the two waterways adjacent to our study site is a place of cultural significance to the local Thungghutti peoples (Thungghutti Natural Regeneration Team, pers. comm. 2024). It was historically a meeting place and is still frequently used as a fishing spot. Prior to European colonisation, the site was maintained by frequent fire, which was applied through mosaic burns approximately annually, after leaving camp to care for country and to encourage the return of favourable flora and fauna. After European colonisation, Cultural burning ceased and has not resumed until recent years. Bushfires affected areas adjacent to the study site in 2001/2002, 2006/2007, 2013/2014, and 2019/2020; however, none directly affected the study reach (NSW DCCEEW SEED Fire History dataset (NSW DCCEEW 2010)). Since colonisation numerous changes to the region have occurred. Industry involving cutting of native red cedar (Toona ciliata) brought European colonisers to the Macleay region in the 1930s (National Parks and Wildlife Service New South Wales (NPWS) et al. 2003). Farming settlements began at the same time, leading to widespread land clearing on floodplains (National Parks and Wildlife Service New South Wales (NPWS) et al. 2003) and to displacement of local Indigenous peoples (National Parks and Wildlife Service New South Wales (NPWS) et al. 2003).

Fig. 1.

Site map showing the location of Bellbrook, NSW Australia, where Nulla Nulla Creek is located. The reach of Nulla Nulla Creek surveyed for this study was originally fitted with eight plots. The test (burned) and control (unburned) areas for this study needed to be retrofitted after a patchy burn. The map shows the post-burn burned and unburned survey areas. The eighth plot and other green areas were excluded from the final data analysis.

The Macleay catchment is in the North Coast Bioregion of NSW. The region has a subtropical climate characterised by hot summers and mild winters. Mean annual temperatures range between 20.1 and 29.5°C. The mean annual rainfall is 962.6 mm (Bureau of Meteorology 2024). Average local temperatures in August 2022, the month the burn took place, ranged between a minimum of 6°C to a maximum of 21°C (BOM 2024). That winter was the wettest on record in the state of NSW, with monthly average August 2022 rainfall of 47 mm. Soils derived from basalt occur to the north of the North Coast Bioregion (SEED, accessed April 2024). Historically, forms of Eastern Riverine Forests and Subtropical Rainforest were likely the most common riparian communities to occur around rivers in the area. Eastern Riverine Forest is dominated by river oak (Casuarina cunninghamiana), bottlebrush (Callistemon viminallis), and black tea-tree (Melaleuca bracteata) on the toe of banks, rainforest-adapted species and blackwood (Acacia melanoxylon) on the mid-bank, and cabbage gum (Eucalyptus amplifolia), forest red gum (Eucalyptus tereticornis) and broad-leaved apple (Angophora subvelutina) on the upper bank (SEED, accessed April 2024). Following colonisation, much of the original vegetation in the region was cleared and replaced with cattle grazing and paddocks largely covered by exotic grasses. Riparian corridors on low lying rivers in the region now generally consist of a mixture of native and exotic species with breaks in connectivity between vegetated patches.

The riparian community on Nulla Nulla Creek is now a highly modified Eastern Riverine Forest consisting of a mixture of native and exotic species, with exotic species being considerably more prominent (NSW SEED, accessed April 2024). The groundcover layer is dominated by the exotic grass broad-leaved paspalum (Paspalum mandiocanum) with scattered patches of native grasses, including blady grass (Imperata cylindrica) and tussock (Poa labillardierei). The midstorey comprises largely of exotic woody weeds with a few individuals of blackwood (Acacia melanoxylon), bottlebrush (Callistemon viminallis), black tea-tree (Melaleuca bracteata) and rainforest-adapted species (Mallotus philippensis, Maclura cochinchinensis). The canopy is mostly absent except for a few riveroak and native rough-leaved elm (Aphananthe philippinensis). Both the canopy and midstorey layer are infested with woody weeds including blackberry, lantana and exotic woody vines, such as madeira vine (Anredera cordifolia), white passionfruit (Passiflora subpeltata), and Dutchman’s pipe (Aristolochia elegans). Nulla Nulla Creek in is moderate geomorphic condition according to the NSW River Styles database (accessed May 2024). The banks are steeped and sloped, likely due to past erosion caused by historic vegetation clearing and subsequent channel widening. The river corridor is now moderately well-vegetated but there are few signs of ongoing erosion. Lateral deposition of cobble stone point bars is present in the river channel, though bench formation is only in the early stages and is in the process of becoming revegetated with canopy species.

Methods

Results

Soil and surface temperatures

Maximum temperatures in burned areas exceeded 399°C at the soil surface and reached 47.1°C at a 10-cm depth (Fig. 2). Average soil temperatures during the burn were 29.1 (s.e. 3.7)°C at 10-cm depth. Sixteen tags recorded a maximum burn temperature of 399°C, 4 tags burned at 260°C, 10 at 149°C, and 10 at 121°C. The remaining tags did not exceed 70°C. Average soil temperature at 10-cm depth in control areas was 12.98 (s.e. 0.15)°C and 10.98 (s.e. 0.004)°C across both burn and control areas over the 24 h before the burn. It took soils approximately 2 days following the burn to return to their baseline temperatures (Fig. 3).

Fig. 2.

Maximum temperatures recorded (a) at the soil surface using painted tags, and (b) at 10- cm depth using buried temperature sensors. Tags with temperature sensitive paints only record maximum surface temperatures and do not reflect temperature changes over time. Tags can only be used once. Soil sensors display the maximum temperature recorded during burning activities.

Fig. 3.

Average with a 95% confidence interval for soil temperature over time at 10-cm depth in burned (B, red line) and unburned (UB, blue line) areas. This data was collected using the buried temperature sensors. Sensors were left in the soil from the 17th August 2022 to the 2nd September 2022 to capture baseline soil temperatures before burning and after burning, respectively. The burn was ignited on 18th August and flames had extinguished by the morning of 19th August, however embers may have continued to burn in some locations.

Groundcover

Groundcover exotic species richness was significantly different between burned and unburned areas (χ² = 16.997, P = 0.0019). Exotic species richness was significantly higher in burned compared to unburned areas at the 3-, 9- and 12-month surveys (Fig. 4). Annual herb cover and richness followed a similar pattern to exotic species richness (Fig. 4). Annual herb richness in burned plots at the 3-month survey was significantly higher than the baseline, 6-, 9- and 12-month surveys (Table 3). Herb richness in unburned areas was significantly higher at the 6-month survey than the 3-, 9- and 12-month surveys (Table 3).

Fig. 4.

(a) Annual herb cover for each of the surveys in burned (B, red) and unburned (UB, blue) areas, and (b) exotics species richness, native species richness, and annual herb species richness for each survey in burned (full line) and unburned (dashed line) areas.

Table 3.Outcomes of generalised linear mixed models for exotic species richness, annual herb species richness, and annual herb cover.

	P-value	z-statistic	Effect size ± s.e.		P-value	z-statistic	Effect size ± s.e.
Exotic species richness
3-month B/UB	<0.01	3.9	1.1 ± 0.3	12-month B/UB	0.01	2.6	0.7 ± 0.3
9-month B/UB	0.006	2.8	0.7 ± 0.3
Annual herb richness
B: 3-month/Baseline	<0.01	7.0	3.6 ± 0.5	UB: 6-month/3-month	0.03	−2.9	−0.6 ± 0.3
B: 3-month/6-month	0.04	2.9	0.5 ± 0.1	UB: 6-month/9-month	0.02	3.0	0.7 ± 0.3
B: 3-month/9-month	0.04	2.8	0.5 ± 0.1	UB: 12-month/9-month	0.009	3.3	−0.9 ± 0.3
B: 3-month/12-month	< 0.01	−4.1	−0.7 ± 0.2
Annual herb cover
3-month B/UB	<0.01	5.8	5.8 ± 1.0	B: Baseline/3-month	<0.01	7.4	6.9 ± 1.0
6-month B/UB	0.006	2.7	2.3 ± 0.8	B: Baseline/6-month	<0.01	6.7	6.3 ± 1.0
9-month B/UB	0.0012	3.2	3.1 ± 1.0	B: Baseline/9-month	<0.01	5.5	5.3 ± 1.0
12-month B/UB	<0.01	3.9	4.1 ± 1.1	B: Baseline/12-month	<0.01	5.2	5.0 ± 1.0
UB: 6-month/Baseline	0.0004	4.1	4.2 ± 1.0	B: 3-month/9-month	0.02	2.9	1.6 ± 0.5
UB: 6-month/3-month	0.008	−3.3	−3.0 ± 0.9	B: 3-month/12-month	0.005	−3.5	−1.9 ± 0.5
UB: 6-month/12-month	0.006	−3.4	−3.1 ± 0.9

P values, z-statistics, and effect sizes are only reported for tests where significance was found. B, burned areas and UB, unburned areas. For example, ‘UB: 6-month/3-month’ indicates a comparison of outcomes between 6- and 3-month surveys in unburned areas.

There were no significant differences in percent cover of exotic groundcover between surveys, however, overall cover was significantly different between burned and unburned areas (χ² = 9.90, P = 0.042; Table 1; Fig. 4), and key species showed significant differences post-treatment. The exotic grass broad-leaved paspalum (Paspalum mandiocanum) decreased in cover after burning and stayed low in cover in burned areas 12 months after burning. Paspalum cover was significantly different in burned and unburned areas (χ² = 12.87, P = 0.012). In burned areas, cover at the baseline survey was significantly higher than the 6-month (z = −3.0, P = 0.03, effect size = −1.1) and 12-month surveys (z = −2.8, P = 0.04; effect size = −1.3)

Annual herb cover was significantly higher in burned compared to unburned areas (χ² = 19.7, P = 0.0006, Fig. 4). Herb cover in burned and unburned areas was significantly different at all times except the baseline survey (Table 3). In the burned areas, the baseline survey was significantly lower in annual herb cover than all other surveys (Table 3). The 3-month survey had significantly higher cover than the 9- and 12-month surveys in burned areas (Table 3). In unburned areas, the cover at the 6-month survey was significantly higher than the baseline, 3- and 12-month surveys (Table 3).

Midstorey

Percentage midstorey exotic cover decreased after herbicide and burning treatments, and was significantly different between burned and unburned areas (χ² = 14.09, P = 0.015; Fig. 5). In burned areas percentage exotic midstorey cover decreased after treatments and was significantly higher at the baseline survey than the 3-month (z = −4.63, P < 0.01, effect size = −0.5) and 6-month (z = −2.99, P < 0.033, effect size = −0.3) surveys. In unburned areas, the percentage exotic midstorey cover was significantly lower at the pre-burn survey, after herbicide treatment, than the baseline survey (z = 2.914, P = 0.042, effect size = 0.2) and 9-month survey (z = 3.43, P = 0.008, effect size = 0.2) (Fig. 5). Cover was significantly higher at the 9-month survey than the 12-month survey (z = −3.29, P = 0.013, effect size = −0.4) in the control areas (Fig. 5).

Fig. 5.

(a) Percentage cover of midstorey exotic species compared to native species cover in burned and unburned areas for each survey, and (b) richness of native and exotic species in burned and unburned areas. B refers to burned areas and UB refers to unburned areas.

Midstorey exotic species richness, decreased after burning and after spraying herbicide in control areas, and stayed at lower levels in treatment areas after burning (Fig. 5). Midstorey exotic richness increased across all treatments after the 3-month survey. Exotic midstorey species richness was significantly different between burned and unburned areas (χ² = 11.9, P = 0.036), and lower at the 3-month survey than the baseline survey in burned areas (z = −3.77, P = 0.0023, effect size = −1.5).

Target woody weeds responses to treatments

Decreases in lantana, blackberry and madeira vine cover were observed, both after herbicide treatment and directly post-burn, and cover remained low until 3-months post-burn (Fig. 6). However, there were no significant differences in the target woody weed cover between burned and unburned areas post-treatments. Midstorey cover returned to pre-burn levels for all three woody weeds by the 12-month survey in burned areas and for madeira vine in control areas. Percentage cover of groundcover seedlings for lantana and madeira vine was very low. No groundcover seedling regrowth of blackberry was recorded (Table A1).

Fig. 6.

Cover for three key woody weeds in the (a) midstorey (plants between 0.5 and 5 m) and (b) groundlayer (plants under 0.5 m) in burned and unburned areas. B refers to burned areas and UB refers to unburned areas.

Overall midstorey shrub cover followed a similar pattern to target woody weed cover and decreased after both burning and herbicide treatment. Midstorey shrub cover was significantly different between burned and unburned areas (χ² = 13.58, P = 0.018; Table 1; Fig. 4) and was lower in unburned areas compared to burned areas at the 6-month survey (z = 2.1, P = 0.03, effect size = 1.0). In burned areas, midstorey shrub cover at the baseline survey was significantly higher than the 3-month (z = −3.8, P = 0.002, effect size = −1.3), 9-month (z = −3.5, P = 0.006, effect size = −1.2), and 12-month (z = −3.8, P = 0.002, effect size = −1.3) surveys. In unburned areas, shrub cover at the baseline survey was significantly higher than the pre-burn (z = 3.5, P = 0.006, effect size = 1.2) and 6-month (z = −4.1, P = 0.0007, effect size = −1.5) surveys (Fig. 2). Groundcover shrub regrowth was also significantly different between burned and unburned areas (χ² = 19.3, P = 0.0007), and was lower in unburned areas than burned areas at the baseline (z = 2.6, P = 0.008, effect size = 16.8) and 3-month surveys (z = 2.0, P = 0.05, effect size = 9.2). In unburned areas, cover at the 6-month survey was significantly higher than the baseline (z = 3.1, P = 0.008, effect size = 12.6) and 3-month survey (z = −2.7, P = 0.05, effect size = −4.8) (Table A2).

As noted previously, canopy cover results are observational only given that low intensity burns do not penetrate the canopy. However, the burn may have affected the base of vines that are invading the canopy. We observed that overall canopy cover decreased after herbicide and fire treatments in the burned area (Fig. 7). Exotic canopy and vine cover did not decrease in the burn or control areas directly after herbicide treatment but did decrease in the burned area after fire treatment. Both decreased in the unburned area at the 3-month survey. Exotic canopy and vine cover returned to pre-treatment levels in burn and exceeded pre-treatment levels in control areas after 12-months.

Fig. 7.

Percentage cover for overall canopy cover, exotic vine canopy cover, and exotic species cover for each survey in burned and unburned areas. B refers to burned areas and UB refers to unburned areas.

Discussion

We found no significant decreases in the cover of lantana, blackberry and madeira vine in plots treated with fire and herbicide, and those treated with herbicide alone. However, we found significant decreases in percentage exotic woody weed cover and richness between the baseline and first post-burn survey at 3-months in burned areas. Therefore, our results support the hypothesis that target woody weed cover is lower in plots treated with fire and herbicide than those only treated with business-as-usual weed control techniques. We also found significant decreases in midstorey shrub cover, of which the three target woody weeds species formed a large component. Both shrub cover and percentage of exotic midstorey species began to increase after the 6-month post-burn survey and by the 12-month survey - there were few differences between the burned and control areas. Our results do not support the hypothesis that native species richness and cover will increase in plots treated with fire more than in those treated with herbicide alone. While midstorey exotic species richness decreased after burning, groundcover exotic species richness increased considerably. Most of this increase was driven by growth of exotic annual species. Few native woody species were observed regenerating after burning.

Our findings suggest that fire and herbicide are more effective at treating woody weeds in a riparian zone than herbicide alone, at least over a relatively short term. Fewer differences between burned and unburned areas were observed after 12 months, suggesting that ongoing management is required to capitalise on the benefit of the initial treatment (Fig. 8). Increases in exotic midstorey cover and richness after the 6-month post-burn survey suggest that ongoing maintenance is needed to curb exotic weed invasion further. Native species richness and cover did not increase after burning, but exotic species richness and cover did. This was driven mainly by an increase in exotic annual herb cover and species richness. Only isolated native species were observed regenerating. Therefore, the risk of creating disturbance and conditions where new invasive species can colonise must be considered when conducting low intensity burns. This suggests that there is a short window within which follow-up treatments are needed to suppress exotic regrowth post-burn. Follow-up treatments are needed within a 3-month window to further control or eliminate invasive woody species. Localised moderate intensity burns may need to be considered to trigger the germination of fire-resistant and pyrogenic native species (Auld and O’Connell 1991). Additionally, a single low intensity burn does not replicate historic, more frequent burn regimes implemented by Indigenous peoples, which remain insufficiently studied in the context of highly modified riparian ecosystems.

Fig. 8.

Target woody weed, Lantana camara, cover (a) before burning and herbicide treatment, (b) immediately after burning (c) 3 months after burning, and (d) 9 months after burning. Source of photos: E. Duxbury.

Low intensity burning is not a set-and-forget technique, but it does provide an important window of opportunity for riparian vegetation management to take place. It also holds the potential to be rolled out over much larger areas than common business-as-usual manual or mechanical removal of exotics. Our findings suggest that low intensity burns using traditional techniques could be readily built into riparian vegetation management plans, particularly in novel riparian ecosystems. However, different indicators and additional herbicide application may need to be considered to establish fire timings in systems affected by climate change and where exotic species are present.

We did not find statistically significant decreases in lantana, blackberry or madeira vine cover after burning and herbicide treatments compared to herbicide treatments alone. However, non-significant decreases in the cover of these species after fire and herbicide treatment were pronounced. Much of the lantana and blackberry at our site was eliminated as a result of burning. Unlike past studies, we did not observe significant blackberry or lantana seedling recruitment post-burn, and the cover of standing vegetation remained low for 12 months after the burn. We found that lantana regrowth was the highest 3 months post-burn but very low overall and dropped below pre-burn levels by 12 months post-burn (Fig. 8). Our burn was conducted at a low temperature, so the disturbance it created may not have been significant enough to facilitate lantana invasion. Young seedlings may have been outcompeted by faster-growing grass and annual species (Laterra et al. 2006). Although we found a short-term decrease in lantana and blackberry cover, further studies are needed to understand the effectiveness of using a combination of herbicide treatment and burning on these two target weed species over the long term.

Previous studies have found that burning may not be effective on its own in controlling blackberry (New South Wales Department of Primary Industries (NSW DPI) 2009). Fire kills herbicide-treated canes, but plants can resprout from crowns and roots (Ainsworth and Mahr 2004, 2006). Ainsworth and Mahr (2006) found inconsistent results when examining blackberry regrowth 2 years after the Victorian Black Summer bushfires. Similarly, lantana can resprout from basal dormant buds during low intensity burns (Swarbrick et al. 1998). However, various studies have shown that moderate to high intensity fires are likely to kill adult lantana plants (Swarbrick et al. 1998) and that annual or biennial burning may be enough to eliminate the species in some vegetation communities such as a sub-tropical or dry eucalypt forest (Osunkoya et al. 2013; Debuse and Lewis 2014). In contrast, Raizada and Raghubanshi (2010) found that smoke may increase lantana seedling germination success and reduce seedling mortality and that the disturbance that fire creates can facilitates lantana invasion (Gentle and Duggin 1997). Osunkoya et al. (2013) found that lantana seedling recruitment may negate any loss of standing vegetation caused by infrequent burning. However, they found in simulations that biennial burning over 4–15 years could control the weed (Osunkoya et al. 2013).

Like exotic shrub species, exotic vine cover decreased to its lowest level after burning in the test areas and after herbicide application in the control areas. Post-treatment exotic vine cover peaked during the 6-month post-burn survey during summer and dropped far below pre-burn levels during the 9- and 12-month post-burn surveys. Differences between these surveys are likely due to seasonal differences in leaf cover (Vivian-Smith et al. 2007). Cover of the target weed madeira vine (Anredera cordifolia) peaked during the 6-month post-burn survey (which was in summer), suggesting this species exploited the disturbance created by burning and then either died off in the subsequent months or grew to heights above 5 m so it was not captured in the midstorey survey. We observed canopy regrowth of madeira vine and other exotic vines that exceeded pre-burn levels by the 12-month post-burn survey, so the latter scenario is more likely. We found no previous research on madeira vine’s response to fire. However, this vine reproduces via aerial tubers that drop from vines and become buried in the propagule bank (Vivian-Smith et al. 2007). Anecdotal reports have observed that these tubers resist fire and exploit disturbances created by fire (Thungghutti Natural Regeneration Team, pers. comm. 2024; Urban Bushland Council WA). Unlike this predicted response, we found higher cover of madeira vine propagule recruitment in unburned compared to burned areas, potentially due to relatively moist conditions in these areas that this species prefers (Vivian-Smith et al. 2007).

Our results do not support the hypotheses that low intensity burning in combination with herbicide treatment of woody weeds will increase native plant species richness and cover. On the contrary, we found higher exotic annual herb cover and species richness after burning. Annual herbs often rapidly colonise disturbed sites and have been found in higher abundances at sites that are burned more frequently (Woinarski et al. 2004). Annual herbs exploit disturbance and are short-lived (Keeley et al. 2005). Percentage of annual herb cover peaked 3 months post-burn in the burned area and decreased in subsequent surveys. It remained higher than the pre-treatment survey for 12 months post-burn. It is possible that the seasonality of the winter burn affected germination of native species, however the pyrogenic species blackwood (Acacia melanoxylon) and blady grass (Imperata cylindrica) can germinate in any season under appropriate conditions (Austraits database accessed Dec 2024). Grass cover decreased slightly in burned areas after the fire treatment. Then, it increased in subsequent surveys, except during the 12-month post-burn survey, which was conducted during a drier winter than the previous year and likely resulted in lower leaf cover from drying and leaf death. It is possible that vegetation succession is occurring, and exotic perennial grasses are succeeding annual herb species as they end their life cycles (Cuneo and Leishman 2014; Thomson et al. 2020). Thomson et al. (2020) found that exotic (and native) annual herbs increased substantially after an unplanned fire with the exotics persisting for 5 years. At our site, we observed decreases in annuals cover after 6 months and the return of perennial grasses, which may inhibit the presence of annual species in the long term (Cuneo and Leishman 2014).

We observed perennial grasses growing in burned areas, potentially outcompeting the regrowth of woody weeds. In native grasslands and savannas, exotic shrubs outcompete native grasses when fire regimes involve infrequent burning (Platt et al. 2015). Frequent fire may preference exotic grasses over exotic shrubs (Silvério et al. 2013). At our site, the cover of the exotic grass species broad-leaved paspalum decreased post-burn and was replaced by a mixture of native and exotic grass species. Amongst these species was a locally significant weed, African pigeon grass (Setaria sphacelate) (New South Wales Local Land Services (LLS) 2022) and Weed of National Significance, red natal grass (Melinis repens) (Department of Primary Industries (DPI) 1999). It is possible that the disturbance caused by the low intensity burn provided the opportunity for these species to establish. Thus, the risk of creating conditions, where weeds that pose greater biodiversity risks than pre-burn species assemblages can establish, must be considered when using fire for vegetation management in highly modified ecosystems (Graham and Taylor 2018).

Very little native species regrowth was observed post-burn aside from ferns that provided the initial fuel load. It was isolated to a few acacia seedlings, a small amount of blady grass (Imperata cylindrica), and a flush of the native annual Sigisbekia orientalis. The acacia seedlings were too small to be identified but were likely the local species blackwood (Acacia melanoxylon) because adult individuals were present at the site. The low overall return of native vegetation could be due to a few factors. Firstly, the river reach had been unfenced and accessible to cattle for many years before it began recovering (landholder, pers. comm.) and there was a much greater quantity of exotic standing vegetation than native standing vegetation, so it is possible that exotic propagules dominate the seedbank. For example, at our site only a few (2–3) individual acacias were observed regenerating in the months after the burn. It is possible that more acacia regeneration was not observed because there were only two parent trees in the burn area, and both were relatively small adult trees, meaning there may have been limited acacia seeds in the propagule bank. Research on riparian seedbanks in the Hunter Valley, which is further south, supports this assumption. There, O’Donnell et al. (2016) found that river reaches in low to moderate geomorphic condition had a high proportion of exotic species propagules in the seedbank, largely dominated by exotic grasses and annuals. Additionally, herbicide residue in the soil may have affected native species germination and seedling establishment (Cornish and Burgin 2005; Fos et al. 2024). In certain greenhouse experiments, glyphosate residue in soils has been found to inhibit seed germination and native seedling establishment, and native species may have been more sensitive to herbicides that exotic species at the study site (Cornish and Burgin 2005; Fos et al. 2024). However, these effects are most prevalent in low absorption soils that were not present at our study site (Cornish and Burgin 2005). Other field-based experiments have found that woody weed cover reduction through herbicide application increases native species presence, particularly when exotic species biomass is removed (Carreiro et al. 2020; Jucker et al. 2020).

The Eastern Riverine Forest is thought to have historically covered the riparian corridor along Nulla Nulla Creek containing certain fire-promoted species (SEED, accessed April 2024). While very little intact cover of this vegetation community exists, components of this community do remain and it is possible that a few fire-promoted species may occur in the seed bank (SEED; Macleay Valley planting guide; Thungghutti Natural Regeneration Team, pers. comm. 2024). Very little Western scientific information is available to advise optimal fire regimes for this vegetation community, however burning at intervals between 7 and 35 years has been suggested (Kenny et al. 2003). Historically, campsite locations at the junction of Nulla Nulla Creek and the Macleay River were burned at up to an annual frequency (Thungghutti Natural Regeneration Team, pers. comm. 2024). However, burn frequency varied depending on weather conditions and the type and amount of vegetation present. Some of the mature woody native vegetation on-site were species associated with the Eastern Riverine Forest. The Cultural burning practice used at this site kept flames away from sensitive species.

Many rainforest communities are considered fire-sensitive and too frequent or hot burns can prevent these communities from regenerating (Kenny et al. 2003; Le Breton et al. 2020). However, there is very little available information about the response to fire for individual rainforest-adapted species present in the Eastern Riverine Forest community. Lilli pilli (Acmena smithii) has some ability to resprout after fire, but its seeds are sensitive to desiccation and unlikely to survive fire (Ashton and Frankenberg 1976). Other species that Eastern Riverine Forest comprises are pyrogenic or fire resistant, including forest red gum (Eucalyptus tereticornis), blady grass (Imperata cylindrica) and blackwood (Acacia melanoxylon), of which the latter two species were present at the study site (Lewis et al. 2012; Cruz et al. 2017; Butler et al. 2021). Additionally, Thungghutti Cultural burning practitioners report eucalyptus and Acacia spp. regenerating after hotter riparian burns (Thungghutti Natural Regeneration Team, pers. comm. 2024).

In our study, the degree of heat transfer to the soil may not have created favourable conditions for the germination of native seeds. Germination of the local species, blackwood (Acacia melanoxylon), is triggered by moderately hot burn temperatures of approximately 80°C (Cruz et al. 2017). Soil temperatures at 10-cm depth reached a maximum of 47.1°C and surface temperatures exceeded 399°C. Therefore, it is possible that seeds in the upper 10 cm of soil achieved temperatures of 80°C in some areas and remained at elevated temperatures for at least 3 h. However, high soil heating at 10-cm depth suggests that the top 2–3 cm would have reached temperatures lethal to seeds (Tangney et al. 2020). Additionally, it is difficult to accurately infer soil temperatures at soil depths above 10 cm because temperatures during fires become decoupled from soil heat transfer except under extreme heating or prolonged heat radiation (Tangneny et al. 2020). Even if germination was triggered by fire, there may not have been suitable temperature or rainfall patterns for young seedlings to survive (Walters et al. 2004), or they may have been outcompeted by fast-growing annuals (Woinarski et al. 2004).

The native species blady grass (Imperata cylindrica) is pyrogenic and relies on fire to spread. Small stands of blady grass were present at the site. This species reproduces through rhizome fragments after fire and by seed (Holm 1977). Seed typically spreads 15 m from mature plants or further (MacDonald 2004). However, despite being a fire-loving species, we did not observe a substantial increase in blady grass cover after burning. This differs from studies outside of Australia where blady grass is a significant invasive species, known to outcompete native species and alter fire regimes (Holzmueller and Jose 2012; Butler et al. 2021). Further investigation into the composition of the seedbank in this region is needed to understand what capacity highly modified sites have for native species regeneration after fire, and what their soil and surface temperature thresholds are for triggering gemination. Further work is also needed on the conditions under which fire may produce mortality of weed and invasive species. To improve the biodiversity conditions at this site, and many other sites like it, it is likely that interventions, such as direct seeding or planting and further weed management are needed in the post-fire window of opportunity (Cuneo and Leishman 2014). Increasing native species presence can allow them to outcompete exotic propagules stored in the seed bank (Ainsworth and Mahr 2006).

Temperatures from our burn were similar to temperatures recorded for similar prescribed burns. Iverson et al. (2004) observed soil surface temperatures of 117°C on average and 381°C maximum during a prescribed burn. Granged et al. (2011) found that soil surface temperatures ranged between 142, 317, and 525°C for low, moderate, and high intensity burns, respectively. We aimed to conduct a low intensity, burn, however, some surface burn temperatures at our site exceeded 399°C, despite being conducted in a wet riparian environment. One soil sensor reached 47°C at 10-cm depth. It is likely that this sensor reached a high temperature because it was buried under a particularly large fuel load of dried lantana canes. Past research in banksia woodland found that mean soil temperatures during burns ranged between 20 and 30°C at 10-cm depth across areas where fine fuel loads had been manipulated to between 8 and 20 t/ha (Tangney et al. 2020). Penman and Towerton (2008) found that during ecological burning in a dry sclerophyll forest most of their sensors at 5-cm depth recorded temperatures below 40°C and only one sensor at 2-cm depth recorded a temperature above 100°C. They noted that the low temperatures recorded may be too low to trigger germination in fire-promoted species. However, further trials with sensors in the upper 5 cm of soil will be needed to determine whether similar temperatures are reached in highly modified riparian systems. Finally, it needs to be noted that fire-promoted species respond to a range of cues, not just heat (Thomas et al. 2003). This study did not investigate soil moisture or smoke-chemical responses to fire, or combinations of cues (e.g. Brown and Van Staden 1997; Thomas et al. 2003).

Conclusions

Our study found that a combination of low intensity burning and herbicide treatment may be effective in reducing woody weed cover in riparian sites in the short term. We have quantified the temperature of a low intensity burn conducted in a moist riparian environment. Creating dry biomass through treated woody weeds was necessary for creating the conditions where a burn could take place due to wet conditions, exacerbated by a La Niña year. The presence of the low-flammability exotic species broad-leaved paspalum further impeded the burn. One advantage of using fire is that it can be applied across a much larger area and more quickly than manual vegetation treatments and can be used in combination with other techniques such as herbicide application. However, it is unlikely that a single treatment with fire and herbicide will be enough to eliminate woody weeds in novel riparian ecosystems in the long-term. Our study found that few native species regenerated after burning, and woody weeds began to return after a 3-month post-burn window. Additionally, burning creates a disturbance that may be exploited by species that pose biodiversity risks. Therefore, integrated long-term interventions combining reduction of weedy biomass with native species restoration initiated within a 3-month window will be essential for any management project utilising fire in highly modified riparian systems. Further interventions could include trialling repeated burns at different temperatures, replicating historic Indigenous burn regimes. Native biodiversity could also be improved by direct seeding, tube stock planting, and follow up burning and herbicide treatments to reduce weedy biomass. Community-led projects utilising local expertise in low intensity burning including Cultural burning techniques hold significant promise as a way to implement integrated rehabilitation plans on-the-ground.