Response of reptiles to weed-control and native plant

Charles Darwin University

Response of reptiles to weed-control and native plant restoration in an arid, grass-invaded landscape

Schlesinger, Christine; Kaestli, Mirjam; Christian, Keith; Muldoon, Shane

Published in:Global Ecology and Conservation

DOI:10.1016/j.gecco.2020.e01325

Published: 01/12/2020

Document VersionPublisher's PDF, also known as Version of record

Link to publication

Citation for published version (APA):Schlesinger, C., Kaestli, M., Christian, K., & Muldoon, S. (2020). Response of reptiles to weed-control and nativeplant restoration in an arid, grass-invaded landscape. Global Ecology and Conservation, 24, 1-18. [e01325].https://doi.org/10.1016/j.gecco.2020.e01325

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal

Take down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Download date: 04. May. 2022

https://doi.org/10.1016/j.gecco.2020.e01325

https://researchers.cdu.edu.au/en/publications/fc67357d-14b0-4e89-b85f-416b14558523


Global Ecology and Conservation 24 (2020) e01325

Contents lists available at ScienceDirect

Global Ecology and Conservation

journal homepage: http: / /www.elsevier.com/locate/gecco

Original Research Article

Response of reptiles to weed-control and native plantrestoration in an arid, grass-invaded landscape

Christine A. Schlesinger a, *, Mirjam Kaestli b, Keith A. Christian b,Shane Muldoon c

a Research Institute for the Environment and Livelihoods, Charles Darwin University, Alice Springs, NT, 0870, Australiab Research Institute for the Environment and Livelihoods, Charles Darwin University, Casuarina, NT, 0909, Australiac Alice Springs Desert Park, Department of Tourism, Sport and Culture, Northern Territory Government, Australia

a r t i c l e i n f o

Article history:Received 1 July 2020Received in revised form 8 October 2020Accepted 8 October 2020

Keywords:ReptilesBuffel grassCenchrus ciliarisInvasive plantsDryland restorationWeed management

* Corresponding author. Research Institute for theSprings, NT, 0870, Australia.

E-mail address: [email protected]

https://doi.org/10.1016/j.gecco.2020.e013252351-9894/© 2020 The Author(s). Published by Elselicenses/by-nc-nd/4.0/).

a b s t r a c t

Introduced grasses are a major threat to dryland ecosystems world-wide because of theirability to transform plant communities and change fire regimes. These structural andfunctional shifts are often assumed to impact wildlife but this has rarely been measureddirectly. Likewise, evaluation of weed removal programs rarely considers benefits to fauna,thereby limiting information that could inform management decisions. We used anexperimental approach to test the impacts of removing invasive buffel grass (Cenchrusciliaris), a globally significant invader of dryland systems, on reptiles, a prominentcomponent of the Australian desert fauna. A combination of mechanical and herbicidetreatment was applied to replicate plots in areas that had been invaded for at least twodecades and changes to ground cover and plant and reptile assemblages were monitoredover six years and compared to still-invaded control plots. Following treatment, nativeplants re-established without the need for reseeding or planting, especially during a periodof high rainfall, when positive effects on reptiles also became apparent. The abundance andspecies richness of reptiles increased at all plots during the mesic period, but less so incontrol plots, and remained higher at treated plots thereafter, although this was onlysignificant at some times. Post-treatment 27 of 36 species were captured more frequentlyin treated plots and only four species, all with very low captures, were captured more oftenin invaded control plots. This consistent trend among species suggests negative impacts ofbuffel grass on reptiles are likely caused by broad factors such as reduced prey or habitatdiversity. Together with concurrent research at the same sites, our results provideexperimental evidence that removing buffel grass from heavily invaded areas, even atsmall scales, benefits a variety of native flora and fauna. Until landscape-scale options areavailable, restoration of smaller areas within buffel-invaded landscapes can help to pre-serve native seed banks and adult plants, reduce fire impacts, and provide patches offavourable habitat for fauna. The creation of ‘islands’ of restored native vegetation deservesfurther consideration as an effective intervention that could help to achieve short andlong-term conservation goals in grass-invaded dryland ecosystems.© 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC

BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Environment and Livelihoods, Charles Darwin University, Alice Springs Campus, Grevillia Drive, Alice

(C.A. Schlesinger).

vier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/

http://creativecommons.org/licenses/by-nc-nd/4.0/

mailto:[email protected]

http://crossmark.crossref.org/dialog/?doi=10.1016/j.gecco.2020.e01325&domain=pdf

www.sciencedirect.com/science/journal/23519894

http://www.elsevier.com/locate/gecco





C.A. Schlesinger, M. Kaestli, K.A. Christian et al. Global Ecology and Conservation 24 (2020) e01325

1. Introduction

Invasion by non-native plants is one of the biggest threats to biodiversity world-wide (IUCN 2000). The most harmfulinvaders are those that become dominant, transform the structure and composition of the invaded ecosystem, and alter keyecological processes (Gaertner et al., 2014). These high impact species are often the focus of management programs, especiallywhere serious adverse impacts of the invader are known or predicted, but weed control is often difficult and time intensiveand requires long-term commitment (Quirion et al., 2018). Decisions about management should balance the effort requiredwith likely benefits (Firn et al., 2015), but for most invasive plant species there is little or no direct empirical evidence linkingmanagement effort to biodiversity gain. Evaluation of weed management programs is usually restricted to monitoringchanges in abundance of the weed species and, less often, to the recovery of native plants; the response of animals to weedremoval is rarely considered (Reid et al., 2009). These shortfalls are compounded when there is an absence of pre-invasionbaseline information on ecological communities and a lack of understanding of invader impact. Better integration be-tween plant and animal focussed research is required to maximise the potential for long-term, cost-effective restoration offunctional ecosystems for diverse components of biodiversity (McAlpine et al., 2016).

Introduced grasses are among the most successful and therefore also most damaging plant invaders world-wide, espe-cially in tropical and desert regions (D’Antonio and Vitousek, 1992, Knapp 1996; Williams and Baruch 2000; Godfree et al.,2017). In low rainfall environments invasive grasses can transform understorey vegetation with relatively sparse structureand diverse composition into dense, single species dominated grasslands (Germano et al., 2001; Franks 2002; Clarke et al.,2005). Once established at high density, more frequent and intense fire is promoted with further impacts on native floraincluding long-lived over-storey plants (D’Antonio and Vitousek, 1992, Knapp 1996; Brooks et al., 2004; McDonald andMcPherson 2011; Brooks 2012; McDonald and McPherson 2013). In combination, these processes lead to long term changein plant assemblages and structure (Franklin and Molina-Freaner 2010). These transformations are often easily observable,especially when an invader becomes dominant over a short time, but indirect impacts on wildlife and more complexecological interactions are difficult to detect, and the extent and causes of these impacts are not well understood (Adair andGroves 1998; Grice et al., 2013).

Buffel grass (Cenchrus ciliaris L. ¼ Pennisetum ciliare L.), native to some regions of Africa and Asia, is a significant invaderworld-wide and has become widespread particularly in south-western North America and inland Australia (Friedel et al.,2011). In Australia, buffel grass probably poses the biggest threat to biodiversity of any invasive grass because of itscomparativelymuchwider distribution (Godfree et al., 2017) and ability to establish and become dominant in diverse habitats(van Klinken and Friedel 2017). Over 70% of the Australian mainland is climatically suitable for the establishment of buffelgrass, and this range is predicted to increase further with climate change, especially in southern coastal areas (Lawson et al.,2004; Scott 2014; Martin et al., 2015). The semi-arid and arid inland regions of Australia are one of the last and largestremaining terrestrial wilderness areas of the world and are of global significance for conservation (Watson et al., 2018); yetbuffel grass already dominates many areas within this region and, left unchecked, will continue to substantially transformAustralia’s dryland ecosystems within upcoming decades. Ranked among the most serious environmental weeds in Australia(Grice 2006; Martin et al., 2015), buffel grass is listed as a likely threatening process for threatened flora (e.g. Palm Valley redcabbage palm, Livistona mariae) and fauna from diverse taxonomic groups including reptiles (e.g. Slater’s skink, Liopholisslateri), mammals (e.g. central rock rat Zyzomys pedunculatus), birds (e.g. Carpentarian grasswren Amytornis dorotheae) andinvertebrates (e.g. desert sand skipper Croitana aestiva and several species of land snails) (Northern Territory Government,2020); however, direct evidence of impacts on fauna, in particular, is limited. Similarly, in dryland ecosystems in NorthAmerica, the risks to native flora and fauna from buffel-fuelled fires is considered to exceed the risk from any other non-nativegrass because of the higher and more constant fuel loads buffel grass produces (McDonald and McPherson 2013). However,despite major concerns about transformation of Sonoran desert ecosystems, including the potential loss of the keystonesaguaro cactus (Carnegiea gigantea) (eg. Rodríguez-Rodríguez et al., 2017), there has been little research on the impacts ofinvasion on fauna, except on Sonoran desert tortoises (Gopherus morafkai) (Gray and Steidl 2015).

Despite widespread concern about negative biodiversity impacts of buffel grass, development of legislation to control orlimit spread has been constrained because of its value as pasture (Friedel et al., 2011), although this is changing in somejurisdictions. Notably, buffel grass is now declared aweed in Arizona, where there is widespread distributionwithin protectedareas, and in the state of South Australia, where buffel has invaded relatively recently and increasing detrimental impacts onfire patterns, productive native pastures, and the cultural values of local Aboriginal peoples are anticipated (Read et al., 2020).Resistance to similar legislation continues across the northern states and territories of Australia, where buffel grass is of mostvalue to pastoralism. Consequently, in these regions, investment into managing buffel grass in invaded areas has been un-dertaken mostly at small scales within conservation reserves and on private land. Although the scale and nature of impacts ofintroduced grasses world-wide necessitates investigation into long-term, broad-scale solutions for managing grass-invadedlandscapes, restoring native vegetation in small areas may have significant conservation value in the shorter term (Hulveyet al., 2017). Knowledge of the effort required to establish and maintain restored ‘islands’, and whether benefits extend todiverse taxonomic groups of flora and fauna, is required to inform management. Information gained from monitoringrestoration efforts can also contribute to understanding how ecological systems are being transformed by invasion.

Reptiles are a prominent component of desert faunal communities, especially in Australia, which has the highest diversityof lizards compared to other deserts globally (Morton and Emmott 2014). Because of their small home ranges and terrestrialhabits, reptiles, and especially lizards and small insectivorous snakes, can potentially benefit from small-scale restoration and

2


comprise a useful case for testing the outcomes of weed removal for vertebrate biodiversity. Relatively few studies worldwidehave investigated the impacts of introduced grasses on reptiles, but the extant literature suggests impacts at community,population, behavioural and physiological levels, including reduced richness and abundance (Abom et al., 2015), avoidance ofinvaded areas (Hacking et al., 2014) and reduction in body condition (Gray and Steidl 2015). The magnitude of impact ofintroduced plants on reptiles is predicted to be greatest where habitat structure is dramatically transformed and the invasivespecies is widely distributed (Martin andMurray 2011). Conversion of open diverse structures to dense grasslands potentiallyreduces thermoregulatory opportunities (Garcia and Clusella-Trullas 2019), obstructs movement or visibility thereby alteringpredator prey interactions and intraspecific interactions (Germano et al., 2012) and is also likely to have direct impacts onprey populations.

Premised upon the likelihood that buffel grass invasion is negatively affecting many reptiles in invaded areas in Australiandryland regions, we predicted that restoring native plant structures and communities would be beneficial for reptiles and thatpositive effects would become evident gradually after initial weed control, coinciding with the re-establishment of nativeplant communities and structures. To test these predictions and evaluate potential positive outcomes from weed removalagainst costs, we removed buffel grass from a series of replicated plots and monitored the response of vegetation and cor-responding changes in reptile assemblages in comparison to paired adjacent invaded control plots where buffel grass was leftunmanaged.

2. Methods

2.1. Study location and experimental design

The study was located in the MacDonnell Ranges bioregion in central Australia at two sites; Simpsons Gap in the TjoritjaWest MacDonnell National Park (approximately 12 kmwest of Alice Springs; 23�430S, 133�430E); and the Alice Springs DesertPark (on the western edge of Alice Springs; 23�420S, 133�490E). Both areas are broadly classified as alluvial flats but havedifferent soil and plant communities. The Desert Park site comprised kandosols (alluvial soils) with lowmoisture onmoderateslopes, supporting a sparse open woodland of ironwood (Acacia estrophiolata), bloodwood (Corymbia opaca) and ghost gum(Corymbia aparrerinja), with scattered fork leafed corkwood (Hakea divaricata), witchetty bush (Acacia kempeana), and deadfinish (Acacia tetragonophylla). The Simpsons Gap site was characterised by alluvial sands and silts with moderate to highwater holding capacity, supporting sparse corkwood (H. divaricata) trees, an abundance of bramble wattle (Acacia victoriae)with emu bush (Eremophila longifolia) and colony wattle (Acacia murrayana) prominent in the shrub layer. When wecommenced our study, buffel grass dominated the ground-storey at both sites but was notably taller and denser at SimpsonsGap. Buffel grass at the Desert Park had occasionally been mown to reduce fuel loads, but was otherwise unmanaged, and atSimpsons Gap was completely unmanaged.

The MacDonnell Ranges region has a continental, arid climate featuring large seasonal and diurnal temperature variations(average 4.0 �Cmin to 19.9 �Cmax in July/mid-winter and 21.6 �Cmin to 36.5 �Cmax in January/mid-summer at Alice Springsairport over 79 years; http://www.bom.gov.au). Mean annual rainfall is 283 mm and there is very high variability among andwithin years; rain can occur at any time of the year, while being most likely in summer. Growth of buffel grass and nativeperennial and ephemeral understorey plants is closely tied to rainfall.

The experiment was set up as a randomised block design with three blocks nested in each of the two sites (Simpsons Gapand Desert Park) and two plots, one treated (actively managed to remove buffel grass) and one left as an unmanaged control,within each block (see Young and Schlesinger 2014 for a map). In total there were six treated plots and six adjacent controlplots which represented the surrounding buffel-invadedmatrix. Plots were 50m� 70m and paired plots within a block wereseparated by a buffer zone (invaded by buffel grass) of approximately 20m. Damagewas caused to one block at Desert Park byfire-fighting vehicles in 2011 making it unsuitable for further use; hence, only five replicate blocks were available fromOctober 2011.

2.2. Management of buffel grass at treated plots

Treated plots were mown in February 2008 with a Kubota L5030 tractor with a Howard HD-B Series slasher avoiding treesand shrubs - including detectable seedlings, large logs and fallen trees. Mowing was undertaken to reduce dead standingmaterial which, following rain, would promote new growth and improve the uptake of herbicide.

Herbicide treatment commenced in late 2008 after the first substantial rain event and was continued intermittently, whenneeded, for five years. Glycophosphate herbicide was applied to individual buffel tussocks using backpack spray kits, whileavoiding regenerating native plants. Spraying was implemented by officers from the Northern Territory Parks and Wildlifeand the Alice Springs Desert Park, and decisions about when to spray were based on their prior experience managing weedsand work schedules. Time (person hours) spent onmanagement and the quantities of diluted chemical used (as dilution ratesvaried among brands) were recorded.

3

http://www.bom.gov.au


2.3. Monitoring changes in understorey plants and ground cover

Buffel grass, native plant, and other ground cover within plots was measured periodically, to coincide with fauna surveys.The exception was February 2008, just after the mowing treatment, when un-mown control plots at Simpsons Gap were notre-measured becausewe assumed there had been negligible change in the threemonths since the previous survey. Cover wasmeasured along three permanently located 50 m transects within each plot using the line-intersect method (Mueller-Dombois and Ellenberg 1974). Ground cover was categorised as; buffel grass, other grass (Family Poaceae), chenopod(Family Chenopodiaceae), all other native plants, grass litter, leaf litter, logs and >5 cm branches lying on the ground, grassstubble (where only the root and base of a grass remained with no foliage), ant pavement or other bare ground. Plant cover atthe level of species or genus (where species were not distinguishable in the field) was recorded from the fourth samplingperiod onwards, once native plants started to substantially recover, to enablemore detailed comparisons of plant compositionacross treatments. These data for plant taxa were pooled into the defined categories for the main analyses. Other than buffelgrass, all plants recorded were native to the region. Along the length of each 50 m transect the ground cover type wasrecorded to the nearest cm (using projected cover for plants) and converted to percent cover. The average percent cover ofeach type across the three transects was used as the cover estimate for each plot.

2.4. Monitoring the response of ground active reptiles

Within each plot (n¼ 6 treated and 6 control), we installed 14 pit traps arranged along two parallel, permanent trap lines.Each trap line comprised 50 m of aluminium fly-wire drift fence (approximately 50 cm high) and seven 20 l plastic bucketpitfall traps, one at either end of the fence, and the others spaced at approximately 8 m intervals in-between. Shelter fortrapped animals was provided inside the trap (thick curved pieces of bark) and outside (by leaning bucket lids against thefence over the bucket on the side corresponding to the dominant sun direction). When not in use pitfall traps were sealed.Drift fences were left in place throughout the study, but small animals were able to cross from one side of the fence to theother when the traps were not in use via gaps in the fence above each bucket (approx. 20 cm high and the width of the bucketlid).

Traps were operated seven times; November 2007 (pre-treatment), February 2008 (immediately after mowing) and thenfive times after herbicide treatment had commenced, approximately annually, between January 2009 and November 2012(n ¼ 6 treated and 6 control plots for five sampling periods; and n ¼ 5 treated and 5 control plots for the final two periods).Trapping was always undertaken during the warmer months (SeptembereMarch) when temperatures are predictably highand reptiles were most likely to be active. Traps at all sites were opened at the same time for four consecutive days and nights(representing 56 trap-nights of effort per plot). Captured reptiles were identified to species and lizards were temporarilymarked with white paint so that individuals could be identified if they were re-captured within the same period. All reptileswere released close to the point of capture within 12 h.

Captures across the two trap-lines within each plot and across the four nights were pooled to derive a single number ofcaptures for each reptile species at each plot for each sampling period. For lizards, known recaptures within a trappingepisode were excluded, but it is possible that some individual snakes were counted more than once as they were not marked.

2.5. Analyses

2.5.1. Ground coverPRIMER-e v7 Ltd and PERMANOVAþ (Quest Research Limited, New Zealand) were used for all uni- and multivariate

analyses. An advantage of using semiparametric PERMANOVA also for univariate dependent variables is that p values aregenerated by permutations with no assumptions of data distributions (Anderson et al., 2008). Bray-Curtis similarity matriceswere generated on square root transformed data. The dependent variables used in ground cover analyses were (a) buffel grasscover, (b) native ground cover (the summed cover of native grasses, chenopods and other native plants) and (c) the multi-variate measure relative ground cover (i.e. the proportional cover of soil surface incorporating all plant and other covercategories listed in section 2.3). We used permutational multivariate analysis of variance (PERMANOVA) with type III sums ofsquares to test for significant effects of management on ground cover. Fixed factors were time, treatment and site, and therandom factor block was nested in site. Principal coordinate ordination (PCO) was used to visualise the relatedness of theground cover compositions between the samples. Pearson correlation vectors were added for those cover variables moststrongly associated with the first two PCO axes.

Species richness comparisons were made on plant taxa (for sampling rounds 4 to 7) and we tested for differences incomposition with PERMANOVA as described above. We then used the similarity percentages procedure (SIMPER; Clarke1993) on the square root transformed plant taxa abundances to identify the main plant taxa responsible for the observedsignificant dissimilarity between treatments.

4


2.5.2. ReptilesWe tested for effects of treatment and time on the composition of the reptile assemblages (multivariate dependent

variable incorporating individual abundances of all species) as well as on the univariate dependent variables total abundanceand species richness. For each of these, the same procedure was followed as for ground cover i.e. Bray Curtis similarities weregenerated on square root transformed data, prior to PERMANOVA and PCO. Fixed factors were time, treatment and site, andthe random factor block was nested in site.

A permutational distance-based linear model e an extension to distance-based redundancy analysis (dbRDA) - wasconducted to determine how well ground cover variables (square root transformed and normalized) could explain themultivariate reptile assemblage (Anderson et al., 2008). A stepwise selection procedure was used to determine the best fittingmultiple regression model using the Akaike Information Criterion c (AICc) as selection criterion.

We also tested for effects of temperature and rainfall on lizard assemblages, again using a distance-based linear model andthe same procedure. Variables considered in the model were average maximum and minimum temperatures for trappingdays, and antecedent rainfall over 30, 60, 90, 180 and 270 days. Variables were tested for multicollinearity and averageminimum temperature was excluded as it showed strong correlation with maximum temperature. Rainfall after 60 and 180days were also excluded due to collinearity, and because they each explained <3% of the reptile assemblage in marginal tests.

3. Results

3.1. Effort expended and timing of management in relation to rainfall and site

Except for initial mowing, the timing of management depended on rainfall. By far the greatest effort required (in time andvolume of herbicide) was in the first summer, following the first substantial rainfall after mowing. Much less effort wasrequired thereafter, even though one of the strongest La Ni~na events on record caused rainfall in the second and thirdsummers to bemuch higher than average, and triggered substantial regrowth of buffel grass at treated sites (Table 1: Fig.1 a&b).

Management effort also varied substantially between sites. At Simpsons Gapmowing took longer (~8 h/ha) because densershrubs and sandier soil made it more difficult to manoeuvre (SG~ 8 h/ha; DP ~ 5 h/ha) and the volume of herbicide used wasmore than double because buffel grass tended to re-establish more quickly and tussocks grew larger at this site (SG: 175 h,2249 L; DP: 77 h, 863 L, per hectare over 5 years; Table 1).

3.2. Effect of the treatment on buffel grass and native plant cover

Slashing in February 2008 temporarily reduced standing cover of buffel grass at treated plots. Re-growth was vigorousfollowing the first major rainfall in summer 2008/2009 but, after herbicide treatment commenced, buffel grass cover wasmaintained at a low level (Fig. 1 c). There were strong effects of treatment (Pseudo-F1,22 ¼ 70.6, P¼ 0.002) and the interactionbetween treatment and time (Pseudo-F6,22 ¼ 13.0, P ¼ 0.001) on buffel grass cover reflecting the reduction in cover over timeat treated plots. At Desert Park buffel grass cover was initially lower (~25%) compared to Simpsons Gap (~40%) but increasedover time at invaded control plots, especially coinciding with high rainfall in 2010 and 2011, peaking at around 45% by the lastsampling period. At Simpsons Gap buffel grass cover was consistently high at control plots (about 40% and up to 60%) exceptwhen it was temporarily reduced following two fires that affected this site in 2011 (Fig. 1 c). Buffel grass cover recovered closeto pre-fire levels within a year. Native plant cover at treated plots was also reduced by the fire, but not to the same extent, asthe intensity and extent of burn in these plots was reduced (for details of how the fires affected the experimental plots seeSchlesinger et al., 2013).

High rainfall in summer 2008/2009 triggered modest germination of native grasses and forbs at all plots. Native groundcover then increased substantially at plots where buffel grass was treated, coinciding with mesic conditions and peaking in2010 (at Simpsons Gap) and 2010/2011 (Desert Park). No corresponding major growth-pulse of native plants was evident atcontrol plots (Fig. 1 d). Reflecting this, PERMANOVA on native plants showed strong effects of time (Pseudo-F6,22 ¼ 50.6,P ¼ 0.001) and the interaction between time and treatment (Pseudo-F6,22 ¼ 9.1, p ¼ 0.001). Pairwise tests of the interactionterm showed no differences in the relative abundance of native plants between control and treated plots for the first three

Table 1Quantity of diluted herbicide used to treat buffel grass per year (JulyeJune) in relation to rainfall at Desert Park (DP) and Simpsons Gap (SG). Average annualrainfall for Alice Springs is 282.8 mm

Year Rainfall (mm) Herbicide (l/ha)

DP SG DP SG Mean

2008/2009 228 311 354 1071 7122009/2010 554 450 161 377 2692010/2011 725 774 184 318 2512011/2012 217 275 163 243 21120012/2013 147 185 0 239 144

5

Fig. 1. Timelines showing (a) monthly rainfall (Alice Springs airport); (b) time spent on management (Simpsons Gap e grey, Desert Park-black); and changes inmean (±se) cover of (c) buffel grass and (d) other ground-storey plants. Plant cover (circles at Simpsons Gap and triangles at Desert Park; control plots black andtreated plots grey) was measured at times coinciding with fauna trapping. In (c) and (d) n ¼ 3 for each site/treatment/time, except n ¼ 2 at Desert Park un-managed and managed plots for the last two sampling times (Oct 2011 and Nov 2012).


sampling rounds (pre and early treatment) followed by significant differences from sampling round four, coinciding with therainfall triggered growth pulse, which persisted up to round seven (P < 0.05 for each).

3.3. Effect of treatment on the composition of the ground cover

Ground cover composition changed through time and in response to treatment and there was a strong interaction be-tween treatment and time (Table 2). The impact of treatment also differed between sites and there was a significant inter-action between site and time. Within sites ground cover and composition at unmanaged control plots and pre/early-treatment plots was similar as indicated by their clustering (Fig. 2 a), but distinct from treated plots which separated outalong the first PCO axis. This axis explained 36.5% of the ground cover variance. Control and pre-treatment plots were,

6

Table 2PERMANOVA results on the effects of treatment, site and time on groundcover with block as random factor nested in site. All permutations were above 970except for site (60 permutations only). ECV is the estimated square root of components of variation which estimates the effect size as % species dissimilarity.The residual unexplained ECV was 8.0. Results for distance-based tests for homogeneity of multivariate dispersions (PermDISP) are shown for the mainfactors. *P < 0.05; **P < 0.01; ***P ¼ 0.001.

Factor Pseudo-F (df) P value ECV PermDISP P value

Treatment 59.0 (1) 0.006** 12.1 0.006**Time 25.5 (6) 0.001*** 11.9 0.61Site 8.7 (1) 0.058 13.3 0.003**Block (random factor) 14.2 (4) 0.001*** 8.0 0.007**Time x Treatment 11.4 (6) 0.001*** 10.9Time x Site 7.9 (6) 0.001*** 9.0Treatment x Site 7.6 (1) 0.027* 5.8


predictably, correlated with high buffel grass cover (Pearson correlation > 0.8; Fig. 2 a). Sites tended to separate along thesecond PCO axis which explained 34.1% of the variance.

Distances between centroids through time by treatment and site (Fig. 2 b) show that pre-treatment and after slashing(times 1&2) ground cover at each sitewas distinct while treated and control plots within each sitewere similar. Ground covercomposition in treated and control plots began to diverge from time three at Simpsons Gap and time five at Desert Park,reflecting regeneration of the native plants at treated plots. Changes in composition were more pronounced in the treatedplots compared to the controls which showed relatively little change across times three to seven along the first PCO axis. Bythe end of the study (times 6& 7), the ground cover at the control plots, which had still been distinct across sites in 2007, hadconverged. In contrast, the composition of ground cover in managed plots remained distinct across the two sites.

The richness of the plant taxa, measured from February 2010 (sample 4) onward, was significantly higher at treated plotswith average 10.5 taxa (sd 3.3) compared to 6.0 taxa (sd 3.1) at control plots (Pseudo-F1,88 ¼ 27, P ¼ 0.005). A total of 49 planttaxa were detected in control plots compared to 53 species in the treated plots showing that diverse plants were still presentin invaded plots albeit at much lower abundance. There was an average 72% dissimilarity in plant taxa between the treatedand untreated plots. Apart from more buffel grass in the control plots, the main taxa contributing to this dissimilarity wereSalsola kali, Enneapogan polyphyllus and Calocephalus platycephalus, which were all more abundant in the treated plots.

3.4. Effect of treatment on the abundance and species richness of reptiles

Over five years we captured 30 lizard and seven snake species and 1125 individual reptiles (excluding lizard recaptures).Lizards were biased toward smaller, predominantly terrestrial species of Gekkonidae, Scincidae, and Pygopodidae, with someprimarily arboreal species from these families also captured. The most common species were the beaked gecko Rhynchoeduraornata (290 captures) and the fat tailed gecko Diplodactylus conspicillatus (273 captures). Agamidae and Varanidae werecaptured only occasionally. The unbanded shovel nosed snake Brachyurophis incinctus (Elapidae) and Centralian blindsnakeRamphotyphlops centralis (Typhlopidae) made up the majority of snake captures with just a few individuals of other largerspecies captured.

Time, treatment and the interaction between treatment and time were significant factors in explaining reptile totalabundance and species richness (Tables 3 and 4). Therewere no differences in the average total abundance or species richnessof reptiles between treated and control plots pre-treatment (time 1), or after mechanical reduction in grass cover (pre-treatment time 2) or just after herbicide treatment began (time 3) but differences began to become apparent thereafter,coinciding with high rainfall and regeneration of native plant structures at treated plots (Fig. 3). The highest abundances andnumber of species were recorded in February 2010, coinciding with the beginning of the 2010 La Ni~na event. The mesicconditions continued beyond the next sampling period, but reptile abundance had dropped off by late 2010, especially atinvaded control plots, and remained low in the drier period that followed. The trend for decreasing abundance and richness ofreptiles at control plots relative to treated plots from February 2010 persisted through subsequent trapping events althoughtotal abundance only differed significantly in the fifth (late 2010) and sixth (late 2011) trapping interval (pairwise tests:P ¼ 0.009 and P ¼ 0.045 respectively), and differences in species richness were only significant in late 2010 (P ¼ 0.037), withsome weak evidence of difference for late 2011 (P ¼ 0.051).

3.5. Effect of treatment on the composition of reptile assemblages

We did not detect a significant interaction between treatment and time on reptile composition. Time and site, and in-teractions between these factors, best explained the composition of reptile assemblages (Table 5), as visualised in Fig. 4 a byclustering of pre-treatment samples (times 1& 2) toward one end of the first PCO axis and general separation between DesertPark and Simpsons Gap samples. Reptile composition was also significantly affected by block, reflecting the high spatialvariability of assemblages, even at small scales. Overall, the factors used in the PERMANOVA did not explain reptilecomposition as well as they explained plant cover composition; 34% of the residual dissimilarity in reptiles could not beexplained by our statistical model compared to 8% residual dissimilarity for plants.

7

Fig. 2. Principal component ordination of ground cover (square root transformed) at Desert Park (DP) and Simpsons Gap (SG) with first and second samplingtimes shown as pre-treatment; (a) with vectors indicating the strength and direction of cover categories most correlated with the PCO axes (Pearson correla-tion > 0.8) and; (b) representing the distance between centroids of ground cover composition between treated and control plots at each site through time withnumbers (1e7) corresponding to times when sampling occurred from November 2007eNovember 2012. Triangles mark the DP and circles the SG site. Opensymbols are treated and filled symbols control plots. Grey stars mark pre-treatment samples (t1-2) at SG and grey crosses pre-treatment samples at DP.


8

Table 3PERMANOVA results on the effects of treatment, site and time on the total abundance of reptiles with block as random factor nested in site. All permutationswere above 970 except for location (60 permutations only). ECV is the estimated square root of components of variation which estimates the effect size as %species dissimilarity. The residual unexplained ECV was 19.9. Results for distance-based tests for homogeneity of multivariate dispersions (Perm DISP) areshown for the main factors. *P < 0.05; **P < 0.01; ***P ¼ 0.001.


Treatment 15.3 (1) 0.006** 13.6 0.003**Time 8.3 (6) 0.001*** 17.7 0.034*Site 2.9 (1) 0.151 5.9 0.816Block (random factor) 1.8 (4) 0.098 5.1 0.517Time x Treatment 2.9 (6) 0.018* 11.8Time x Site 1.2 (6) 0.053 9.0Treatment x Site 1.3 (1) 0.301 2.9


The plot of distances between centroids shows the interaction between the main factors in our analytical model withrespect to reptile composition (Fig. 4b). Differences between sites are evident and were maintained through the study butthere was no conclusive trend with respect to increasing or decreasing similarity between treated and control plots within asite.

The absence of compositional changes in reptile assemblages between treated and control plots, despite significant dif-ferences in total abundance and species richness in some post-treatment trapping periods, suggests possible broad impactsacross species. To check for such trends, we pooled data across all treated and control sites for pre-treated (times 1 & 2) andpost-treated (times 3& 7) periods and used a nonparametric paired sign-test to assess the null hypothesis that the proportionof positive signs (more captures) and negative signs (fewer captures) in treated sites compared to untreated sites was 0.5,with each species representing a sample, paired across treatments. Consistent trends among species were apparent post-treatment (Fig. 5). Pre-treatment, 13 species were captured more often at plots allocated to the treated group, and 10 werecaptured more often in the control group with three species captured in equal numbers; that is there was no significantdifference between treatment and control plots (sign test, n ¼ 23, p ¼ 0.14). In contrast, post-treatment 27 species werecaptured more often at treated plots compared to only four species with higher captures at control plots where buffel grassremained, while five species had equal captures across treatments (Fig. 5; sign test, n¼ 31, p < 0.0001). Although therewas noevidence that removal of buffel grass impacted the most frequently captured species post-treatment, the gecko Diplodactylusconspicillatus, all the other abundant species (�10 captures) were captured more often in treated plots (Fig. 5).

3.6. Environmental variables associated with reptile abundance

A distance-based linear model with ground cover categories as explanatory variables showed that bare ground bestexplained reptile composition accounting for 11% of variation (P ¼ 0.001). The presence of buffel grass stubble (5.5%,P ¼ 0.001) and native grasses (4.8% P ¼ 0.004) also contributed significantly to the model although explaining only a smallfraction of the variation.

A distance-based linear model on climatic variables independent from our experimental factors showed that temperatureand rainfall explained some of the temporal patterns in the reptile data. Marginal tests showed that average T max explained10.0% of the variation in reptile assemblages (P ¼ 0.001) and rainfall for the last 30, 90 and 270 days each explained 3.5e6%(P < 0.02 for all). The multivariable model with average maximum temperature and rainfall for the last 30 and 90 days provedthe best fit but explained only 16.1% of the total variation in reptile assemblages.

4. Discussion

4.1. Response of native plants and reptiles to weed removal

Our results support the prediction that weed management would have overall positive outcomes for reptiles coincidingwith the re-establishment of native plants and provide direct and indirect evidence of negative impacts of buffel grass in-vasion on this group. Typically, the response of reptiles to habitat change, and especially to changes in plant cover, depends onlife history traits and foraging and thermoregulatory behaviour of each species. For example, in spinifex (Triodia sp.) grass-lands that dominate much of the Australian arid zone, different lizard species are associated with different levels of cover,often related to successional stages post-fire (Masters 1996; Schlesinger 1997; Dickman et al., 1999; Letnic et al., 2004; Dalyet al., 2008). In contrast to what might be expected where there is progressive divergence in the structure and composition ofground cover, we did not detect significant changes in composition of reptile assemblages between treated and control plots.Instead, in the post-treatment period, differences in total abundance and species richness between treated and control sitesreflected a trend among most species for fewer captures in still invaded sites.

The similarity in composition of reptile assemblages between treated and control plots may also be accounted for, in part,by a lack of colonisation opportunity, especially if species particularly sensitive to buffel grass invasion had already dis-appeared from the local area. In the absence of deliberate replanting and re-seeding, or re-introduction of animals, the

9

Table 4PERMANOVA results on the effects of treatment, site and time on the species richness of reptiles with block as random factor nested in site. All permutationswere above 970 except for location (60 permutations only). ECV is the estimated square root of components of variation which estimates the effect size as %species dissimilarity. The residual unexplained ECV was 17.66. Results for distance-based tests for homogeneity of multivariate dispersions (Perm DISP) areshown for the main factors. *P < 0.05; **P < 0.01; ***P ¼ 0.001.


Treatment 23.5 (1) 0.009** 9.3 0.068Time 3.5 (6) 0.002** 9.7 0.011*Site 1.9 (1) 0.054 3.4 0.021*Block (random factor) 1.5 (4) 0.204 3.3 0.412Time x Treatment 2.2 (6) 0.039* 8.2Time x Site 0.4 (6) 0.957 0Treatment x Site 3.7 (1) 0.084 4.6

Fig. 3. Timeline showing mean (±se) number of (a) individual reptiles and (b) reptile species captured at treated (grey) and control (black) plots (n ¼ 6 treatedand control plots for the first five sampling periods and n ¼ 5 for last two periods). Buffel grass cover was mechanically reduced in January 2008 and herbicidetreatment of re-sprouting buffel grass commenced in February 2009. Significant differences from pair-wise tests are indicated * P < 0.05; **P < 0.01.

Table 5PERMANOVA results on the effects of treatment, site and time on the reptile assemblage with block as random factor nested in site. All permutations wereabove 970 except for location (60 permutations only). ECV is the estimated square root of components of variation which estimates the effect size as %species dissimilarity. The residual unexplained ECV was 34.4. Results for distance-based tests for homogeneity of multivariate dispersions (PermDISP) areshown for the main factors. Block was a random factor nested in Site. *P < 0.05; **P < 0.01; ***P ¼ 0.001.


Treatment 2.0 (1) 0.178 6.8 0.272Time 5.1 (6) 0.001*** 23.9 0.007**Site 8.0 (1) 0.041* 22.0 0.001***Block (random factor) 2.3 (4) 0.003** 10.8 0.519Time x Treatment 1.4 (6) 0.090 9.4Time x Site 1.7 (6) 0.006** 14.4Treatment x Site 1.9 (1) 0.175 9.0


composition of restored communities is constrained by species still occurring in the pre-restoration landscape and by op-portunities for recolonization from more distant areas. The re-establishment of native plants at our treated plots indicatesthat a diverse native seed bank had persisted post-invasion. Further replenishment of the seedbank likely occurred as theexperiment progressed, through increased local production of seed, or from dispersal by wind or animals. For reptiles, giventhe small home ranges of most species, and the unlikelihood of their dispersal over long distances (Stow et al., 2001;Kanowski et al., 2006), the species captured at treated plots were almost certainly limited to those present in the invadedlandscape at the start of the study. Hence, the higher abundance of reptiles at these plots most likely reflected selection for

10

Fig. 4. Principal component ordination of reptile abundance data (square root transformed) at Desert Park (DP) and Simpsons Gap (SG) with first and secondsampling times shown as pre-treatment; (a) with vectors indicating species most correlated with the PCO axes (Pearson correlation >0.6); and (b) representingdistances between centroids of the reptile assemblage composition at treated and control plots at each site through time with numbers (1e7) corresponding totimes when sampling occurred from November 2007eNovember 2012. Triangles mark the DP and circles the SG site. Open symbols are treated and filled symbolscontrol plots. Grey stars mark pre-treatment samples (t1-2) at SG and grey crosses pre-treatment samples at DP.


11

Fig. 5. Total number of captures of reptile species in control (black) and treated (grey) plots post-treatment (trips 3e7). Species are ordered from most abundantto least abundant.


restored native habitat - or against invaded habitat - by individuals with nearby home ranges, and possibly lower repro-duction and survival in still invaded compared to restored plots.

4.2. Impacts of invasive grasses on reptiles

From 2010, following the initial boom of reptile activity, total abundance and richness of reptiles declined in control plotsto below levels recorded at commencement of the study, whereas at treated sites, abundance and species richness returned tolevels similar to those in the pre-treatment period. Although background variability in captures among years precludesdefinitive conclusions, this pattern suggests that the evident positive impacts of weedmanagement on reptiles may reflect, atleast in part, ongoing decline in invaded areas. Despite having been invaded for at least two decades, transformation of theplant community in invaded plots was ongoing during our study, especially at the Desert Park where buffel grass covercontinued to increase. At Simpsons Gap buffel grass had probably reached maximum density a decade prior to our research(Clarke et al., 2005) and no obvious further change in cover was apparent. However, ongoing changes in the abundance and

12


composition of shrubs and trees were observed in invaded areas at this site, including the loss by fire of several corkwood(Hakea divaricata) trees estimated to bewell over 100 years old. This gradual decline in abundance of large trees and shrubs inbuffel-invaded areas, mirrored by the loss of long-lived, fire sensitive cacti in Sonora (Rodríguez-Rodríguez et al., 2017), isexpected to continue as mature plants are repeatedly impacted by more intense fire and new recruitment is constrained(Miller et al., 2010; Schlesinger et al., 2013; Edwards et al., 2019). These changes will progressively reduce the availability ofmicrohabitats important to certain reptiles, including perch sites for semi-arboreal sit-and-wait predators, and crevices underbark, logs, or leaf litter. If reptiles were continuing to decline at control plots and the surrounding invaded areas during ourstudy, then this suggests invasion impacts were ongoing and may foretell further negative effects on this group in buffel-invaded areas as plant communities continue to transform.

Because no comparable uninvaded areas remain in our study region, and there are limited baseline data on reptilecommunities in openwoodland habitats in arid Australia, it is no longer possible to assess the full effect of change associatedwith buffel grass invasion in this habitat. This challenge is common in highly modified landscapes (Jellinek et al., 2014).However, weed-removal experiments can provide evidence of the impacts of an invasive species on biodiversity (Adair andGroves 1998; Lambert et al., 2017), and are less likely to be confounded by site differences than multisite comparisons be-tween invaded and uninvaded areas. The relatively consistent response of reptile assemblages to weed removal and re-establishment of native plants in our experiment suggests a broad mechanism by which buffel grass affects most reptiles,rather than nuanced impacts on individual species resulting from their habits and life history.

Mechanisms bywhich invasive grasses affect reptiles are obscure (Abom et al., 2015; Carter et al., 2015) but the importanceof habitat structure, especially for lizards, is long established (Sexton 1958, 1959; Heatwole 1966; Pianka 1966) and change inthe thermal environment appears to be a proximal cause of detrimental impacts when habitat structure is altered (Garcia andClusella-Trullas 2019). However reduced richness and abundance of reptiles in grass-invaded areas can occur evenwhere thethermal environment is unchanged (Abom et al., 2015). Most small terrestrial vertebrates of arid environments are adapted tosparsely vegetated habitats (Germano et al., 2001), such that the dense cover of an invasive grass is likely to be problematic.Once dense swards develop, there may not only be a reduction of thermoregulatory opportunities but also obstruction ofefficient movement (Germano et al., 2012), and diminished visibility which reduces detection of prey or predators (McKinneyet al., 2015). Similar to many other invasive grasses, the cover and height attained by buffel grass is usually much greater thanthat of the replaced native plant communities. Our finding that bare ground was the cover variable that best explainedvariation in the reptile community, despite ground cover being impacted bymultiple factors including treatment, rainfall andfire, is consistent with the importance of open spaces for many arid-adapted species. Overall, the experimental treatmentsand the variables in our analyses explained only a small proportion of the temporal and spatial variation of the reptilecommunity, reflecting both complexities in design and the challenges of determining the mechanisms driving changes infaunal assemblages in desert environments of high temporal and spatial variability (Germano et al., 2012).

In addition to altered habitat structure, the most plausible explanation for negative impacts of buffel grass in multiplereptile species is reduced prey availability. Indirect trophic links between plant diversity and insectivorous reptiles can bedifficult to untangle, especially where temporal variation in invertebrate abundance and diversity is extreme. Nevertheless,the importance of prey availability in determining activity, growth and reproduction of arid zone lizards is long-established(Ballinger 1977; Dunham 1978, 1981; Schoener and Schoener 1978) and direct measures of available prey have been found tobetter explain regional differences between lizard communities than rainfall and habitat (Pastro et al., 2013). Buffel grassinvasion is associated with reduced abundance and diversity of invertebrates (Binks et al., 2005; Smyth et al., 2009; Bonneyet al., 2017) although existing research has mostly focussed on ants. Conclusive assessment of the impact of buffel grass onprey for insectivorous reptiles would require analysis of a broader array of invertebrate groups and consideration of temporalpatterns in their abundance in relation to season and rain.

In the arid environments of the southern hemisphere long dry spells punctuated by big rains, the latter often linked to theLa Ni~na phase of the El Ni~no southern oscillation, are the norm. The fundamental importance of these rainfall patterns inregulating plant growth and animal activity is evident throughout our results, affecting the efficacy of buffel grass man-agement, the restoration of native plants in treated areas and reptile captures at different times. Unlike somemore ephemeraland mobile fauna, reptiles tend to persist in low numbers throughout the landscape during drier periods (Pastro et al., 2013),although seasonal reproduction is delayed and activity and body condition reduced (Dickman et al., 1999; Greenville andDickman 2005; Schlesinger et al., 2010). Activity and reproduction then quickly resume when prey becomes more abun-dant. High rates of reproduction, juvenile survival and recruitment during resource booms are almost certainly crucial to thelocal persistence of reptile populations during more pervasive dry times. However, our results suggest that buffel grass maydisrupt these temporal patterns, limiting the opportunities associated with big rain events.

During the La Ni~na rain event, there was a strong growth pulse of diverse native plants at treated plots but not at buffel-dominated control plots. If extrapolated to the wider landscape and over time, this suggests that the abundance and diversityof native plant-based food resources associatedwith high rainfall is substantially reduced in heavily invaded areas. This can beexpected to impact directly on primary consumers including many invertebrates and, in turn, on higher order consumers.Consistent with this expectation, differences in abundance and species richness of reptiles between treatments first becameapparent at the beginning of the mesic period, when activity increased in both treated and control plots, but not as much inthe latter. Significant differences in reptile abundance between treatments were detected for the following two years,coinciding with a decrease in captures, especially at control plots. This negative trendmay indicate that the expected increasein invertebrate abundance and diversity associated with the mesic conditions was not only reduced, but also did not persist

13


for as long at the invaded plots. The response of plants to rainfall pulses is heterogeneous in space and time according to theability of different species to access soil moisture of varying depth and duration (Schwinning and Sala 2004). Because nativeplant communities are substantively replaced by a single species in areas invaded by buffel grass, the variety and timing ofgrowth responses to soil moisture pulses and associated plant resources are inevitably suppressed. We think it likely thisreduces both the variety of consumers that can benefit, and the time over which resource availability remains elevated. Theconsequent reduced ability of primary and higher order consumers to benefit from sustained high rainfall periods may proveto be one of the most serious consequences of buffel grass invasion for reptiles and other native fauna in arid Australia.

4.3. Management implications and conservation benefits of weed removal

Our research has demonstrated direct benefits of small-scale grassy weed removal for native fauna, coinciding withrestoration of native plant structure and composition. The natural re-establishment of native grasses and forbs as a result ofmanagement shows that a diverse native seed bank was still present and, in areas where this is the case e includingpotentially throughout much of central Australia, restoration of native plant communities in buffel-invaded areasmay requirerelatively low effort. That a diverse suit of reptiles was still present within the invaded landscape was also encouraging forrestoration, and consistent with findings at another Australian site (Dittmer and Bidwell 2018). Less positively, our researchsuggests that reptiles are detrimentally impacted by buffel grass invasion, and that invaded areas may experience ongoingdecline even after two decades. Other research at the same experimental sites has documented higher species richness andabundance of ants (Bonney et al., 2017) and increased foraging by birds (Young and Schlesinger 2014) in plots where buffelgrass was removed. Together, these results provide strong evidence of the value of buffel grass management for diversecomponents of the native flora and fauna even in heavily invaded areas and where management can only be undertaken at asmall scale. They also lend further weight to calls for buffel grass to be recognised world-wide as being among the mostpressing environmental issues in the regions it has invaded, and for management to be better supported (Brean 2019; Readet al., 2020).

A direct but unplanned outcome of weed removal and native plant regeneration at our treated plots was reduction ingrassy fuel and lower impacts of wildfire (Schlesinger et al., 2013). The potential for small, restored areas to act as fire refugiacould be a significant co-benefit of restoration. For example, trees that survive in unburned or low-severity burn patches canfunction as refugial seed sources for regeneration of surrounding areas (Downing et al., 2019). Green firebreaks are strips ofvegetationwith low flammability grown at strategic locations in the landscape and are increasingly being used world-wide tomanage fire (Cui et al., 2019). The use of restored native vegetation as native green firebreaks to disrupt fire invasion feedbacksin invaded dryland environments is a more novel concept (Porensky et al., 2018) that has the advantage of simultaneouslysupporting the persistence of native plant communities and providing resource ‘islands’ for native fauna. In areas invadedwith buffel grass, experimentation with green breaks of similar size to standard fire breaks would be worthwhile, withpriority given to areas with significant natural and cultural heritage, and areas where long-lived trees with high habitat valuemay be at risk (Schlesinger and Judd 2019; Westerhuis et al., 2019).

The negative economic, social and environmental costs of high impact invasive grasses (Grice et al., 2013; Setterfield et al.2013, 2018; Godfree et al., 2017; van Klinken and Friedel 2017; Read et al., 2020) point to the need for innovative, sustainable,cost-effective, landscape scale management approaches (Sutton et al., 2019). However, for some species, including buffelgrass, prospects for broad scale management are still constrained in many jurisdictions because of their continued economicvalue for pastoral industries (Grice et al., 2012). While these conflicting values remain unresolved, the creation and main-tenance of smaller areas where native vegetation has been restored offers multiple benefits, and over the longer term mayprovide repositories for diversity. We concur with Longland and Bateman (2002) and Hulvey et al. (2017) that the creation ofrestoration islands should be given further consideration as interventions to help achieve long and short-term conservationgoals in dryland ecosystems.

At a practical level, low moisture availability and extreme variability in weather pose challenges for restoration of aridecosystems. Uptake of herbicides requires active plant growth and is therefore dependent on season and rainfall, and re-establishment of native plant communities is similarly dependent on the amount and timing of rainfall after weed treat-ment. On the positive side, re-incursion by the invader is restricted to wetter times so minimal management is required forlong periods. The effort required for management will vary substantially with topography and soil type, as was evident at oursites, and this precludes precise estimates of costs. Nevertheless, our results show that upkeep takes much less effort thaninitial weed treatment, even when high rainfall encourages re-incursion, and highlight the widely recognised need forsustained effort in weed management to capitalise on the original investment. Some pre-treatment actions, such as slashing,are not weather dependent, can improve uptake of chemicals by plants during the growth phase, and significantly improvethe effectiveness of treatment. Whereas herbicide application is not a viable option for managing buffel grass over largescales, especially in areas already heavily invaded, our research shows that at smaller scales, when spraying can be targeted toavoid native species, it can be effective in restoring native plant communities, with positive local impacts on fauna. However,where spot spraying is not an option, the detrimental effect of herbicide on native species is problematic (Farrell and Gornish2019). Investigations into buffel-specific bioherbicides which show a much lower or nil toxicity compared to synthetics andminimise damage to native vegetation (Marsico et al., 2019; Masi et al., 2019) are of particular importance as these productswould significantly improve the efficiency and safety of future restoration endeavours in the regions of the world where thispervasive invader now dominates.

14


Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could haveappeared to influence the work reported in this paper.

Acknowledgements

This work was supported by Charles Darwin University, the Alice Springs Desert Park and Northern Territory Parks andWildlife. We thank the many people who assisted with field work, especially Parks and Wildlife officers from the SimpsonsGap ranger station and the Desert Park who were also responsible for buffel grass management at the field sites. Specialthanks to Clive, Alexandra and Joe Rosewarne. Stephen Morton and Margaret Friedel provided valuable comments on themanuscript.

References

Abom, R., Vogler, W., Schwarzkopf, L., 2015. Mechanisms of the impact of a weed (grader grass, Themeda quadrivalvis ) on reptile assemblage structure in atropical savannah. Biol. Conserv. 191, 75e82.

Adair, R.J., Groves, R.H., 1998. Impact of environmental weeds on biodiversity: a review and development of a methodology. In: Canberra ACT (Ed.), E. A.National Weeds Program.

Anderson, M.J., Gorley, R.N., Clarke, K.R., 2008. PERMANOVAþ for PRIMER: Guide to Software and Statistical Methods. PRIMER-E, Plymoth, UK.Ballinger, R.E., 1977. Reproductive strategies: food availability as a source of proximal variation in a lizard. Ecology 58, 628e635.Binks, R., Cann, A., Perks, S., Silla, A., Young, M., 2005. The Effect of Introduced Buffel Grass (Cenchrus Ciliaris L.) on Terrestrial Invertebrate Communities in

the Pilbara Region, Western Australia. BSc (Hons). University of Western Australia, Perth.Bonney, S., Andersen, A., Schlesinger, C., 2017. Biodiversity impacts of an invasive grass: ant community responses to Cenchrus ciliaris in arid Australia. Biol.

Invasions 19, 57e72.Brean, H., 2019. Ecologists Issue Call to Action after Buffelgrass Fire Scorches Saguaros in Catalinas. Tucson.com. Aug 30, 2019 Updated Jul 24, 2020, News.

https://tucson.com/news/local/ecologists-issue-call-to-action-after-buffelgrass-fire-scorches-saguaros-in-catalinas/article_c630a967-366b-5c37-998e-4a188bda5489.html.

Brooks, M.L., 2012. Effects of high fire frequency in creosote bush scrub vegetation of the Mojave Desert. Int. J. Wildland Fire 21, 61e68.Brooks, M.L., D’Antonio, C.M., Richardson, D.M., Grace, J.B., Keeley, J.E., DiTomaso, J.M., Hobbs, R.J., Pellant, M., Pyke, D., 2004. Effects of invasive alien plants

on fire regimes. Bioscience 54, 677e688.Carter, E.T., Eads, B.C., Ravesi, M.J., Kingsbury, B.A., 2015. Exotic invasive plants alter thermal regimes: implications for management using a case study of a

native ectotherm. Funct. Ecol. 29, 683e693.Clarke, K.R., 1993. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18, 117e143.Clarke, P.J., Latz, P.K., Albrecht, D.E., 2005. Long-term changes in semi-arid vegetation: invasion of an exotic perennial grass has larger effects than rainfall

variability. J. Veg. Sci. 16, 237e248.Cui, X., Alam, M.A., Perry, G.L., Paterson, A.M., Wyse, S.V., Curran, T.J., 2019. Green firebreaks as a management tool for wildfires: lessons from China. J.

Environ. Manag. 233, 329e336.D’Antonio, C.M., Vitousek, P.M., 1992. Biological Invasions by exotic grasses, the grass/fire cycle, and global change. Annu. Rev. Ecol. Systemat. 23, 63e87.Daly, B.G., Dickman, C.R., Crowther, M.S., 2008. Causes of habitat divergence in two species of agamid lizards in arid central Australia. Ecology 89, 65e76.Dickman, C.R., Letnic, M., Mahon, P.S., 1999. Population dynamics of two species of dragon lizards in arid Australia: the effects of rainfall. Oecologia 119,

357e366.Dittmer, D.E., Bidwell, J.R., 2018. Herpetofaunal species presence in buffel grass (Cenchrus ciliaris) versus native vegetation-dominated habitats at Uluṟu-

Kata Tjuṯa National Park. Austral Ecol. 43, 203e212.Downing, W.M., Krawchuk, M.A., Meigs, G.W., Haire, S.L., Coop, J.D., Walker, R.B., Whitman, E., Chong, G., Miller, C., 2019. Influence of fire refugia spatial

pattern on post-fire forest recovery in Oregon’s Blue Mountains. Landsc. Ecol. 34, 771e792.Dunham, A.E., 1978. Food availability as a proximate factor influencing individual growth rates in the iguanid lizard Sceloporus merriami. Ecology 59,

770e778.Dunham, A.E., 1981. Populations in a Fluctuating Environment: the Comparative Population Ecology of the Iguanid Lizards Sceloporus Merriami and Uro-

saurus Ornatus, vol. 158. Miscellaneous Publication of the Museum of Zoology University of Michigan, pp. 1e62.Edwards, K.M., Schlesinger, C., Ooi, M.K.J., French, K., Gooden, B., 2019. Invasive grass affects seed viability of native perennial shrubs in arid woodlands. Biol.

Invasions 21, 1763e1774.Farrell, H.L., Gornish, E.S., 2019. Pennisetum ciliare: a review of treatment efficacy, competitive traits, and restoration opportunities. Invasive Plant Sci.

Manag. 12, 203e213.Firn, J., Martin, T.G., Chad�es, I., Walters, B., Hayes, J., Nicol, S., Carwardine, J., 2015. Priority threat management of non-native plants to maintain ecosystem

integrity across heterogeneous landscapes. J. Appl. Ecol. 52, 1135e1144.Franklin, K., Molina-Freaner, F., 2010. Consequences of buffelgrass pasture development for primary productivity, perennial plant richness, and vegetation

structure in the drylands of Sonora, Mexico. Conserv. Biol. 24, 1664e1673.Franks, A.J., 2002. The ecological consequences of Buffel Grass Cenchrus ciliaris establishment within remnant vegetation of Queensland. Pac. Conserv. Biol.

8, 99e107.Friedel, M.H., Grice, A.C., Marshall, N.A., van Klinken, R.D., 2011. Reducing contention amongst organisations dealing with commercially valuable but

invasive plants: the case of buffel grass. Environ. Sci. Pol. 14, 1205e1218.Gaertner, M., Biggs, R., Te Beest, M., Hui, C., Molofsky, J., Richardson, D.M., 2014. Invasive plants as drivers of regime shifts: identifying high-priority invaders

that alter feedback relationships. Divers. Distrib. 20, 733e744.Garcia, R.A., Clusella-Trullas, S., 2019. Thermal landscape change as a driver of ectotherm responses to plant invasions. Proceedings of the Royal Society B

286, 20191020.Germano, D.J., Rathbun, G.B., Saslaw, L.R., 2001. Managing exotic grasses and conserving declining species. Wildl. Soc. Bull. 29, 551e559.Germano, D.J., Rathbun, G.B., Saslaw, L.R., 2012. Effects of grazing and invasive grasses on desert vertebrates in California. J. Wildl. Manag. 76, 670e682.Godfree, R., Firn, J., Johnson, S., Knerr, N., Stol, J., Doerr, V., 2017. Why non-native grasses pose a critical emerging threat to biodiversity conservation, habitat

connectivity and agricultural production in multifunctional rural landscapes. Landsc. Ecol. 32, 1219e1242.Goverment’, N.T, 2020. Threatened Species of the Northern Territory.Gray, K.M., Steidl, R.J., 2015. A plant invasion affects condition but not density or population structure of a vulnerable reptile. Biol. Invasions 17, 1979e1988.Greenville, A.C., Dickman, C.R., 2005. The ecology of Lerista labialis (Scincidae) in the Simpson Desert: reproduction and diet. J. Arid Environ. 60, 611e625.Grice, A., 2006. The impacts of invasive plant species on biodiversity of Australian rangelands. Rangeland Journal - RANGELAND J 28.

15

http://refhub.elsevier.com/S2351-9894(20)30866-0/sref1














https://tucson.com/news/local/ecologists-issue-call-to-action-after-buffelgrass-fire-scorches-saguaros-in-catalinas/article_c630a967-366b-5c37-998e-4a188bda5489.html

https://tucson.com/news/local/ecologists-issue-call-to-action-after-buffelgrass-fire-scorches-saguaros-in-catalinas/article_c630a967-366b-5c37-998e-4a188bda5489.html










































































Grice, A., Friedel, M., Marshall, N., Klinken, R., 2012. Tackling contentious invasive plant species: a case study of buffel grass in Australia. Environ. Manag. 49,285e294.

Grice, A.C., Vanderduys, E.P., Perry, J.J., Cook, G.D., 2013. Patterns and processes of invasive grass impacts on wildlife in Australia. Wildl. Soc. Bull. 37,478e485.

Hacking, J., Abom, R., Schwarzkopf, L., 2014. Why do lizards avoid weeds? Biol. Invasions 16, 935e947.Heatwole, H., 1966. Factors affecting orientation and habitat selection in some geckos. Zeitschrift fur Tierpsychologie 23, 303e314.Hulvey, K.B., Leger, E.A., Porensky, L.M., Roche, L.M., Veblen, K.E., Fund, A., Shaw, J., Gornish, E.S., 2017. Restoration islands: a tool for efficiently restoring

dryland ecosystems? Restor. Ecol. 25, S124eS134.IUCN, 2000. IUCN Guidelines for the prevention of biodiversity loss caused by alien invasive species. In: I. U. F. C. O. Nature. Gland Switzerland.Jellinek, S., Parris, K.M., McCarthy, M.A., Wintle, B.A., Driscoll, D.A., 2014. Reptiles in restored agricultural landscapes: the value of linear strips, patches and

habitat condition. Anim. Conserv. 17, 544e554.Kanowski, J.J., Reis, T.M., Catterall, C.P., Piper, S.D., 2006. Factors affecting the use of reforested sites by reptiles in cleared rainforest landscapes in tropical

and subtropical Australia. Restor. Ecol. 14, 67e76.Knapp, P.A., 1996. Cheatgrass (Bromus tectorum L) dominance in the Great Basin Desert: history, persistence, and influences to human activities. Global

Environ. Change 6, 37e52.Lambert, K.T.A., Reid, N., McDonald, P.G., 2017. Does the removal of Lantana camara influence eucalypt canopy health, soil nutrients and site occupancy of a

despotic species? For. Ecol. Manag. 394, 104e110.Lawson, B., Bryant, M., Franks, A., 2004. Assessing the potential distribution of buffel grass (Cenchrus ciliaris L.) in Australia using a climate-soil model. Plant

Protect. Q. 19, 155e163.Letnic, M., Dickman, C.R., Tischler, M.K., Tamayo, B., Beh, C.L., 2004. The responses of small mammals and lizards to post-fire succession and rainfall in arid

Australia. J. Arid Environ. 59, 85e114.Longland, W.S., Bateman, S.L., 2002. Viewpoint: the ecological value of shrub Islands on disturbed sagebrush rangelands. J. Range Manag. 55, 571e575.Marsico, G., Ciccone, M., Masi, M., Freda, F., Cristofaro, M., Evidente, A., Superchi, S., Scafato, P., 2019. Synthesis and herbicidal activity against buffelgrass

(cenchrus ciliaris) of (±)-3-deoxyradicinin. Molecules 24, 3193.Martin, L.J., Murray, B.R., 2011. A predictive framework and review of the ecological impacts of exotic plant invasions on reptiles and amphibians. Biol. Rev.

86, 407e419.Martin, T.G., Murphy, H., Liedloff, A., Thomas, C., Chad�es, I., Cook, G., Fensham, R., McIvor, J., van Klinken, R.D., 2015. Buffel grass and climate change: a

framework for projecting invasive species distributions when data are scarce. Biol. Invasions 17, 3197e3210.Masi, M., Freda, Clement, Cimmino, A., Cristofaro, M., Meyer, S., Evidente, A., 2019. Phytotoxic activity and structureeactivity relationships of radicinin

derivatives against the invasive weed buffelgrass (cenchrus ciliaris). Molecules 24, 2793.Masters, P., 1996. The effects of fire-driven succession on reptiles in spinifex grasslands at Uluru National Park, Northern Territory. Wildl. Res. 23, 39e47.McAlpine, C., Catterall, C.P., Nally, R.M., Lindenmayer, D., Reid, J.L., Holl, K.D., Bennett, A.F., Runting, R.K., Wilson, K., Hobbs, R.J., Seabrook, L., Cunningham, S.,

Moilanen, A., Maron, M., Shoo, L., Lunt, I., Vesk, P., Rumpff, L., Martin, T.G., Thomson, J., Possingham, H., 2016. Integrating plant- and animal-basedperspectives for more effective restoration of biodiversity. Front. Ecol. Environ. 14, 37e45.

McDonald, C.J., McPherson, G.R., 2011. Fire behavior characteristics of buffelgrass-fueled fires and native plant community composition in invaded patches.J. Arid Environ. 75, 1147e1154.

McDonald, C.J., McPherson, G.R., 2013. Creating hotter fires in the Sonoran desert: buffelgrass produces copious fuels and high fire temperatures. FireEcology 9, 26e39.

McKinney, M.A., Schlesinger, C.A., Pavey, C.R., 2015. Foraging behaviour of the endangered Australian skink (Liopholis slateri). Aust. J. Zool. 62, 477e482.Miller, G., Friedel, M., Adam, P., Chewings, V., 2010. Ecological impacts of buffel grass (Cenchrus ciliaris L.) invasion in central Australia - does field evidence

support a fire-invasion feedback? Rangel. J. 32, 353e365.Morton, S.R., Emmott, A.J., 2014. Lizards of the Australian deserts: uncovering an extraordinary ecological story. Hist. Record. Aust. Sci. 25, 217e226.Mueller-Dombois, D., Ellenberg, H., 1974. Aims and Methods of Vegetation Ecology. John Wiley & Sons, New York.Pastro, L.A., Dickman, C.R., Letnic, M., 2013. Effects of wildfire, rainfall and region on desert lizard assemblages: the importance of multi-scale processes.

Oecologia 1e12.Pianka, E.R., 1966. Convexity, desert lizards, and spatial heterogeneity. Ecology 47, 1055e1059.Porensky, L.M., Perryman, B.L., Williamson, M.A., Madsen, M.D., Leger, E.A., 2018. Combining active restoration and targeted grazing to establish native

plants and reduce fuel loads in invaded ecosystems. Ecol. Evol. 8, 12533e12546.Quirion, B., Simek, Z., D�avalos, A., Blossey, B., 2018. Management of invasive Phragmites australis in the Adirondacks: a cautionary tale about prospects of

eradication. Biol. Invasions 20, 59e73.Read, J.L., Firn, J., Grice, A.C., Murphy, R., Ryan-Colton, E., Schlesinger, C.A., 2020. Ranking buffel: comparative risk and mitigation costs of key environmental

and socio-cultural threats in central Australia. Ecol. Evol. 1e19, 00.Reid, A.M., Morin, L., Downey, P.O., French, K., Virtue, J.G., 2009. Does invasive plant management aid the restoration of natural ecosystems? Biol. Conserv.

142, 2342e2349.Rodríguez-Rodríguez, L., Stafford, E., Williams, A., Wright, B., Kribs, C., Ríos-Soto, K., 2017. A Stage Structured Model of the Impact of Buffelgrass on Saguaro

Cacti and Their Nurse Trees.Schlesinger, C., Judd, B., 2019. The summer bushfires you didn’t hear about, and the invasive species fuelling them. In: The Conversation, March 12, 2019 5:

49am AEDT, Environment þ Energy. http://theconversation.com/the-summer-bushfires-you-didnt-hear-about-and-the-invasive-species-fuelling-them-112619.

Schlesinger, C., White, S., Muldoon, S., 2013. Spatial pattern and severity of fire in areas with and without buffel grass (Cenchrus ciliaris) and effects onnative vegetation in central Australia. Austral Ecol. 38, 831e840.

Schlesinger, C.A., 1997. Fire studies in mallee (Eucalyptus spp.) communities of western New South Wales: reptile and beetle populations in sites of differingfire history. Rangel. J. 19, 190e205.

Schlesinger, C.A., Christian, K.A., James, C.D., Morton, S.R., 2010. Seven lizard species and a blind snake: activity, body condition and growth of desertherpetofauna in relation to rainfall. Aust. J. Zool. 58, 273e283.

Schoener, T.W., Schoener, A., 1978. Estimating and interpreting body-size growth in some Anolis lizards. Copeia 1978, 390e405.Schwinning, S., Sala, O.E., 2004. Hierarchy of responses to resource pulses in arid and semi-arid ecosystems. Oecologia 141, 211e220.Scott, J., 2014. Australian Rangelands and Climate Change. Ninti One Limited and CSIRO, Alice Springs, Alice Springs.Setterfield, S.A., Rossiter-Rachor, N.A., Adams, V.M., 2018. Navigating the fiery debate: the role of scientific evidence in eliciting policy and management

responses for contentious plants in northern Australia. Pac. Conserv. Biol. 24, 318e328.Setterfield, S.A., Rossiter-Rachor, N.A., Douglas, M.M., Wainger, L., Petty, A.M., Barrow, P., Shepherd, I.J., Ferdinands, K.B., 2013. Adding fuel to the fire: the

impacts of non-native grass invasion on fire management at a regional scale. PloS One 8, e59144.Sexton, O.J., 1958. The relationship between the habitat preferences of hatchling Chelydra serpentina and the physical structure of the vegetation. Ecology

39, 751e754.Sexton, O.J., 1959. Spatial and temporal movements of a population of the painted turtle, Chrysemys picta marginata (Agassiz). Ecol. Monogr. 29, 137e140.Smyth, A., Friedel, M., O’Malley, C., 2009. The influence of buffel grass (Cenchrus ciliaris) on biodiversity in an arid Australian landscape. Rangel. J. 31,

307e320.Stow, A.J., Sunnucks, P., Briscoe, D.A., Gardner, M.G., 2001. The impact of habitat fragmentation on dispersal of Cunningham’s skink (Egernia cunninghami):

evidence from allelic and genotypic analyses of microsatellites. Mol. Ecol. 10, 867e878.

16
























































































http://theconversation.com/the-summer-bushfires-you-didnt-hear-about-and-the-invasive-species-fuelling-them-112619

http://theconversation.com/the-summer-bushfires-you-didnt-hear-about-and-the-invasive-species-fuelling-them-112619
































Sutton, G.F., Canavan, K., Day, M.D., den Breeyen, A., Goolsby, J.A., Cristofaro, M., McConnachie, A., Paterson, I.D., 2019. Grasses as suitable targets for classicalweed biological control. BioControl 64, 605e622.

van Klinken, R.D., Friedel, M.H., 2017. Unassisted invasions: understanding and responding to Australia’s high-impact environmental grass weeds. Aust. J.Bot. 65, 678e690.

Watson, J., Venter, O., Lee, J., Jones, K., Robinson, J., Possingham, H., Allan, J., 2018. Protect the last of the wild. Nature 563, 27e30.Westerhuis, E.L., Schlesinger, C.A., Nano, C.E.M., Morton, S.R., Christian, K.A., 2019. Characteristics of hollows and hollow-bearing trees in semi-arid river red

gum woodland and potential limitations for hollow-dependent wildlife. Austral Ecol. 44, 995e1004.Williams, D.G., Baruch, Z., 2000. African grass invasion in the americas: ecosystem consequences and the role of ecophysiology. Biol. Invasions 2, 123e140.Young, L., Schlesinger, C., 2014. Habitat use and behaviour of birds in areas invaded by buffel grass (Cenchrus ciliaris L.) and in restored habitat. Wildl. Res. 41,

379e394.

17

















Documents

Response of reptiles to weed-control and native plant