CM

Ca

b

c

a

ARR1AA

KCCRDRS

1

aipot(he

5

c

0d

Agriculture, Ecosystems and Environment 148 (2012) 44– 64

Contents lists available at SciVerse ScienceDirect

Agriculture, Ecosystems and Environment

jo u r n al hom ep age: www.elsev ier .com/ locate /agee

arbon management of commercial rangelands in Australia:ajor pools and fluxes

hristopher Deana,b,∗, Grant W. Wardell-Johnsonb, Richard J. Harperc

School of Biological, Earth and Environmental Sciences, University of NSW, Sydney, NSW 2052, AustraliaCurtin Institute for Biodiversity and Climate, School of Science, Curtin University, GPO Box U1987, Perth, WA 6845, AustraliaAlcoa Chair of Sustainable Water Management, School of Environmental Science, Murdoch University, Murdoch, WA 6150, Australia

r t i c l e i n f o

rticle history:eceived 7 April 2011eceived in revised form9 November 2011ccepted 21 November 2011vailable online 16 December 2011

eywords:arbon sequestrationarbon emissionangelandseforestationegrowthoil organic carbon

a b s t r a c t

Land-use emissions accompanying biomass loss, change in soil organic carbon (�SOC) and decomposingwood-products, were comparable with fossil fuel emissions in the late 20th century. We examine therates, magnitudes and uncertainties for major carbon (C) fluxes for rangelands due to commercial grazingand climate change in Australia. Total net C emission from biomass over 369 Mha of rangeland to-date was0.73 (±0.40) Pg, with 83% of that from the potentially forested 53% of the rangelands. A higher emissionestimate is likely from a higher resolution analysis. The total �SOC to-date was −0.16 (±0.05) Pg. Carbonemissions from all rangeland pools considered are currently 32 (±10) Tg yr−1—equivalent to 21 (±6)%of Australia’s Kyoto-Protocol annual greenhouse gas emissions. The �SOC from erosion and deforesta-tion was −4.0 (±1.6) Tg yr−1—less than annual emissions from livestock methane, or biomass attrition,however it will continue for several centuries. Apart from deforestation a foci of land degradation wasriparian zones. Cessation of deforestation and onset of rehabilitation of degraded rangeland would allowSOC recovery. If extensive rehabilitation started in 2011 and erosion ceased in 2050 then a �SOC of−1.2 (±0.5) Pg would be avoided. The fastest sequestration option was maturation of regrowth forestin Queensland with a C flux of 0.36 (±0.18) Mg ha−1 yr−1 in biomass across 22.7 Mha for the next 50 yr;equivalent to ∼50% of national inventory agriculture emissions (as of mid 2011); and long-term seques-

tration would be 0.79 (±0.40) Pg. Due to change in water balance, temperature and accompanying fireand drought regimes from climate change, the forecast �SOC from the forested rangelands to 0.3 m depthwas −1.8 (0.6) Pg (i.e. 38 (12)% of extant SOC stock) resulting from a change in biomass from 2000 to 2100.For improved management of rangeland carbon fluxes: (a) more information is needed on the locationof land degradation, and the dynamics and spatial variation of the major carbon pools and fluxes; and (b)

ed b

freer data transfer is need

. Introduction

Since establishment of the IPCC in 1988 there has been enthusi-sm to address anthropogenic climate change by reducing growthn atmospheric greenhouse gasses (GHG). However, annual anthro-ogenic carbon (C) efflux has not decreased and urgency has arisenwing to increasing positive feedback from climate change andhe preference to stabilise CO2-e concentration below 550 ppmvAnderson and Bows, 2008). Managing C fluxes requires compre-

ensive accounting, of both past activity and future options. Forxample, the measurement of reduced C sink efficiency has been

∗ Corresponding author. Present Address. PO Box 3520, Rundle Mall, Adelaide, SA000, Australia. Tel.: +61 8 8268 8230.

E-mail addresses: cdeanspace@gmail.com, christopher.dean@unsw.edu.au,hristopher.dean@curtin.edu.au (C. Dean).

167-8809/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.agee.2011.11.011

etween government departments, and to the scientific community.© 2011 Elsevier B.V. All rights reserved.

questioned due to the similar effect from uncounted land-use emis-sions (Gloor et al., 2010).

The net anthropogenic reduction in woody biomass for pasture,urbanisation, and crops (agricultural, forestry and energy) from thelate Pleistocene to the present day, has had a trend of increas-ing influence on global warming (Edney et al., 1990; Caseldineand Hatton, 1993; Rudiman, 2003; Salinger, 2007; Olofsson andHickler, 2008; Metz, 2009; Pinter et al., 2011). Future land-use pro-cesses may emit a further 100 Pg of C (Schimel, 1995). Soil organiccarbon (SOC) concentration is generally correlated with long-termbiomass stock (Jackson and Ash, 1998; Wynn et al., 2006), primarilydue to root turnover and litterfall. Thus, in the long-term, declinein biomass generally causes decline in SOC. Emissions from bothbiomass and SOC, due to land degradation, contribute to climate

change, and in places there is feedback to further land degradation(Sivakumar, 2007).

Around 40–60% of anthropogenic emissions since pre-industrialtimes remain in the atmosphere (House et al., 2002; Elmegreen

C. Dean et al. / Agriculture, Ecosystems a

Fig. 1. Comparison of major sources of anthropogenic C emissions.Data sources: deforestation and fossil fuels, Steffen (2009); deforestation plus wood-pi

RFo2u(eClHSoerwZpCwdc

wAdhAotpdA

cEshdtmiee

used to determine the commercial grazing land in GIS compati-ble format. In all calculations carbon was considered to be 50% byweight of dry biomass.

roducts and soil change, Houghton (2008). (N.B. Houghton assumed zero changen soil carbon with timber harvesting.)

afelski et al., 2009) and continue to contribute to climate change.uture anthropogenic impact from land-use is forecast to be 9–22%f the total anthropogenic increase in CO2 to 2100 (Sitch et al.,005). Emissions from fossil fuel usage overtook those from land-se in 1959 (±7) (Woodwell et al., 1983). From more recent dataHoughton, 2008; Steffen, 2009) and by assigning polynomials tomission trends (Fig. 1) it appears that it was in 1913 (±10) that the

efflux from fossil fuel combustion, exceeded that from biomassoss due to anthropogenic deforestation and forest degradation.owever, when decomposition of wood-products and change inOC (�SOC) from land-use are included then fossil fuel emissionsnly overtook emissions from land-use in 1988 (±20). Fossil fuelmissions may have only surpassed those from land-use even moreecently than that, as Houghton (2008) had assumed zero �SOCith timber harvesting, which is unlikely (e.g. Diochon et al., 2009;

ummo and Friedland, 2011). Emissions from land-use and wood-roducts both have positive future trends. Thus, for the purposes of

management, land-use influences on SOC are significant and areorthy of scrutiny, especially as their emissions can be less imme-iate (and hence less easily measured) than those from fossil fuelombustion.

This study is centred on the Australian commercial rangelands,hich differ from those elsewhere (e.g. the grasslands of Southmerica and the Great Plains of the USA), in that they are pre-ominantly shrubland and woodland with lesser amounts of scrub,eath and herbland (Carnahan, 1977; Luly, 1993). Nevertheless,ustralia and the USA share similarities in the industrialisationf their arid and semiarid rangelands (e.g. investment, specula-ion, social mores and exports) (Heathcote, 1969). There are alsoarallels between the historical livestock-driven environmentalegradation processes in Australia, the Americas and southernfrica (Pickup, 1998).

The Australian commercial rangelands are essentially dedi-ated to the export market of livestock products, whereas prior touropean settlement (1788) they were used by subsistence-basedocieties. The accompanying changes in Australia’s native forestsave principally been through degradation, timber extraction,eforestation, regrowth and woody-thickening. Ongoing degrada-ion of Australian rangelands emits CO2, whereas less-degrading

anagement would reduce those emissions; furthermore, revers-

ng degradation could sequester C (Howden et al., 1991; McKeont al., 1992; Glenn et al., 1993; Walker and Steffen, 1993; Henryt al., 2002; Hill et al., 2006).

nd Environment 148 (2012) 44– 64 45

Estimates of �SOC accompanying rangeland deforestation andregrowth in Australia have high uncertainty (Henry et al., 2002),which is one reason they were not included in Australia’s adoptionof the Kyoto Protocol GHG accounting. Similarly, �SOC accompa-nying slower processes such as woody re-colonisation or forestdegradation from erosion, are not currently included; also as theycorrespond to ‘grazing management’, which was excluded fromAustralia’s Kyoto-Protocol accounting. Such fluxes are examinedin this study in order to provide estimates of scientific relevance,rather than to be guided by an administrative selection.

In recent decades the highest rates of deforestation in Australiawere in the State of QLD, which maintains the most comprehensiverecords in Australia of deforestation from LANDSAT remote-sensing(Queensland Department of Natural Resources, 2000). Emissionsfrom biomass due to deforestation over the last decade were∼11 Tg-C yr−1 (from ∼0.375 Mha yr−1), and the deforestation rateis decreasing (DNRW, 2008). Emissions from deforestation in QLD,from all sources, are likely to remain approximately at 2006 levels(Raison et al., 2009). Changed forest use in QLD offers the highest,net sequestration in the rangelands nationally, through reduceddeforestation and managed regeneration, but also with biodiver-sity plantings and plantations (Bray and Golden, 2008; Fenshamand Guymer, 2009).

This study examines (a) the magnitudes and timescales of majoremissions and sequestration for the commercial rangelands ofAustralia with change of forest cover, and (b) the effect of climatechange on SOC; and we illustrate the options and hurdles in recog-nising, assessing and managing the larger carbon pools and fluxes inrangelands. As a subset of the national rangelands, we examine thecommercial rangelands in QLD, where management has the mostpractical capacity to reduce emissions.

2. Materials and methods

2.1. Study region

The Australian rangeland zone covers 661 Mha (Donohue et al.,2005). Excluding from that zone, reserves and mesic arable land,the commercial livestock properties occupy ∼369 (±5) Mha (Fig. 2).Subsections of that area were selected for examination of livestockwatering points and forest regrowth (Section 2.2). State and Ter-ritory cadastre maps, combined with federal land-use maps were

Fig. 2. Distribution of C in potential forest aboveground biomass across Australia,showing intersection with the commercial rangelands and areas of deforestation.

4 ems a

(icot(tui

bbidwhCp−

2

2

dpadswt(eiApmatRvlaexrrsidt

gtprN

2

w(lsl

6 C. Dean et al. / Agriculture, Ecosyst

We used the Australian Government’s definition of forestDCCEE, 2010): a stand of trees covering at least 0.2 ha, attain-ng at least 2 m high at maturity and with at least 20% projectedanopy cover. That top height is low compared with that typicalf temperate forests but the basic density of several major Aus-ralian semiarid species is comparatively high, near 1000 kg m−3

Ilic et al., 2000). A projected canopy cover of 20% correspondso approximately 11% foliage projected cover (FPC)—a thresholdsed in remote-sensing analyses of rangelands to determine which

mage pixels represent forest cover (Scarth et al., 2008).Distance from water [for livestock] is the major influence on

iomass, via reduced growth and dieback, with erosion elsewhereeing the next major influence (Sparrow et al., 2003). From qual-

tative observation of rangeland condition and associated carbonynamics across Australia (Dean, 2005; Roxburgh et al., 2006) itas noticed that watering points were frequently close to areas ofigher biomass, and that the majority of those were riparian zones.onsequently we examined the proximity of creeks and wateringoints; the most readily accessible data was for ∼6 Mha centred at22.98◦S, 135.27◦E in the southern Northern Territory.

.2. Nationwide emission/sequestration

.2.1. Net �C to-dateAn estimate of the total C emitted from biomass to-date

ue to commercial grazing, was obtained from the pre- andost- (i.e. extant) European-colonisation biomass surfaces (Berrynd Roderick, 2006) (termed B&R hereafter). Those layers wereerived from growth modelling under past and present atmo-pheric CO2 concentrations and from remote sensing; resolutionas (0.05 × 0.05◦, i.e. ∼5.5 × 5.5 km pixels). Forest extents for those

wo periods were available in GIS polygon form at 1:1,000,000 scaleDEWR, 2007). The B&R layer of extant biomass represented netffects on biomass, such as from deforestation, land degradation,nfrastructure, regrowth, improved pasture and woody-thickening.

higher resolution layer (0.0025 × 0.0025◦, i.e. ∼250 × 250 mixels) of forest potential biomass was obtained from the Com-onwealth Department of Climate Change and Energy Efficiency,

s used in the National Carbon Accounting System (NCAS) and inhe model FullCAM (Richards and Brack, 2004; Waterworth andichards, 2008). The NCAS layer was of aboveground biomass. Aalue of 15% (representing root biomass) was added to the NCASayer, this being an average of 10.5% for ash eucalypts (Feller, 1980)nd 20% for temperate, high-biomass eucalypt forests (Mokanyt al., 2006). This value was therefore conservative as (a) more-eric environments (such as semiarid rangelands) have a higheroot:shoot ratio, e.g. 1.84 and 0.642 for shrubland and savannahespectively (Mokany et al., 2006), and (b) the global average plantpecies root:shoot ratio is 0.333 (Niklas and Enquist, 2002), as usedn Berry and Roderick (2006). Effects of a higher root:shoot ratioerived from regional allometrics (Henry et al., 2002) was alsoested.

One estimate of change in rangeland biomass due to commercialrazing was calculated by the B&R pre-commercial-biomass fromhe B&R extant-biomass layer. Another estimate was from multi-lying the ratio of extant/pre-commercial from the B&R layers (i.e.elative change) by the NCAS layer and subtracting that from theCAS layer, to get absolute change in biomass, resampled to 250 m.

.2.2. Annual fluxesEmission estimates were available for: (a) biomass loss

ith deforestation, savannah burning and livestock methane

Robertson, 2003); (b) diesel fuel usage (Rolfe, 2002); (c) biomassoss with deforestation (Henry et al., 2002); (d) biomass andoil pools modelling (Hill et al., 2006); (e) overgrazing of mulgaands (Moore et al., 2001); and (f) biomass loss with deforestation

nd Environment 148 (2012) 44– 64

(DNRW, 2008). Error margins were not given for most of thoseestimates but typical variation between different reported valuesfor each emission type was in the order of 30%—the error marginwe assigned when tallying emissions. National rangeland fluxeswere tallied from contributions (a) to (f) above and from our owncalculations for �SOC.

For the estimate of emissions from biomass due to QLDdeforestation we used only the rangeland portion from DNRW(2008)—10.2 Tg yr−1 from 0.343 Mha yr−1. QLD emissions fromdeforestation are ∼65% of the national total (Raison et al., 2009). Weobserved qualitatively, substantial rangeland woody-vegetationclearing in both NSW and NT in the form of thinning, fence lineclearing, pasture development and mustering laneways. Clearingrates in NSW have recently increased (Fensham et al., 2011). There-fore if QLD deforestation rates decrease further then the figure of10.2 Tg yr−1 will remain for several years equitable to a nationaltotal.

Estimates for C emissions from livestock excreta and savan-nah burning of 13 and 1.8 Tg yr−1 respectively (Cook et al.,2010) included undisclosed amounts from outside the commercialrangelands. Consequently, we used instead the rangeland-specificestimates of 8.2 and 3.5 Tg yr−1 (Robertson, 2003).

SOC efflux following deforestation was estimated from ear-lier reports. Following deforestation in QLD the SOC to 1 m depthdropped by 3.42 Mg ha−1 over ∼15 yr (Harms et al., 2005). (Thatrate is not perpetual but would slow over two millennia). Usingthe recent deforestation rate of 0.343 Mha yr−1 for QLD (DNRW,2008), which was conservative when extrapolated over the past30 yr, gave a State-wide SOC emission from deforestation in QLDof 2.3 (±0.7) Tg yr−1. Another estimate of annual SOC efflux fromdeforestation was 10% of annual lost biomass (Raison et al.,2009). That proportion is half of that calculated here but it maynot have included previous deforestation and therefore wouldcease to be tallied if deforestation ceased; whereas the value wepropose includes effects from earlier deforestation and would con-tinue for several centuries, although at a continually decreasingrate (Dean et al., 2012).

The fate of SOC in eroded soil worldwide constitutes an areaof great uncertainty. With erosion in Australian rangelands thereis a net loss of nutrients (Sparrow et al., 2003), which suggeststhat there would be a net SOC emission from eroded soil, ratherthan merely SOC translocation. Australian rangelands currentlylose soil with a flux of 5.6 Mg ha−1 yr−1 (Loughran et al., 2004).Soil loss from the mostly un-commercialised sites is in the orderof 1 Mg ha−1 yr−1, similar to the upper limit of soil formation(Loughran et al., 2004). Thus, a conservative estimate for soil loss,due to rangeland commercialisation is 4.6 Mg ha−1 yr−1. Calcula-tion of SOC efflux due to erosion Australia-wide was based on thatearlier work and adjusted to fit a logistic curve (slower at onset andcompletion). Assuming a typical but conservative SOC concentra-tion of 0.5% by weight, and a conservative SOC loss-fraction fromeroded soil of 20% (Lal, 2004; Lal et al., 2004) gives, for the 369 Mhaof commercial rangelands, SOC efflux at 1.7 (±0.9) Tg yr−1 (assum-ing a 50% error margin). Loss of dissolved organic carbon (DOC)from the soil was assumed to be part of the SOC erosion loss andwas not modelled separately.

We calculated two total annual emissions: (a) ‘Total1’included erosion plus losses from recent deforestation and dieselusage, and (b) ‘Total2’ = ‘Total1’ + methane emissions from live-stock + savannah burning.

To gauge the C fluxes associated with sequestration in therangelands two scenarios were examined: (a) rehabilitation of

degraded rangelands nationwide starting in 2011, using rate datafrom woody-thickening (Hibbard et al., 2003), and (b) regrowth ofrecently deforested land in QLD. For the purpose of showing theimpact of ameliorative management, two subsets of rehabilitation

ems a

wc

wwgCo

2

pCJfS2tmdtpumGsr

Gr0f�a2at

2

t(ahots(EmAg2tvmipmt

rbp

C. Dean et al. / Agriculture, Ecosyst

ere considered: (a) erosion ceased in 2050, and (b) erosionontinued.

The process of sequestering SOC through establishment ofoody vegetation in semiarid climates on degraded land is slow,ith an asymptote near 2000 yr (Hibbard et al., 2003). Error mar-

ins for SOC sequestration on degraded rangeland were high.onsequently our estimates were intended as a guide to the orderf magnitude.

.3. Climate change affect on �SOC

We calculated �SOC accompanying the changes in annual tem-erature and rainfall resulting from climate change, but withoutO2 fertilisation, following the method of Dean and Wardell-

ohnson (2010). Comprehensive modelling of climate change foundorest biomass to be most responsive to change in precipitation, andOC more responsive to change in temperature (Cowling and Shin,006) but the link between biomass and SOC would mean that long-erm change in rainfall would also affect SOC. Owing to the probably

ore-pronounced impact of climate change on forest ecosystemsue to increased fire (Gonzalez et al., 2010) we chose to examinehe forested areas only and not the non-timbered grasslands. Tem-erature and rainfall layers for the future climate were calculatedsing Ozclim (CSIRO, 2007) with settings from the global climateodels ‘ECHAM5/MPI-OM’, ‘Miroc-M’, ‘CSIRO-Mk3.5’ and ‘ECHO-’; under medium climate sensitivity and with the IPCC emissioncenarios ‘A1B’ and ‘A1FI’. The choice of climate models followedecommendations for Australia in Perkins et al. (2007).

SOC data to 0.3 m for Australia, for the year 2000, was available inIS raster format (from Barson et al., 2002). The potentially forested

angelands contained 69% of the rangeland SOC to 0.3 m. SOC to.3 m depth for the years 2000 and 2100 was calculated using theormula: SOC = −546 − (0.00756MAT3) − 134.54ln(MAP), R2 = 0.75,

= 74.34, where MAT is mean annual temperature and MAP is meannnual precipitation (Bird et al., 2002). The calculated SOC layer for100 was divided by that for 2000 to give relative change. The rel-tive change was multiplied by the SOC layer (Barson et al., 2002)o obtain the absolute value of SOC in 2100, to 0.3 m.

.4. Regrowth in QLD

Remnant and non-remnant forest ecosystem data from vege-ation mapping were available for 2003 from the QLD HerbariumRosemary Niehus, 2009; personal communication). Data were notvailable for some parts of western QLD, however deforestationad been mostly in the eastern, higher-rainfall country. The areaf analysis was 22.66 Mha, consisting of 114,749 remnant ecosys-em polygons. Extant biomass of regrowth was calculated fromatellite-derived maps of FPC at 25 × 25 m resolution, as of 2007John Armston, Dave Harris, and Craig Shephard; QLD Dept. ofnvironmental and Resource Management; 2009; personal com-unication). FPC was converted to stand basal area (formula fromrmston et al., 2009) and stand basal area was converted to above-round and belowground biomass (formulae from Henry et al.,002). Those three formulae were developed for mature vegeta-ion rather than for regrowth, which meant that calculated biomassalues for juvenile regrowth (which can have early canopy closure)ay have been too high. However, as the aim was to calculate miss-

ng biomass and hence sequestration potential, that aspect wouldrovide an underestimate rather than overestimate, i.e. it wouldost likely provide a conservative estimate of possible sequestra-

ion.

FPC and NCAS biomass values were compared for selected,

emnant, regional ecosystems, because in those locations extantiomass should have equalled potential biomass. To reliably com-are the NCAS and FPC biomass layers the following were excluded:

nd Environment 148 (2012) 44– 64 47

areas deforested since 2002, mixed ecosystem types, and polygonssmaller than 2 ha (to avoid edge effects). Also, a 100 m interiorbuffer for all ecosystem polygons was subtracted to accommodatethe error margin of ±25 m in the FPC layer. Those exclusions left506 remnant ecosystem polygons for comparison between biomassderived from FPC and NCAS. There was substantial variation forindividual polygons but good agreement at the State-level. Theaverage ratio of NCAS/FPC State-wide was 1.11 (SD 0.56). The dif-ference from a ratio of 1 was not significant and therefore biomassderived from FPC and NCAS was treated as equivalent. The potentialbiomass of all regrowth forest polygons, State-wide, was subse-quently obtained from the NCAS layer.

Biomass growth typically follows a logistic curve, but param-eterisation of such a curve requires knowledge of when growthbegan, which was not available for much of the regrowth. Alter-natively, the growth until about 85% maturity can be roughlyapproximated as linear (e.g. Fensham and Guymer, 2009). Completerecovery of regrowth has been estimated to require 70 yr (Fenshamand Guymer, 2009), whereas 85% of maximum biomass for woody-thickening has been estimated at 140 yr (Hibbard et al., 2003). Asa compromise between those two studies we assumed that 85% ofmature biomass would be achieved in 100 yr. Faster growth ratesare likely where live root stock remains belowground after defor-estation (e.g. lignotubers), but slower growth is likely from seed,especially on degraded land. The biomass present after 50 yr wascalculated assuming a linear growth rate. Where regrowth had beenre-cleared, the most recent deforestation event was used as thestarting point. An approximate, linear sequestration rate for QLDwas calculated from the difference between the extant biomass andthat in 50 yr time for all regrowth polygons; those differences weresummed and the total divided by 50 to get annual sequestrationflux.

3. Results

3.1. Nationwide net �C to-date

Approximately 50% of Australia’s potential forest biomass (fromthe NCAS layer) was in the commercial rangelands, approximately50% of the rangelands were potentially forested (Table 2a), and∼70% of the rangeland biomass (both extant and pre-commercial)was in the potentially forested rangelands (Fig. 2, Table 1). Thus thepotentially forested areas have the capacity to hold the majority ofC in biomass and correspondingly, the majority of SOC.

From the B&R layers resampled to 1000 m pixels the C in poten-tial biomass was 6.50 Pg. The potential biomass from the NCASlayer was 6.59 Pg (Fig. 2). Thus the two values for potential biomassdiffered by only ∼2%.

The NCAS layer had approximately 20 times the spatial resolu-tion of the B&R layer and arguably had more accurate pixel values.On close examination the NCAS layer had not differentiated muchof the biomass of the Mitchell Grass downs (Astrebla spp.) of about32 Mha in northern Australia from that of nearby forest. Biomassfor those two vegetation types was however differentiated in theB&R layers. Thus when the ratio of extant/pre-commercial from theB&R layers was applied to the NCAS layer the calculated emissionfrom the grasslands was higher than their pre-commercial stocks.Consequently, the NCAS layer was not further used in combinationwith the B&R layers. The high potential in the NCAS layer for thegrasslands may have arisen due to the layer’s independence of anenvironmental characteristic that gave preference for grass rather

than tree cover (e.g. clay expansion/contraction or historical fireregime).

The value used for C in potential biomass in the commercialrangelands was therefore taken from the B&R layers: 6.5 (±0.5) Pg.

48 C. Dean et al. / Agriculture, Ecosystems and Environment 148 (2012) 44– 64

Table 1Areas, carbon in potential biomass (derived from Richards and Brack, 2004) and deforestation (derived from DEWR, 2007) for Australian forests and Australian commercialrangelands.

Land-use/cover Area (Mha) C in potential biomass (Pg) % of rangelands % of forests % of Australia

Area Carbon Area Carbon Area Carbon

Rangeland 369 6.593 100 100 49 33Potential forest in rangeland 195 4.595 53 71 46 28 26 23Potential forest in Australia 426 16.15 100 100 57 80

TpA

a1

sTrdctgaflrshi

tthwa

pb0o

F(Qi

Deforested Australia 95.2 5.897

Deforested rangeland 21.9 1.050

Australia 752 20.24

hat corresponds to approximately one quarter of Australia’s totalotential C in biomass (of 23.6 Pg), from approximately 50% ofustralia’s land area.

From the deforestation estimate using the national vegetationssessment (DEWR, 2007) and the NCAS layer, Australia had lost.05 Pg of C, being 23% of its forested rangeland biomass (Table 1).

From the B&R layers resampled to 1000 m pixels, the extanttock of C in biomass on the commercial rangelands was 6.08 Pg.hus a net 0.418 Pg had been emitted. But from the B&R layersesampled to 250 m resolution the loss was only 0.414 Pg. Thatifference of ∼1% was typical of error margins involved in GIS cal-ulations using the national data sets. They were in part due torimming of pixels when converting from large pixels (∼5.5 km) ineographic coordinates (degrees) to projected coordinates (metres)nd then overlapping with GIS layers originally derived in polygonorm in projected coordinates. From close examination of the B&Rayers we estimated the errors in total extant biomass stock for theangelands to be ∼15%, but up to ∼50% for individual sites such asalt lake edges and coastal tidal zones. Those latter areas probablyad high error margins due to problems of capturing variation in

nter-annual or seasonal growth by remote-sensing.The emission calculated from the B&R layers was a net value and

herefore included all effects such as land degradation, deforesta-ion, regrowth, woody-thickening and CO2 fertilisation. Australiaad lost 6% of its rangeland C in biomass due to commercial grazing,ith 0.345 Pg (83%) of the loss from potentially forested rangeland

nd 0.069 Pg from never-forested rangeland (Table 2b, Fig. 3).An average of the two methods of biomass loss—(a) NCAS-

otential plus deforestation from DEWR (2007), and (b) differenceetween the B&R layers: (1.05 and 0.414 Pg respectively)—gave.73 (±0.4) Pg for C efflux from biomass due to commercialisationf the rangelands.

ig. 3. Change in C in biomass since onset of commercial grazing in the rangelandsderived from Berry and Roderick, 2006). Note that the major losses are from easternLD, and that deforestation is most often incomplete at the pixel level (0.05 × 0.05◦ ,

.e. ∼5.5 km × 5.5 km).

22 −37 13 −2911 −23 5 −7 3 −5

100 100

The highest State or Territory biomass losses following com-mercialisation of the rangelands were from QLD, from the mostlydeforested eastern section (Tables 2a and 2b, Fig. 3). QLD held morerangeland biomass than all the other States and Territories com-bined and 53% of that QLD biomass was on freehold land (in 27% ofthe area of QLD rangelands), which contained more biomass thanany of the other States individually.

From the modelling of soil erosion the estimate of SOC range-land losses to-date was 0.086 (±0.022) Pg and from deforestationto-date the SOC loss was 0.07 (±0.04) Pg—a total SOC efflux of 0.16(±0.05) Pg. We found qualitatively that erosion appeared more pro-nounced in overgrazed areas and near tracks, (both vehicle andlivestock tracks), and especially pronounced near where trackscrossed riparian zones, extending upstream from the crossing. Asa qualitative observation, the consequent erosion in riparian zonescaused the highest, localised negative impact on trees and on soilstability (Fig. 4).

Of 602 randomly selected watering points across ∼6 Mha in thesouthern Northern Territory, 31% were within 100 m of a creek and68% were within 500 m of a creek (Fig. 5). Thus there appearedto have been a distinct preference for managers to locate livestockwatering points near creeks. Also, natural watering holes are inher-ently centred in riparian zones. The majority of herbaceous feed indryer times is in riparian zones, which naturally attracts higher con-centrations of livestock. Qualitatively, that spatial concentration ofimpact was more prominent during dry periods prior to rain, andespecially during droughts. Throughout most of semiarid Australia,where woody biomass was sparse, the biomass was often higher inriparian zones, due to topographic flow (e.g. Figs. 6 and 7). Both thelocation of watering points and the concentration of available feedcontributed to high riparian-zone impact.

In several experimental [livestock] exclosures established in the1960s in WA we observed qualitatively that buffel grass (Cenchrusciliaris L. (syn. Pennisetum ciliare (L.) Link)) had much higherbiomasses than outside the exclosures, especially under shrubs andtrees. Buffel grass is an African species introduced into Australiafor livestock forage and to reduce soil instability resulting fromovergrazing, following similar introduction into North America.Buffel biomass was also high in riparian areas (Fig. 7) although itwas reduced to some degree by livestock grazing. However, in theabsence of high livestock numbers, and where native macropod(kangaroos and wallabies) numbers were high, the buffel biomassappeared sufficiently high to represent an increased fire risk. Nearwhere grazing properties bordered rural townships that risk wassometimes reduced by prescribed burning, but that activity alsoreduced biomass of some native mature trees.

3.2. Climate change affect on �SOC

The forecast �SOC with climate change for the forested range-lands (Fig. 8, Tables 3a and 3b) showed substantial variability owingto (a) differences between climate models, (b) differences betweenglobal emission scenarios, and (c) areas with low extant stocks

C. Dean et al. / Agriculture, Ecosystems and Environment 148 (2012) 44– 64 49

Fig. 4. Examples of slow riparian biomass- and SOC-loss from overgrazing and tracks, southern Northern Territory. (a) Erosion and stages of gradual biomass decline due toover grazing, combined with a vehicular crossing downstream. Note persistence of buffel grass in lower right photo. (b) Degradation from cattle at natural waterholes. Upperphoto: widespread soil disturbance, no litter cover, no young Gidyea (Acacia cambagei) but colonisation by a less-palatable thorny acacia species (lower right). Lower photo:bank erosion and tree decline around a spring, including degradation from both cattle and feral herbivores.

50 C. Dean et al. / Agriculture, Ecosystems and Environment 148 (2012) 44– 64

Table 2aAreas of potentially forested and never-forested, Australian commercial rangeland (in Mha).

Forest cover status NSW NT QLD SA VIC WA Total Total%

Never-forested 12.74 40.47 62.96 26.69 0.01 31.67 174.5 47Potentially-forested 18.75 29.77 74.03 16.28 0.20 55.42 194.5 53

Total 31.49 70.24 136.99 42.97 0.22 87.09 369.0 100

Table 2bChanges (derived from Berry and Roderick, 2006) in C in biomass (in Tg) since commercialisation of potentially forested and never-forested Australian rangeland.

Forest cover status NSW NT QLD SA VIC WA Total Total%

Never-forested −13.58 −19.67 −17.40 −1.80 0.00 −16.43 −68.89 17Potentially-forested −28.70 −50.61 −223.74

Total −42.28 −70.29 −241.13

F∼

atbfca

TE

TSt

ig. 5. Histogram of number of watering points vs. distance from nearest creek, for6 Mha of the Northern Territory. Note the concentration close to creeks.

ccruing high variability between 2000 and 2100 from the equa-ion of Bird et al. (2002). The largest differences nationally were

etween the models CSIRO-Mk3.5 and Miroc-M, mainly due to dif-erences in forecast rainfall in northeast Australia. The area of mostonsistency between models was the central-west of WA (knowns the Pilbara and Gascoyne), where all models forecast a drop in

able 3axtant (2000) SOC stocks (to 0.3 m), stocks in 2100 and percentage change for the foreste

State In 2000 CSIRO-Mk3.5 ECHAM5

A1B A1FI A1B A1FI

Tg % Tg �(%) Tg �(%) Tg �(%) Tg

NSW 222 8 107 −52 63 −72 145 −35 104

NT 417 14 192 −54 105 −75 270 −35 197

QLD 1523 51 818 −46 544 −64 1005 −34 775

SA 182 6 38 −79 15 −92 99 −46 65

VIC 6.0 0.2 3.1 −49 1.5 −75 4.9 −18 4.3

WA 610 21 174 −71 88 −86 343 −44 226

Total 2961 100 1332 −55 817 −72 1868 −37 1371

able 3btate-wide totals for SOC stocks for 2100 (to 0.3 m) in the forested rangelands. Standard dhe two emission scenarios (Table 3a).

State A1B A1FI

Tg �(%) Tg

NSW 175 −21 151

NT 289 −31 228

QLD 1123 −26 959

SA 114 −37 93

VIC 4.8 −19 4.2

WA 318 −48 207

Total 2020 −32 1642

−3.90 −0.57 −37.25 −344.8 83

−5.70 −0.58 −53.69 −413.7 100

rainfall. Owing to the consistency for that area its average forecastwhen all four models were combined was large, as there was nosuperposition of differently signed fluxes. In areas where differentmodels diverged most, the net average �SOC was closer to zero. SAshowed the highest standard deviation, due to the lower, average,extant SOC inducing more relative variability.

3.3. Annual emissions

The annual emissions calculated were compared with thosefrom the literature (Fig. 9). Nationally, the two-highest singularemissions were biomass loss from deforestation in QLD and nation-wide methane emissions from livestock (the latter is the maindifference between ‘Total2’ and ‘Total1’). Even though �SOC wasa relatively minor flux on an annual basis (comprising −2.3 Tg yr−1

from deforestation and −2.1 Tg yr−1 from erosion) it was half theefflux from biomass from deforestation in QLD and equivalent to3% of Australia’s Kyoto-Protocol annual GHG emissions (totalling149 Tg yr−1 of C-equivalent (DCCEE, 2011)). Furthermore, �SOC

takes longer than biomass emissions accompanying deforestation(due to the multi-centennial half-life of some SOC pools (e.g. Liaoet al., 2006; Dean et al., 2012)) and thereby �SOC will accruea higher total emission in the long-term, than the short-term

d rangelands from different climate models and emission scenarios.

ECHO-G Miroc-M

A1B A1FI A1B A1FI

�(%) Tg �(%) Tg �(%) Tg �(%) Tg �(%)

−53 229 3 226 2 217 −2 211 −5−53 336 −20 291 −30 356 −15 320 −23−49 1342 −12 1273 −16 1325 −13 1244 −18−64 151 −17 133 −27 168 −8 159 −13−29 5.6 −5.8 5.4 −10 5.7 −4.4 5.5 −7.3−63 343 −44 216 −65 402 −34 297 −51

−54 2406 −19 2145 −28 2474 −16 2236 −24

eviations (in brackets) were assigned from the variation across the four models and

Average

�(%) Tg �(%)

−32 163 (66) −27 (11)−45 258 (86) −38 (13)−37 1041 (301) −32 (9)−49 104 (99) −43 (24)−30 4.5 (1.5) −25 (8)−66 261 (104) −57 (23)

−45 1831 (597) −38 (12)

C. Dean et al. / Agriculture, Ecosystems and Environment 148 (2012) 44– 64 51

Fig. 6. Distribution of forest biomass (shown as FPC (foliage projected cover)) inminimally forested rangelands. Biomass is concentrated in riparian zones. FPC values>hC

−lw

brelifltte

FrnpmtCtl

10 are forest. This region is west of the most concentrated deforestation. The upperalf of the map is the Mitchell Grass downs ecosystem type and the lower half is thehannel Country ecosystem type—similar ecosystems to the latter are widespread.

2.3 Tg yr−1. Also, if national destocking was instantaneous theand degradation would not cease immediately and SOC efflux

ould continue. Thus, the SOC efflux is one of the major effects.From the approximation of nationwide rangeland soil reha-

ilitation (Fig. 10) if erosion was allowed to continue whileehabilitation that started in 2011 was ongoing, then the total SOCmission (to 0.3 m depth) due to commercialisation of the range-ands would be in the order of 1.3 (±0.5) Pg, peaking in 2400,ncluding 1.2 (±0.5) Pg after 2011. The error margins for those SOC

ux estimates are high (possibly around 90%) because the calcula-ion was based on single values for the national SOC loss by erosion,he percentage of C in rangeland soil, and the loss rate of SOC fromroded soil.

ig. 7. Distribution of forest biomass relative to riparian zones in minimally forestedangelands. Upper: biomass distribution typical of ecosystems similar to the ‘Chan-el Country’. The lighter-green trees are in moister riparian zones (top centre), theale green shade on the soil surface in the riparian zone (e.g.: arrow) is buffel grass,edium green are the eucalypts, and the darker grey-green are acacias. Lower:

ypical riparian buffel infestation in commercial grazing land of ‘Channel Country’.onsiderable high dry biomass level, during drought period, but with green tips dueo recent light showers. (For interpretation of the references to colour in this figureegend, the reader is referred to the web version of the article).

Fig. 8. Change in SOC to 0.3 m with climate change, for the potentially forested com-

mercial rangelands between 2000 and 2100. Average from the four climate changemodels and two emission scenarios. (a) Percentage change. (b) Standard deviation(as a percentage) of the average.

If rehabilitation began in 2011 and all erosion was curtailed by

2050, then the SOC efflux since onset of rangeland commercialisa-tion would be limited to 0.15 (±0.14) Pg, and 0.062 (±0.056) Pg from2011 to 2050. Thus, a SOC emission of 1.2 − 0.062 = 1.16 (±0.5) Pg

Fig. 9. Estimates of annual emissions from Australian rangelands. Lighter columnsrepresent only some component pools, and darker columns represent all pools.Values from the present work are the right hand four columns.

52 C. Dean et al. / Agriculture, Ecosystems and Environment 148 (2012) 44– 64

Fig. 10. SOC for the whole Australian commercial rangeland (to 0.3 m) with current erosion and future rehabilitation. Emissions are much less if erosion is arrested in 2050than if erosion is allowed to continue concomitant with rehabilitation. Recovery time is in the order of 2000 yr owing to the inherently slow rate of SOC sequestration.

Ffy

ca

3

i

TCl

ig. 11. Forest change in southeast QLD’s commercial rangelands. Upper photo:resh broad-scale deforestation, fraction of potential present is zero. Lower photo:oung regrowth forest amongst multi-event, patchy clearing.

ould be avoided by initiating rehabilitation in 2011 and ceasingnthropogenic erosion by 2050.

.4. Sequestration in QLD regrowth

The area of regrowth forest (Figs. 11 and 12, Tables 4a and 4b)n the commercial rangelands of QLD was 22.7 Mha., i.e. 29% by

able 4aarbon sequestration capacity in biomass in regrowth forests on commercial range-

and in QLD, partitioned into areas in terms of percentage of potential present.

Extant biomass (% of potential) Area (Mha) Emitted carbon = potential− extant (Tg)

0 ≤ 10 2.1526 127.2810 ≤ 20 9.3213 398.3020 ≤ 30 4.5299 139.5530 ≤ 40 3.3501 73.9640 ≤ 50 2.1486 40.0450 ≤ 60 0.5610 10.1360 ≤ 70 0.2187 2.8770 ≤ 80 0.1341 1.4180 ≤ 90 0.1046 0.7090–100 0.0484 0.37

Total 22.57 794.6

Fig. 12. Map of potential C sequestration in QLD regrowth forest.

area of QLD’s potentially forested rangelands had potential tosequester C via forest regrowth; which corresponds to 16% of allQLD’s rangelands. The largest regrowth cohort (by area) was thatwith only 10–20% regrowth of potential biomass present (Table 4a).

Table 4bCarbon sequestration capacity in biomass in regrowth forests on commercial range-land in QLD, and growth rate.

Extant 0.179 (±0.054) PgPotential 0.974 (±0.292) PgEmitted = potential − extant = what

can be sequestered0.794 (±0.397) Pg

Sequestration over next 50 yr 0.410 (±0.205) PgSequestration rate (next 50 yr) 8.2 (±4.1) Tg yr−1

Sequestration flux (next 50 yr) 0.36 (±0.18) Mg ha−1 yr−1

ems a

Dlr

itrtdpeAl

lsw(mewt

ifo8c

0geh0ibtNttb

4

4

rftmrpiF(Sob5

o(lbo

C. Dean et al. / Agriculture, Ecosyst

eforestation in QLD to date corresponded to ∼92% of range-and deforestation nationwide and was therefore representative ofangeland deforestation nationally.

The highest potential was consistently in the higher rainfall zonen eastern QLD (Fig. 12). The spatial distribution of potential seques-ration was more dependent on the clearing activity than on theainfall gradient, because eastern QLD generally had a lower frac-ion of its potential present. That spatial distribution may have beenue to more re-clearing events in the east, accrued over a longereriod of usage or more intensive usage in the east. Average prop-rty size was smaller in the higher rainfall areas generally acrossustralia, which may have been an indicator of the intensity of

and-use.From the employed approximations the errors for QLD range-

and regrowth were ∼30% for extant stocks and ∼50% forequestration rates. The extant C in biomass of the regrowthas 0.179 (±0.054) Pg and it could reach a maximum of 0.974

±0.292) Pg: the average regrowth forest was at ∼30% of itsaximum biomass. The net C efflux from biomass from QLD defor-

station to date was therefore 0.794 (±0.397) Pg (Table 4b), whichas the amount that theoretically could be recovered as seques-

ration.The C sequestration rate (with the linear growth rate approx-

mation) was 0.36 (±18) Mg ha−1 yr−1. If all the extant regrowthorest was allowed to mature then the C sequestered in biomassver 50 yr would be 0.410 (±0.205) Pg, i.e. a sequestration rate of.2 (±4.1) Tg yr−1. That rate assumes no detriment to growth fromlimate change effects, such as increased water deficit or fire risk.

The regrowth calculations above assumed a root:shoot ratio of.15. The regional allometrics from Henry et al. (2002) for above-round and belowground biomass, when applied to the remnantcosystems, yielded an average root:shoot ratio of 0.33. Using thatigher ratio gave a net C efflux from past QLD deforestation of.939 (±0.469) Pg, i.e. 18% higher than 0.794 (±0.397) Pg. However

f 0.33 was used as the root:shoot ratio then the assumption of NCASiomass being equal to FPC biomass was no longer valid, becausehe NCAS/FPC biomass ratio of 1.11 became 1.28, indicating thatCAS biomass values may be too high. The root:shoot ratio was

hus a major cause of uncertainty and it was reasonable to retainhe initial estimate of the C that could be sequestered in biomass,y maturation of regrowth, at 0.794 (±0.397) Pg.

. Discussion

.1. Emissions

Under commercial grazing and climate change the forestedangelands provide major rangeland C fluxes. The effect on C stocksrom commercial grazing in the forested rangelands is similar tohe global trend for ‘woodland’ type rangeland (Ellis, 2011). Four

ain avenues to reduce C emissions from rangelands have beenecommended (Henry et al., 2002): (a) reduced deforestation, (b)rotection and enhancement of SOC, (c) reforestation, and (d)

ncreasing the scientific knowledge involved in these processes.or the first three avenues we found that: (a) an emission of 10±3) Tg yr−1 could be avoided, (b) an emission of 1.53 (±0.5) Pg ofOC could be avoided by rangeland rehabilitation and cessationf erosion, and (c) 0.794 (±0.397) Pg of C could be sequestered iniomass in QLD regrowth (or 8.2 (±4.1) Tg yr−1 of C over the next0 yr). The fourth avenue is discussed in more detail in Section 4.4.

The discrepancy between the C in biomass emission estimatef 0.414 Pg from all commercial rangeland using the B&R layers

Table 2b) and the 1.05 Pg from deforestation alone using the NCASayer and DEWR (2007) (Table 1) was most likely due to a com-ination of factors: (a) deforestation being incomplete within anyne 0.05 × 0.05◦ pixel; (b) NDVI (normalised vegetation index) of

nd Environment 148 (2012) 44– 64 53

young regrowth or its grass replacement being similar to or greaterthan the NDVI of mature forest; and (c) possible overestimationof CO2 fertilisation in Berry and Roderick (2006). The B&R layersinherently included biomass loss due to, for example, overgrazing,fence-line clearing, laneway clearing, firewood collection, cuttingfor fodder and small-scale illegitimate felling; and biomass gaindue to thickening (although this has been shown to be minimal onnet for Australia (Witt et al., 2006; Fensham et al., 2011)); whereasthe method of determining biomass change form the DEWR (2007)and NCAS layers referred more to broad-scale deforestation alone.Additionally, if a root:shoot ratio of 0.333 had been used in place of0.15 for the deforestation estimate the emission would be 1.22 Pgrather than 1.05 Pg. The magnitude of the loss of C in biomass (0.73(±0.40) Pg) due to rangeland commercialisation can be comparedwith the nationwide loss of 3.3 Pg of C (from Berry and Roderick,2006) of which the major component was deforestation of arableland (with higher average biomass) for intensive farming, forestryand infrastructure.

Higher precision data, in the form of regional allometrics and25 m pixels, yielded a C emission from biomass due to deforesta-tion in QLD of 0.79 (±0.40) Pg (Tables 4a and 4b), which was higherthan the nationwide rangeland emission [from all sources] of 0.73(±0.4) Pg calculated from lower precision data. Furthermore, fromthe higher precision calculations, the average sequestration poten-tial for QLD regrowth forest was 35 Mg ha−1, but from the B&Rlayers that value was 3.4 Mg ha−1; i.e. nearly an order of magnitudelower. Those comparisons suggest that the value for national range-land biomass loss using the B&R layers was too low, and thereforethat the calculated value of 1.05 Pg of C emitted to date from defor-estation (Table 1) was more accurate. As neither could be verifiedfurther with the available data, there was no reason to nominate analternative value for C emission from biomass due to deforestationin QLD, other than the 0.73 (±0.40) Pg.

The net C efflux that accompanies erosion, due to commercial-isation of the rangelands, is dependent on: (a) the net balancebetween rates of loss and soil formation (pedogenesis), (b) thecontribution of biomass growth to soil carbon formation with andwithout commercialisation, and (c) erosion rates with and withoutcommercialisation.

Determining the fraction of rangeland erosion that is natu-ral and that which is anthropogenic is not straightforward (e.g.Dregne, 1995). For the erosion estimate in the absence of com-mercialisation we used the soil loss rate from minimally disturbedsites (1 Mg ha−1 yr−1), which is similar to the maximum rate ofsoil formation (pedogenesis) (Loughran et al., 2004), i.e. pedoge-nesis may just keep up with natural erosion. That gives the portionof rangeland erosion (of the 5.6 Mg ha−1 yr−1 nationwide) that isanthropogenic as 80%.

Evidence for soil loss due to commercialisation is from:(a) historical, qualitative reports; (b) current fence-line con-trasts in paddock soil condition; (c) point-source erosion fromanthropogenic activity; and (d) quantification of broad-scaleanthropogenic contributions to wind erosion.

Degradation of Australia’s rangelands due to grazing prac-tices has occurred since the onset of commercialisation in the1850s, when overstocking was widely practised (Wheeler andHutchinson, 1973; Harrington et al., 1979; Williams and Calaby,1985; Barnes, 1997; McKeon et al., 2004). After a decade ofcommercialisation there was substantial erosion, loss of perennialgrasses and saltbush, shrub removal and salinity; and within30 yr there was woody shrub proliferation and scalded plains(Barnes, 1997; Scott, 2001). Severe erosion continued for several

decades. During droughts, denudation and erosion prevailed nearwatering points, and vegetation further declined with topsoil lostin the wind. Stocking rates were maximised until negative effectswere detected, and reduced when short-term economics became

5 ems a

uiBsvIt1oamdbiiu

awaiar(ppo“ht

bbdS2atM

ppeTisnstimc

ptrcerba

opia

4 C. Dean et al. / Agriculture, Ecosyst

nsustainable (Barnes, 1997; James et al., 1999). Degradation wasnitially focussed around natural watering points (Little, 1964;eard, 1990) but became more diffuse with bore usage, financialetbacks of overgrazing (with regrowth on defunct properties),ehicular tracks, shrub cutting for fodder, and deforestation.mprovements in land condition were evident from the mid twen-ieth century, after the introduction of stocking limits (e.g. Barnes,997; Watson et al., 2007). Today amelioration of erosion is reliantn the reporting of the most prevalent degradation effects, whichre qualitatively assessed (on-ground and by remote sensing),ostly from long-term changes in perennial grass cover or in shrub

emographics (e.g. Dean, 2005; Watson et al., 2007). Soil carbon oriomass carbon stocks per se are unmonitored, other than change

n biomass carbon in QLD from broad-scale deforestation. Changen soil carbon due to deforestation is longer-term and currentlynmonitored during routine inspections, nationwide.

Contemporary assessment suggests that vegetation degradationnd erosion affects over 55% of Australian commercial rangeland,ith portions for the different States being 100%, 80%, 50%, 40%

nd 40% for NSW, SA, QLD, WA and NT respectively (Woods, 1984),ncluding for example on 85.3 Mha within western QLD (Dawsonnd Turner, 1982). The rate of soil loss is estimated to be cur-ently greater than soil formation in 70% of Australia’s rangelandsLoughran et al., 2004), and >1 Mg ha−1 yr−1 (the upper limit ofedogenesis) in over 84% of WA, with highest rates in regions whereedogenesis is low (George, 2001). The most quantitatively rig-rous pre/post-commercialisation comparison for Australia is theaccelerated erosion index”, which showed that wind erosion isigher with commercial rangeland grazing in the over half of QLDhat is naturally susceptible to dust storms (Leys et al., 2001).

Ground cover (including plant and tree litter, grass, and micro-iotic crusts) stabilises soil against water and wind erosion, and cane used as a surrogate of soil condition, with SOC efflux potentiallyoubled in its absence (Perry, 1977; Eldridge and Greene, 1994;hao et al., 1994; Roose and Barthès, 2006; Zuazo and Pleguezuelo,008). Soil erosion increases with decreased cover levels, suchs through overgrazing, and cover levels >10–40% are necessaryo prevent increased soil movement (Miles and McTainsh, 1994;

cIvor et al., 1995).Annual water runoff is at least doubled with conversion to exotic

asture after deforestation (with generally higher runoff in heavyluvial events), and consequently erosion is also increased (Scanlant al., 1991; Thornton et al., 2007; Zuazo and Pleguezuelo, 2008).he frequency and intensity of heavy pluvial events is forecast toncrease with climate change—a trend evident in central Australiaince ca. 1970 (Pickup, 1998; Gallant et al., 2007; Pittock, 2009). Theumber of deforested rangeland sites measured in the continentaltudy by Loughran et al. (2004), under-represented the deforesta-ion in QLD, NSW and NT and therefore that estimate of erosions possibly low for Australian rangelands as a whole. That would

ake our estimate of SOC loss with rangeland erosion most likelyonservative.

Eroded soil is either carried by wind or water, with these vectorsrevailing to different degrees. In each case there are three loca-ions relevant to quantifying SOC flux: the erosion site, the transitoute and the depositional site. Although the fate of SOC eroded byommercial rangeland grazing has been widely studied (e.g. Butlert al., 2003; Lal, 2004; Paull et al., 2006; Polyakov and Lal, 2008) itemains a topic of current scientific research, with a major concerneing whether it represents a net loss to the rangeland system, or

redistribution.Erosion from overgrazing, deforestation and various methods

f attrition of woody vegetation incur a range of changes in soilroperties which negatively affect biomass production (examples

n: Christian and Perry, 1969; Harrington et al., 1979; Dawsonnd Turner, 1982; Murrell, 1984; Pickup and Chewings, 1986;

nd Environment 148 (2012) 44– 64

Pickup, 1990; Chartres et al., 1992; McMahon et al., 1992; Sparrowet al., 1997; Kinloch and Friedel, 2005). Conversely, impacts severeenough to limit microbial community dynamics can limit soil respi-ration (limiting SOC efflux) (Grandy and Robertson, 2007), althoughunder these circumstances, any emitted SOC is not replenished, dueto an absence of biomass.

Biomass and nutrients are lowered in eroded areas (unlesswoody plants can re-colonise) and biomass can be higher in depo-sitional areas where nutrients accumulate. However the nutrientsand water in the deposition zone can become decoupled from veg-etation, with net nutrient loss, and the biomass on the relativelysmall deposition zone does not balance that lost from the erodedarea (Sparrow et al., 1997, 2003; Tongway et al., 2003). A net nutri-ent loss involves C efflux (for organic nutrients).

The fate of the transported carbon will depend on oxygen avail-ability (such as though physical agitation, moisture and oxidants)and temperature. Eroded SOC in sediments in aquatic systems con-tributes to the net carbon stock (Jacinthe et al., 2001; McCarty andRitchie, 2002; Smith et al., 2005; Paull et al., 2006; Smittenberget al., 2006). Emissions therein increase with increasing tempera-ture, (with rates being substrate dependent) and within microbialtolerance limits (Weston and Joye, 2005; Doering et al., 2011).Anoxic, waterlogged soils and lower temperatures, decrease therate of organic decomposition (Trumbore, 1997; McCarty andRitchie, 2002; Stolbovoi, 2002; McIntyre et al., 2009). Salt lakesand wetlands Australia-wide, on commercial rangeland propertiesand conservation reserves downstream of commercial rangelands,are examples of such anoxic sites. Other catchments drain seawardthrough major rivers.

There is preferential erosion of the surface soil and organic frac-tions of soils (which are lighter), and of microbial carbon, and thusthe translocated carbon may be more labile than that which is noteroded (Gregorich et al., 1998; Mora et al., 2007). In that situation,if the deposited SOC has no additional protection mechanism, it canhave two to four times the efflux than of the eroded site (Mora et al.,2007). Translocated SOC can incur fast initial losses, e.g. 15% within0.27 yr, at least ∼12% emitted in transit and with higher losses forlonger transit times, and less than half the translocated SOC mayreach riparian transport (Polyakov and Lal, 2008). Carbon losses perunit mass are more from the larger particles than from finer frac-tions due to scission of the larger aggregates, preferentially wherethe carbon is located and the scission is achieved by flow-relatedshear stresses and raindrop impact, providing surfaces for microbialattack (ibid).

Jacinthe et al. (2002) found SOC losses of 29–35% from erodedsoil, with nearly half of that lost within 0.055 yr. Lal (2001) andJacinthe et al. (2001) estimated that 20% of translocated SOC is emit-ted as CO2. For net SOC efflux from soil loss, that value of 20% (whichwe used) would thus appear to be conservative (Lal et al., 2004) andtherefore appropriately used in this study.

Our calculation of a SOC efflux of 1.7 (±0.9) Tg yr−1 owing to netsoil loss from erosion for the commercial rangelands, based on thesoil erosion work of Loughran et al. (2004), does not include SOCemitted with erosion where the translocation is only within singlestudy sites (which ranged from 400 to 3500 m), i.e. where there wassoil movement but no net soil loss—those emissions are additional.This constitutes another reason for our calculated SOC efflux witherosion to be considered a conservative value.

Similar calculations covering similar expanses and climateselsewhere are scarce. However, for Rajasthan (India) in the200–500 mm rainfall zone. Singh et al. (2007) found continuallydeclining SOC due principally to land management effects. From

1972 to 2002 their SOC efflux averaged 0.49% yr−1 to 1 m in thearid region, ranging from 0.14 to 1.1% yr−1, dependent on soiltype. Assuming an intermediate emission of 0.6% yr−1, and apply-ing that to the most degraded of Australia’s rangeland (i.e. 5% of the

ems a

3rlt

taPricGS

waeatctTr

4

4

al0Getcod(eao

rttlpsr

‘edtom(oaldwa2

c

C. Dean et al. / Agriculture, Ecosyst

69 Mha) then the annual C efflux from the top 0.3 m of Australianangeland is 11 Tg yr−1. That value is about six times our calcu-ated 1.7 Tg yr−1, but within expected error margins consideringhe approximations used.

The calculated annual C emissions (‘Total1’ and ‘Total2’) fromhe commercial rangelands of 20 (±7) Tg yr−1 and 32 (±10) Tg yr−1

re equivalent to 13 (±4)% and 21 (±6)% of Australia’s Kyoto-rotocol annual GHG emissions, i.e. approximately identical to theeported Agriculture sectors’ emission (DCCEE, 2011). ‘Total2’ isdentical to that of Robertson (2003), although he included a highomponent from deforestation (20.5 Tg yr−1 from the 1999 nationalHG inventory), which was replaced in our tally by the efflux fromOC and mulga lands grazing (not included in his estimate).

Australian national GHG emissions increase annually, and heree used the 2011 value of 546 Tg of CO2-e (DCCEE, 2011). They

re currently reported according to Kyoto Protocol rules, how-ver for the land sector this only includes agricultural emissionsnd net emissions from reforestation, afforestation and deforesta-ion. Additionally �SOC is only included when the party concernedhooses to include Tier-3 carbon fluxes. Australia has elected noto include Kyoto Article 3.4 land use activities in its Kyoto account.hus, several of the emissions calculated here are not currentlyeported and hence several major fluxes are omitted.

.2. Sequestration

.2.1. Sequestration in biomassMoore et al. (2001) proposed that rehabilitation (by destocking

nd fire suppression) of the mulga lands (Acacia anuera shrub-and, often around 3 m high, occupying 170 Mha) could sequester.273 Mg ha−1 yr−1 in biomass and necromass carbon over ∼60 yr.arnaut (2008) proposed C sequestration at 68 Tg yr−1 over “sev-ral decades” for “restoration” of the mulga lands, correspondingo 0.4 Mg ha−1 yr−1. From exclosure studies, Witt et al. (2011)alculated 0.22 (0.02) Mg ha−1 yr−1 in biomass for rehabilitationf the mulga lands. Those three rates are comparable with thaterived for mesquite (Prosopis spp.) thickening in Texas: 0.52±0.12) Mg ha−1 yr−1 in SOC and aboveground vegetation (Archert al., 2004). Afforestation of rangeland through technology toccess groundwater under hardpan can achieve net sequestrationf at least 1.1 Mg ha−1 yr−1 (Yamada et al., 2003).

Our C sequestration flux of 0.36 (±0.18) Mg ha−1 yr−1 in QLDegrowth biomass is higher than for rangeland rehabilitation andhe exclosure studies, but lower than for thickening, and closest tohe Garnaut (2008) estimate. Juvenile regrowth may have mature,ive root stock from before deforestation (e.g. brigalow (Acacia har-ophylla) has lignotubers). Therefore the growth rate for regrowthhould be faster than where colonisation (or re-colonisation) isequired, such as for thickening or rehabilitation.

Arrest of further deforestation and re-clearing has been termedforgone development’ and a monetary cost allocated (Davidsont al., 2006). However the cost of C efflux accompanying furthereforestation was not included. Also, prior to the high deforesta-ion rates of the 1990s in Queensland, avoidance of deforestationf eucalypt forest was recommended by the Queensland Depart-ent of Primary Industries as it precipitates woody thickening

Turner and Beeston, 1978; Pressland, 1984). Although the purposef deforestation is to provide improved grazing pasture, the benefitsre not always perpetual; for example: (a) higher productivity onlyasting for a few decades; (b) phosphorous levels decreasing to pre-eforestation levels within 10 yr; and (c) some nutrients increasinghile others decreasing, and sodicity increasing, due to reduced

ctivity of deep-set tree roots (Sangha et al., 2005; Kaur et al., 2006,007; Radford et al., 2007).

Raison et al. (2009) estimated that a C emission of 1.9 Tg yr−1

ould be avoided by not re-clearing regrowth on freehold land in

nd Environment 148 (2012) 44– 64 55

QLD. Moreover, allowing regrowth to mature could make a signifi-cant contribution to Australia’s climate change mitigation effort:the 8.2 (±4.1) Tg yr−1 (calculated here) would offset either: (a)Australia’s fugitive emissions, or (b) ∼50% of Australia’s agricul-ture emissions, reported in the national inventory (as of mid 2011,DCCEE, 2011). There would also be biodiversity, soil rehabilita-tion and downstream water quality benefits. Recovery of forestcover, through managed reforestation, natural regrowth, or woody-thickening on degraded land, is however, likely to reduce pasturecover (at least for several decades) because tree basal area is neg-atively correlated with herbaceous cover (Beale, 1973). (Althoughsome tree and shrub cover is beneficial for grass growth on more-xeric sandy soils (Daly and Hodgkinson, 1996)). A middle-groundscenario with proportions of both herbage cover and tree covercan provide a compromise. Clumped trees (rather than isolatedtrees) and spatial variability in tree cover (for a given basal area) isbeneficial as it can greatly increase soil nutrients, SOC and pastureproduction (Scanlan, 2002; Eldridge and Wong, 2005). In terms ofbiomass carbon, optimisation is also possible through cover by afew mature paddock trees, which can have high biomass but lesscanopy cover than numerous saplings with high canopy cover (forthe same basal area); similar to the effect of shrub cover limitinggrass growth more than does tree cover (Daly and Hodgkinson,1996). However, this balance of mature trees and grass requiresseveral decades for self-thinning and maturity of the regrowth (orwoody thickening), which can be stymied under poor environ-mental conditions and by browsing. Meanwhile, reduced pastureproduction would prevail. Also, once an exotic grass is established,especially in nutrient-rich sites, then initiating reforestation maybe problematic (Semple and Koen, 2003; Prober et al., 2011).

4.2.2. Sequestration in soilFor rangeland rehabilitation by destocking, based on aver-

age literature values of sequestration in soils, Gifford and McIvor(2009) estimated SOC sequestration flux at 0.07 Mg ha−1 yr−1 and0.14 Mg ha−1 yr−1 for degraded mulga lands and all Australianrangeland, respectively (with an error margin of a factor of10). That sequestration was to occur over 40 yr. From exclosureexperiments Witt et al. (2011) calculated SOC sequestration of0.049 Mg ha−1 yr−1 to 0.3 m, for rehabilitation of the mulga lands.From modelling destocking Harper et al. (2007) estimated SOCsequestration at 0.0091 (±0.0061) Mg ha−1 yr−1 over 5 yr. Thosestudies did not include landscape-scale effects such as ongoingriparian erosion (including streambank erosion, gullying, sheet ero-sion and grass removal, exacerbated by periods of drought (Condon,1972)), and therefore did not represent the net SOC flux at thelandscape-scale. The estimate of Harper et al. (2007) was mostlikely lower as it was a very short-term flux and it may haveincreased with time, following a logistic curve.

The slower flux of SOC arises from the longer average half-life of SOC than that of dead biomass. Significant contributions toSOC come from coarse and fine root turnover (Lorenz et al., 2007;Kirschbaum et al., 2008)—these are not replenished after decom-position following deforestation. Due to the long half-lives of someSOC pools, for SOC to reach potential at the site-level (i.e. maximumSOC storage), several centuries are required (Hibbard et al., 2003)with SOC accruing over successive cycles of tree maturation andsenescence (Dean et al., 2011). The SOC sequestration componentof rehabilitation modelled here used the time-dependent outputfrom the modelling of woody-thickening by Hibbard et al. (2003),which had a logistic curve (starting slowly, faster in mid-range and

slowing towards 2000 yr). Their average rate for SOC sequestrationamongst woody vegetation was 0.15 (±0.04) Mg ha−1 yr−1, which ismost similar to the Australian rangeland rate of Gifford and McIvor(2009).

5 ems a

vbUe4ftantc

bccndbt(Sa

4

auictsiAIam

toswmf

n

aeasrfrsaaRssr

6 C. Dean et al. / Agriculture, Ecosyst

Both emission (from erosion) and sequestration (from woodyegetation) were included in our calculation of �SOC with reha-ilitation, therefore yielding a time-variable rate for net SOC flux.pon destocking it is unlikely that instances of severe riparianrosion would be curtailed without active intervention (see Section.4) and aeolian erosion may only cease after seed germinationollowing rain, possibly combined with gradual litter accumula-ion. Thus, after onset of rehabilitation nation-wide, both emissionnd sequestration would overlap. In that situation the calculatedet effect of a reduced C emission down to 0.36 (±0.32) Pg, ratherhan sequestration per se, is the flux relevant to national climatehange mitigation efforts.

Disturbances that cause tree death can liberate otherwise sta-le SOC, through the ‘priming effect’ thereby producing additionalarbon efflux (Dijkstra and Cheng, 2007; Fontaine et al., 2007). Theombination of influences means that the SOC stocks can declineotably through frequent deforestation or timber harvesting. SOCynamics are complicated and �SOC accompanying change iniomass requires comprehensive experimentation, especially closeo mature trees (e.g. Throop and Archer, 2008) or to deep levelse.g. Young et al., 2005), where deep-rooted plants contribute toOC from root-turnover, and to where decomposition residues mayccumulate through eluviation in coarse-textured soils).

.3. Climate change effects

The often-perceived paradigm of an increase in biomass withnthropogenic global change (including climate change, land-se change and atmospheric pollution) is based mostly on: (a)

ncreased CO2 causing faster photosynthesis, (b) increased CO2ausing increased water use efficiency, (c) possible nitrogen fer-ilisation from pollution, and (d) in some locations, longer growingeason due to warming. There is suggestion that the anthropogenicncrease in CO2 has, at least until recently, increased biomass inustralia (e.g. Howden et al., 2001; Berry and Roderick, 2006).

ncreased biomass in the Australian rangelands due to the CO2nd N fertilisation effects of further global change has also beenodelled (e.g. Carter et al., 2004), but not measured.The balanced chemical equation below represents the forma-

ion and disassociation of cellulose (the main constituent by weightf terrestrial vegetative biomass). The equation, with an equilibriaign to indicate that it goes in forwards and reverse directions, isritten without showing the intermediaries (such as glucose, soilicrobial pathways, and methane under anaerobic decomposition)

or visualisation purposes.

6CO2 + n5H2O

+ EnergyPhotosynthesis or catalysis

�Combustion or decomposition

(C6H10O5)n + n6O2 ↑ (1)

For a vascular plant, the effects of the reagents (e.g. CO2, H2O)re linked through a range of secondary processes dependent onnvironmental conditions, which may act neutrally, positively, ors environmental stressors, on the forward and reverse reactions;uch that an increase in a reagent (e.g. CO2) does not produce theesult indicated by the equation alone. For example, in one CO2ertilisation experiment, the N uptake by grass from the more-ecalcitrant SOC was increased, leaving more-labile SOC, and nethorter half-lives of ecosystem carbon (Gill et al., 2006). Similarly,lthough the formula we used for determining change in SOC as

function of temperature and rainfall (Bird et al., 2002) had an

2 of 0.75, it was not tested for use in rapid temporal change. Theubstitution of time for space may be true to some degree but is aimplification of the ecosystem dynamics of climate change, whichequire further consideration.

nd Environment 148 (2012) 44– 64

Some of the secondary processes have been described for range-land vegetation (e.g. Walker and Steffen, 1993; Polley et al., 2011).The more abiotic ecological factors associated with climate change,such as: (a) drought stress (longer and hotter droughts), whichlimits the forward reaction in Eq. (1); and (b) increased fire fre-quency, which favours the backwards reaction—both of which havebeen forecast for Australia by 2100 (Pitman et al., 2007; Pitmanand Perkins, 2008; McAlpine et al., 2009; Williams et al., 2009; Liuet al., 2010)—are usually not included in the more-singular “pro-cess based” models (such as modelling the effects of increase inonly CO2 and N).

Potential evaporation is forecast to increase for most ofAustralia’s rangeland, which could be expected to affect soilmoisture, surface runoff, water supply and water demand (i.e.water vapour deficit) (e.g. Bates et al., 2008). An example ofmodelling of intermediate complexity—including changes in CO2,temperature, rainfall and drought, but not including change infire regime—revealed for northeast America, a variety of spatiallydependent responses, with repeated episodes of dieback and sig-nificantly reduced biomass, but in some locations eventual forestgrowth upon quadrupling of CO2 concentration (Solomon, 1986).We now know that such a high CO2 concentration would inducedestructive positive feedback, and that temperature increase isoccurring faster than in the modelling by (Solomon, 1986), whichaccommodated ecological adaptation of local taxa.

Determination of increased biomass with anthropogenic cli-mate change has been based on a range of models or observations,such as: (a) lower forest biomass during glacial periods (with theirinherently low CO2 concentrations, (e.g. Street-Perrott et al., 1997)),(b) high photosynthesis rates in some CO2 enrichment experiments(Saxe et al., 1998), (c) forests where water was not strongly limit-ing (Boisvenue and Running, 2006), and (d) change in vegetationindices in remotely sensed imagery (e.g. Piao et al., 2011).

Some such forecasts of increased biomass with climate changemay be discounted, because (a) the causes and possibly somecharacteristics of anthropogenically-induced climate change aredifferent to those of previous glacial retreat, (b) measurementsusing flux towers alone would neglect efflux of [aqueous] dissolvedorganic carbon and [airborne] volatile aerosols (Beringer et al.,2007) and thereby overestimate net ecosystem productivity, and(c) some measured biomass increase was due to inherent forestrecovery from past disturbance (Heath et al., 1993; Cook et al., 2005;Chave et al., 2008; Muller-Landau, 2009).

Moreover there is ample evidence for a decrease in biomass withongoing anthropogenic climate change, albeit initiating at differ-ent times in different locations. Damaging influences of continuedanthropogenic climate change upon several extant ecosystemshave been modelled and observed, e.g. increase in annual dry period(i.e. drought effects), fire impacts, ozone damage, disease and pestvulnerability, and decomposition rates (Saxe et al., 1998; Barberet al., 2000; Calder and Kirkpatrick, 2008; Kurz et al., 2008; vanMantgem et al., 2009; Wallace et al., 2009; Williams et al., 2009;Carnicer et al., 2011; Hofmockel et al., 2011; Piao et al., 2011).There is evidence of widespread (e.g. global, Eurasia) and regional(e.g. southwest WA, Iberian Peninsula) decreased biomass througheither: (a) downturn in growth (which can incur a concomitantnet reduction in biomass if decomposition rates or fire frequenciesdo not also decrease), or (b) decreased forest density, or increasedmortality; due to reduced rainfall or increased drought effects; (e.g.Wallace et al., 2009; Zhao and Running, 2010; Carnicer et al., 2011;Piao et al., 2011). Drought effects will worsen in many regionsthat are already water-limited, and they have been observed in

areas previously not considered as being water limited (e.g. Allenet al., 2009). Such negative aspects are likely to detract from sus-tainability of existing forests and shrublands, and their abilityto store carbon; and will thereby induce net carbon efflux (i.e.

ems a

pe

idtipaddpcito

(mTr(da[t

cieWseft

(Oqgtmme2lotncsscaaee(i

0tic‘C

C. Dean et al. / Agriculture, Ecosyst

ositive feedback to climate change, cf. Houghton (1997), Kashiant al. (2006)).

Some specific examples of reduced biomass [while under thenfluence of increasing CO2, N and temperature] are: (a) since 1990ecreased forest cover in unmanaged areas and reduced recovery inhinned areas (Wallace et al., 2009) resulting from reduced rainfalln southwest Western Australia, since ca. 1975 due in a significantart to anthropogenic climate change (Bates et al., 2008; Pitmannd Perkins, 2008) and; (b) reduced growth of forests in Eurasiauring the growing season since ca. 1997, also concomitant withecreased rainfall (Piao et al., 2011). Notably in the latter example,rior to 1997 there was increased growth, possibly due to climatehange affects, before the affects became too stressful (and swing-ng the chemical equation (Eq. (1)) more to the reverse directionhan the forward). Thus, for some locations, the ‘honeymoon period’f anthropogenic climate change has passed.

Carnicer et al. (2011) noted that the observed negative effectsi.e. constituting reduced growth or reduced biomass) of cli-

ate change, were more pronounced in more xeric localities.he reduced soil moisture (from evapo-transpiration affects) willeduce resistance to erosion, thereby decreasing productivityThorne et al., 2005; Eldridge et al., 2011b). That suggests that theryer parts of Australia are more likely to be affected. Addition-lly, vulnerability to drought increases with land degradation andhuman] population increase (Pickup, 1998)—which is relevant tohe majority of Australian rangeland.

Net primary productivity may increase with increase in pre-ipitation, however soil has high sensitivity to temperaturencrease—increasing emissions from SOC and decreasing netcosystem productivity (Peng et al., 1998; Xuyang et al., 2008).here SOC is forecast to decrease, there could be net ecosystem C

equestration if increase in precipitation is sufficient (e.g. Dohertyt al., 2009), however no increase in precipitation was consistentlyorecast for Australian rangelands amongst the different modelsested here.

Positive feedback will increasingly augment climate changeScholze et al., 2006; Denman et al., 2007; Piao et al., 2011).ne mechanism of positive feedback is that, increased fire fre-uency (such as forecast for Australia over the next century)enerally changes vegetation types to those with more dispersedree coverage, enlargement of tree hollows, increased proportion of

ore fire-tolerant species, and increased grass cover, i.e. changingature or mesic forest to savannah or grassland—a more-fire-prone

cosystem (Cloudsley-Thompson, 1975; Cook et al., 2005; Pittock,009) which has a lower ecosystem carbon. That process is the

ikely pathway of the �SOC calculated here. Increased frequencyf fire initially acts to remove young tree recruits and in the longererm removes the more mature trees (which become more vul-erable to fire with age) (Liedloff et al., 2001). However, underonditions of minimally intense fires, drought tolerance of somehrub species (such as phreatophytes) may allow dominance overome grasses that are without significant belowground storageomponents (White et al., 2009). Other positive feedback mech-nisms are: (a) reduced site-quality (e.g. through soil nutrient lossnd reduced clay content) from more-frequent fire and increasedrosion (e.g. McIntosh et al., 2005; Gough et al., 2007; Williamst al., 2009), (b) lower local rainfall due to reduced forest biomassLi et al., 2000; McAlpine et al., 2009); and (c) increased efflux fromncreased SOC decomposition (Kirschbaum, 2004).

Carter et al. (2004) forecast an increase in rangeland SOC, to.3 m depth, of ∼4% up to 2070, based on CO2 fertilisation of vege-ation growth, change in rainfall and temperature, but ignoring the

ncreased fire frequency and the increased drought severity fore-ast for Australia (e.g. Williams et al., 2009). That modelling forecastsmall’ losses in woody biomass under drier conditions. Based onO2 fertilisation alone (without change in temperature or rainfall)

nd Environment 148 (2012) 44– 64 57

they found a 5% increase in carbon and concluded that N availabil-ity was limiting. Phosphorous is also notably limiting over much ofAustralia and it is unlikely to be bolstered by global change. Withmore comprehensive modelling to 2100 (although still without fireand drought), the forecast change in SOC for Australia was 0.6%and −6.4%, for low- and high-emission climate change scenariosrespectively (Grace et al., 2006), with highest losses in the mid- tonorthern-latitude rangelands. Also with comprehensive modelling(although without modelling change in drought or fire) an increasein SOC was forecast for Patagonian shrubland, peaking around 2050but decreasing thereafter under a high-emission scenario and nonet gain thereafter for the low-emission scenario (Carrera et al.,2007).

Another model of vegetation change, based on temperatureand rainfall change with climate change (as was our modellingof �SOC), forecast an extensive expansion of the arid interiorof Australia by 2070, at the expense of eucalypt forests (Adams-Hosking et al., 2011). A reduction in the northern rangeland forestsby desert expansion has been forecast (Zeng and Yoon, 2009). WithSOC generally being correlated with woody biomass (Jackson andAsh, 1998; Woomer et al., 2004; Wynn et al., 2006), these changeswould eventually incur SOC efflux. McKeon et al. (1998) forecast aSOC efflux of 50% from Queensland rangelands, from modelling ofglobal change effects (although without increase in fire), accompa-nying a temperature increase of 2.7 ◦C, which is likely over muchof Australia by 2100 under a high emission scenario (Pitman andPerkins, 2008). Our SOC efflux figure of 32 (9)% (Table 3b) for QLDfor the coming century is of a similar order of magnitude to theirvalue, within likely error margins. More generally, the magnitudeof our modelled change in SOC is within expected ranges whenmajor ecosystem effects (such as fire and drought) are considered,although the timing of the flux must also be a major consideration.

Our method of modelling the influence of climate change onSOC was based on differences between ecosystems, and dependenton mean annual temperature and rainfall. However, because it wasderived from a spatially diverse set of measurements (Bird et al.,2002) then it inadvertently also reflected differences in ecosystemcharacteristics such as fire and drought regimes. Fire and droughtare some aspects forecast to have increasingly negative impacts onmuch of Australian carbon stocks under climate change (discussedabove). The link between ecosystem change through fire has beenapplied previously in climate change modelling (e.g. Krawchuket al., 2009). The modelling of temporal change, using a formuladesigned for spatial variability, assumes that the ecosystems willeffectively relocate over time (although not necessarily by migra-tion but possibly by change in relative species abundance).

The SOC measured in development of the relationship of Birdet al. (2002) was inherently dependent on biomass, and con-sequently the modelled �SOC (−38 (12)% to 0.3 m) inherentlyreflected change in biomass. However, decrease in SOC is gener-ally significantly delayed after decrease in biomass, due in partto the surge of decomposing biomass after tree mortality, and tothe long half-life of some major soil pools (e.g. from 250 to 400 yr,Liao et al., 2006; Neff et al., 2009). For example, for replacementof forests by prairie due to fire during the Holocene epoch, theSOC took at least two millennia to stabilise (Baker et al., 1996) butsome of that duration may have been due to the gradual natureof change in biomass, whereas anthropogenic climate change isoccurring relatively rapidly.

Thus, rather than the calculated 38 (12)% drop in SOC happen-ing over a century, it may well take several centuries. The forecast4% increase in SOC from Carter et al. (2004) ignored some aspects

of anthropogenic climate change that are shown to be currentlyactive. That increase will mostly likely be reduced. Also, consid-ering the observed, globally widespread decrease in biomass andin biomass growth, if there is any further biomass increase due to

5 ems a

ciiollbts

c(cMfptStA

4

stsaanMtdtribi

idtpzldreAWcadaairrb

srfs‘b

8 C. Dean et al. / Agriculture, Ecosyst

limate change in Australian rangeland then it may only happenn the very near future. Consequently, any increase in SOC due toncrease in biomass would be indistinguishable against the dropf 38 (12)% in the longer term. That calculated efflux of SOC is notikely to occur in one century, but it can be stated as a decreaseikely to occur over several centuries resulting from the change iniomass occurring from 2000 to 2100. Additionally, if the ecosys-ems are unable to adapt then our forecast �SOC may be a best-casecenario, with higher efflux likely.

Our estimated total SOC efflux for Australia due to the climatehange impact on rangeland ecosystems to 2100 was 1.8 (0.6) Pga significant quantity). It was not possible to say which climatehange model or IPCC scenario was the most accurate. The CSIRO-k3.5 model consistently forecast higher emissions: 55% and 72%

or A1B and A1FI respectively. If that model and scenarios areroven to be the most accurate then emissions from SOC are likelyo be higher than the average. Current trends are above the IPCCRES scenario A1FI (Pittock, 2009), i.e. influences will be strongerhan the average used here, in which case the SOC efflux fromustralian rangelands could well be in the order of 1–2 Pg by 2100.

.4. Sequestration: accounting, threats and enhancement

For successful rangeland C sequestration four avenues have beenuggested (McKeon et al., 1992): (a) a more reliable estimate ofhe expanse of degradation, (b) monitoring the changes in carbontock with time, (c) demonstration of the attainable sequestration,nd (d) logistics for decreasing stock numbers. They also described

variety of effects of future climate change, which were mostlyegative for both livestock and carbon sequestration. Gifford andcIvor (2009) concluded that before reliable estimates of seques-

ration can be made much more information was required on SOCynamics, such as accompanying land degradation and rehabilita-ion. Calculations here showed that uncertainty in the root:shootatio (i.e. in root biomass) was a major contributor to uncertaintyn emissions and sequestration. The correlation between long-termiomass and SOC (Jackson and Ash, 1998; Wynn et al., 2006) would

ndicate accompanying uncertainty for �SOC.For sustainable management in the rangelands more effort

s needed in assessing the extent, location and severity of landegradation (CIE, 2000) (these being sites for potential seques-ration). Historically, rangeland research has focussed on pastureroduction for livestock rather than tree biomass or riparianone condition, or C fluxes. State-held, detailed data on range-and condition (at the paddock and property-levels), watering pointevelopments, and on livestock stocking numbers, are of highlyestricted access (even between government departments), andven for non-freehold properties leased from the government.ssessing rangeland degradation necessitates remote-sensing (e.g.allace et al., 2006) however its adoption requires: (a) better

oordination between the government departments (e.g. leasedministration, conservation, livestock trade, and infrastructureevelopment), (b) innovation to find spectral-band combinationspplicable to high spatial-variability in terrain and vegetation,nd with minimal ground-truthing (e.g. Dean, 2005), (c) purchas-ng imagery of sufficient temporal and spatial resolution, and (d)esources for sufficient ground-truthing. In a review of Australianangeland studies Foran (2007) estimated that it would be ∼2030efore adequate spatial methods were applied nationwide.

The difference between the estimates for C emissions from live-tock of 8.2 and 13 Tg yr−1 (Robertson, 2003; Cook et al., 2010espectively) (with the latter including an undisclosed amount

rom outside the commercial rangelands) indicates another majorource of uncertainty in annual emissions (here equal to 15% ofTotal2’). This is augmented by the lack of data on livestock num-ers, with restrictions on data access even between government

nd Environment 148 (2012) 44– 64

departments (Greg O’Reilly, NT government, 2005; personal com-munication).

Complexities arise for carbon accounting in rangelands, from:(a) high 3D spatial heterogeneity of SOC stocks (e.g. higher underwoody vegetation and at deeper levels where phreatophytes accessdeep-set ground water), (b) long timelines for change, (c) multi-ple land uses, (d) contribution from aeolian, surface water, andgroundwater transport of carbon, and (e) probable fire increaseaccompanying climate change and population increase. SOC maydecrease due to high rates of fuel-wood extraction (Kaul et al.,2009). The Australian rangeland forests are used also for timberextraction, both for firewood collection and commercial forestry(on private estates, where the extraction volumes are more inde-terminate than in State forests (Henry et al., 2002) and hence soare the impacts on carbon stocks). For example, sandalwood (San-talum spicatum) has been extracted (primarily for its aromatic oil)from Western Australian since ca. 1850, but is now sourced from areduced area of ∼50 Mha of rangeland. From 1850 to 1900 approxi-mately 0.175 Tg were extracted, roughly equal to the stock availablein 1970 (Kealley, 1991). (This is however only around a thousandthof the biomass lost in QLD deforestation). From data in (Kealley,1991) and (Kealley, 1989), from 1984 to 2011, the harvestable stockwas reduced by 36%. Sustainability of extraction is in doubt dueto poor regeneration compared with the extraction rate (Kealley,1991). Due to sandalwood’s sensitivity to fire, grazing by livestockand feral herbivores, diminished stocks and the need for succes-sive years of good rainfall for regeneration (Kealley, 1991); climatechange is expected to impact negatively on sandalwood sustain-ability, and a long-term impact on both biomass and SOC to somedegree would be expected.

Buffel grass is highly prized by many pastoralists for livestockproduction. However, due to its impact on biodiversity (Fairfax andFensham, 2000; Clarke et al., 2005; Jackson, 2005; Smyth et al.,2009; McDonald and McPherson, 2011), and on fire regimes (dis-cussed below), and its invasive capacity, it is a declared pest plantor major environmental weed in Australia, with the potential tocolonise over half of the mainland (e.g. Grice and Martin, 2006;Friedel et al., 2008; SA-ALNRM, 2009). Increased grazing pressure,deforestation, and thinning, are linked to colonisation of Aus-tralian commercial rangeland by buffel grass (Ludwig and Tongway,2002; Eyre et al., 2009); and encroachment is facilitated by nativegrasses’ previous lack of selective pressure from large mammaliangrazers—presenting a point of advantage for buffel grass under live-stock grazing (Hodgkinson et al., 1989; Ludwig and Tongway, 2002;Foxcroft et al., 2010).

Buffel grass infestations can reduce ecosystem biomass and SOCby attracting grazing pressure, depriving shrubs of moisture, andincreasing the area burnt and fire intensity during wildfire (e.g.causing increased mortality of woody vegetation) (Ibarra-Floreset al., 1999; Williams and Baruch, 2000; Clarke et al., 2005; Kinget al., 2008; Miller et al., 2010). For example, from modelling fuelloads and fire histories in central Australia King et al. (2008) foundthat with buffel grass, for a fire of moderate size (500 ha), the fre-quency of occurrence increased by 40%, and for fires of moderatefrequency, the area burnt increased five-fold, but with greater pro-portional increases for larger, less-frequent fires.

The mechanisms of buffel grass’s fire promotion and resilienceto fire include: (a) maintaining high loads of flammable biomass, (b)more intense fires than to which native ecosystems are accustomed(McDonald and McPherson, 2011), (c) a relatively high, below-ground component that quickly resprouts after fire (Ward et al.,2006; Rao and Allen, 2010), and (d) a high surface-to-volume ratio

compared with woody vegetation (increasing flammability). Also,anthropogenic nitrogen fertilisation can preferentially increasegrowth of invasive grasses over some native grasses, to the extentof greatly increasing fire risk (Rao and Allen, 2010). A cycle of

ems a

idMwtgas

((wspcf

mstsc

(bWaa2hwtsfifitwlsacwm

edadawp

acKiaeglrcria

C. Dean et al. / Agriculture, Ecosyst

nvasive-grass, fire, grassland-dominance then establishes, to theetriment of woody biomass (e.g. Westoby et al., 1989; Búrquez-ontijo et al., 2002). This will also contribute to lower SOC, asoody biomass is correlated with SOC (including for woody-

hickening (Eldridge et al., 2011a)). Modelled increase in grassrowth due to CO2 fertilisation (even without buffel grass), and theccompanying increase in fire, has been shown to reduce carbontocks in grazed, timbered rangelands (Howden et al., 2001).

As buffel grass congregates in the [more moist] riparian zonesFig. 7), and under shrubs and CWD, i.e. in favourable micrositesMcDonald and McPherson, 2011)—which is where the rangelandoody biomass is often concentrated (Figs. 6 and 7)—then it con-

titutes a threat to rangeland C stocks. The increased fire threat alsorevails in remnant woodland that neighbours deforested areasolonised by buffel, as the higher wind speed in the latter promotesaster fire spread (Freifelder et al., 1996).

As fire is expected to increase in severity and frequency with cli-ate change in Australian rangelands, then buffel grass presents a

ynergistic, increased danger of C efflux. Thus it will also contributeo climate change positive feedback. It may also spread furtherouth via the extended aridity of the higher emission scenarios oflimate change (Steel et al., 2008).

Buffel maintains biomass and spreads after destockingD’Antonio and Vitousek, 1992). Possible reasons for the higheriomass of buffel that we observed inside livestock exclosures inA are: (a) minimal grazing pressure in the exclosures, and (b)

dditional soil and nutrients blown in from the nearby grazed areand accumulation under small plants and shrubs (Spooner et al.,002). The exclosures were isolated from fire by large expanses ofeavily grazed land [that would not carry fire well]. However, ifhole properties were destocked (e.g. for financial carbon seques-

ration schemes) then they would not be isolated from ignitionources and spread of fire, and would consequently be at increasedre risk after high accumulation of buffel grass. Reduced grass

uel loads accompanying rangeland commercialisation have beenmplicated in woody-thickening (e.g. James et al., 1999). Similarly,hrough its increased fire effects, high grass biomass can reduceoody biomass in the absence of grazing (e.g. Werner, 2005). In the

ong-term that higher fire risk, could offset carbon gains from large-cale destocking, and thereby limit the fruitfulness of such venturesimed at carbon sequestration. If properties were destocked forarbon sequestration, it may be necessary to retain some artificialatering points, for at least the grazing pressure exerted by nativeacropods.Feral herbivores (e.g. camels, goats, pigs, buffalo and horses

tc.) may not be constrained by destocking, standard fencing, norecommissioning of artificial water points if springs etc. remainedccessible. Consequently ferals, which can contribute to land degra-ation, would thereby continue to detract from rangeland biomassnd SOC in some locations. Fencing of natural watering pointsould threaten the livelihood of larger native animals and thereforeossibly of native ecosystem functions.

Reducing grazing pressure alone, especially if feral herbivoresre present, is insufficient to restore native grasses and in someases trees, and exotic grasses must also be controlled (Semple andoen, 2003; Allcock and Hik, 2004). Methods of buffel management

nclude control of dominance, eradication, reduction of spread,nd suppression (Jackson, 2004; Franklin et al., 2006; Friedelt al., 2008; Brenner, 2010). Once tractable solutions for the buffelrass and feral animal issues are found, especially where they areikely to have significant influence over carbon stocks such as iniparian areas, then significant emissions from rangelands could be

urtailed and sequestration encouraged. The specific C efflux fromiparian erosion alone, currently has high uncertainty: a prioritys to quantify its spatial distribution and intensity, for both carbonccounting and for monitoring rangeland condition. Without

nd Environment 148 (2012) 44– 64 59

destocking and widespread land rehabilitation, a major source ofC emissions can still be addressed by limiting riparian land degra-dation (being regions of higher C concentration). Mechanisms forrestoring riparian ecosystems are discussed in Bainbridge (2008)and they begin with problem-area isolation. That reinforces theimportance of directly addressing some major issues described inthis study: (a) the juxtaposition of artificial watering points andriparian areas, (b) the erosion foci of vehicular or livestock crossingsof riparian areas, (c) buffel grass infestations in riparian zones.

One management alternative, that would simultaneously pro-mote reduced deforestation for grazing, rangeland degradation,and livestock emissions, while engendering rangeland seques-tration through forest regrowth and rehabilitation, would be tosubstitute the management of rangelands for protein productionfrom livestock [for human consumption], by in vitro protein pro-duction (e.g. Edelman et al., 2005; Hopkins and Dacey, 2008). Thatoption would however require a major administrative intervention.

5. Conclusions

Biomass and soil carbon are both major pools in rangelandcarbon accounting, although the latter has a longer timescale.Livestock methane is a significant short-term contributor. Thepotentially forested rangelands govern the major C stocks andfluxes in both biomass and soil carbon. Net annual C emissions fromAustralian rangelands due to commercial grazing are currently 32(±10) Tg yr−1 including all sources. Due to deforestation and landdegradation since commercialisation of the rangelands in 1850, theC efflux from biomass in the commercial rangelands is likely to behigher than 1 Pg.

The fastest C sequestration option, QLD regrowth of 0.79(±0.40) Pg, is less than half of the emission from SOC due to theaffects of future climate change (1.8 (0.6) Pg). The latter may takelonger than one century to manifest but depending on the trajectoryof climate change it may be >1.8 Pg. The climate change effect is ofsimilar magnitude to the annual global emissions from fossil fuels,and to global deforestation plus wood-product decomposition andemissions from soil (each ∼1.5 Pg yr−1) (Fig. 1). Rehabilitation ofAustralia’s rangeland and cessation of erosion from commercialgrazing will prevent the forecast SOC emission of 1.53 (±0.5) Pg (byaround 2400) adding to the global total. (Time-lines are in the orderof several centuries for SOC sequestration through woody growth(Hibbard et al., 2003)). That rehabilitation will only be possible byattention to major threats such as riparian erosion and buffel grass.The emission-limiting and sequestration options presented herewere made independent of future climate change effects and con-sequently their quantitative values are likely to vary from thosegiven here, towards reduced sequestration capability.

There is a prevalent lack of precision in rangeland C fluxes butthe knowledge base will slowly improve through comprehensiveremote-sensing and experimentation, especially long-term stud-ies of biomass and very long-term studies for SOC, through deepsampling near woody vegetation, and corresponding modelling ofcarbon dynamics. Additionally, increased transfer of information onfactors affecting rangeland condition and the effects, are requiredbetween government departments and from there to the scientificcommunity.

Acknowledgements

We would like to thank the following individuals and agenciesfor their assistance: Martin Taylor (WWF), Ian Watson and Gary

Bastin (CSIRO), David Eldridge (UNSW); Nikki Fitzgerald and RobertWaterworth (Dept. of Climate Change and Energy Efficiency), Dept.Env. and Conservation (WA), Dept. Env. & Heritage (SA), Dept. ofEnv. Nat. Resources & the Arts (NT), Dept. Env. & Nat. Resources

6 ems a

(abaRbL

R

A

A

A

A

A

A

B

B

B

B

B

B

B

B

B

B

B

B

B

B

B

B

C

0 C. Dean et al. / Agriculture, Ecosyst

QLD), Forest Products Commission (WA), Queensland Herbarium;nd various pastoral lease holders. The manuscript was improvedy suggestions from two anonymous referees and the Editor. Wecknowledge financial support, in part, from: Rural Industries andesearch Development Corporation and the World Wildlife Fund;oth were without influence on the study design or on this report.ibrary facilities were provided by UNSW.

eferences

dams-Hosking, C., Grantham, H.S., Rhodes, J.R., McAlpine, C., Moss, P.T., 2011. Mod-elling climate-change-induced shifts in the distribution of the koala. Wildl. Res.38, 122–130.

llcock, K.G., Hik, D.S., 2004. Survival, growth, and escape from herbivory are deter-mined by habitat and herbivore species for three Australian woodland plants.Ocealogia 138, 231–241.

llen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier,M., Kitzberger, T., Rigling, A., Breshears, D.D., Hogg, E.H., Gonzalez, P., Fensham,R., Zhangm, Z., Castro, J., Demidova, N., Lim, J.-H., Allard, G., Running, S.W.A.S.,Cobb, N., 2009. A global overview of drought and heat-induced tree mortal-ity reveals emerging climate change risks for forests. For. Ecol. Manage. 259,660–684, doi:10.1016/j.foreco.2009. 09.001.

nderson, K., Bows, A., 2008. Reframing the climate change challenge in light ofpost-2000 emission trends. Philos. Trans. R. Soc. A 366, 3863–3882.

rcher, S., Boutton, T.W., McMurtry, C.R., 2004. Carbon and nitrogen accumulationin a savanna landscape: field and modeling perspectives. In: Shiyomi, M., Kawa-hata, H., Tsuda, A., Awava, Y. (Eds.), Global Environmental Change in the Oceansand on Lands. Terrapub, Tokyo, pp. 359–373.

rmston, J.D., Denham, R.J., Danaher, T.J., Scarth, P.F., Moffiet, T.N., 2009. Predictionand validation of foliage projective cover from Landsat-5 TM and Landsat-7 ET+imagery. J. Appl. Remote Sens. 3, doi:10.1117/1.3216031.

ainbridge, D.A., 2008. A Guide for Desert and Dryland Restoration. A new Hope forArid Lands. Island Press, The Center for Resource Economics, Washington, D.C.

aker, R.G., Bettis III, E.A., Schwert, D.P., Horton, D.G., Chumbley, C.A., 1996. Holocenepaleoenvironments of Northeast Iowa. Ecol. Monogr. 66, 203–234.

arber, V.A., Juday, G.P., Finney, B.P., 2000. Reduced growth of Alaskan white sprucein the twentieth century from temperature-induced drought stress. Nature 405,668–673.

arnes, N.E.J., 1997. Historical Changes in Land Condition in the South AustralianArid Pastoral Zone. Department of Environmental Sciences and Management.The University of Adelaide, Adelaide, p. 89.

arson, M.M., Malafant, K., Skjemstad, O.J., Royle, S., Janik, J.L., Spouncer, L.R., Merry,R.M., 2002. Improving understanding of Australia’s soil carbon sink. Bureau ofRural Sciences, Commonwealth of Australia, Canberra.

ates, B.C., Hope, P., Ryan, B., Smith, I., Charles, S., 2008. Key findings from the IndianOcean Climate Initiative and their impact on policy development in Australia.Clim. Change 89, 339–354, doi:10.1007/s10584-007-9390-9.

eale, I.F., 1973. Tree density effects on yields of herbage and tree components insouth west Queensland Mulga (Acacia aneura F. Muell.) scrub. Trop. Grassl. 7,135–142.

eard, J.S., 1990. Plant Life in Western Australia. Kangaroo Press, Kenthurst, NewSouth Wales, Australia.

eringer, J., Hutley, L.B., Tapper, N.G., Cernusak, L.A., 2007. Savanna fires and theirimpact on net ecosystem productivity in North Australia. Global Change Biol.13, 990–1004.

erry, S.L., Roderick, M.L., 2006. Changing Australian vegetation from 1788 to 1988:effects of CO2 and land-use change. Aust. J. Bot. 54, 325–338.

ird, M.I., Wynn, J.G., Vellen, L., Cowley, J., Zhou, Y., 2002. Soil Carbon and Carbon Iso-topes Inventory from Australian Sandy Soils [unpublished report to the CRC forGreenhouse Accounting]. Research School of Earth Sciences, Australian NationalUniversity, Canberra, Australia.

oisvenue, C., Running, S.W., 2006. Impacts of climate change on natural forestproductivity—evidence since the middle of the 20th century. Global Change Biol.12, 862–882, doi:10.1111/j.1365-2486.2006.01134.x.

ray, S.G., Golden, R., 2008. Scenario analysis of alternative vegetation managementoptions on the greenhouse gas budget of two grazing businesses in north-easternAustralia. Rangeland J. 31, 137–142.

renner, J.C., 2010. What drives the conversion of native rangeland to buffelgrass(Pennisetum ciliare) pasture in Mexico’s Sonoran Desert?: the social dimensionsof a biological invasion. Hum. Ecol. 38, 495–505, doi:10.1007/s10745-010-9331-5.

úrquez-Montijo, A., Miller, M.E., Martínez-Yrizar, A., 2002. Mexican grasslands,thornscrub, and the tranformation of the Sonoran Desert by invasive exotic buf-felgrass (Pennisetum ciliare). In: Tellman, B. (Ed.), Invasive Exotic Species of theSonoran Region. The University of Arizona Press and The Arizona-Sonora DesertMuseum, Tuscon, pp. 126–146.

utler, L.D., Cropper, J.B., Johnson, R.H., Peacock, G.L., Shaver, P.L., Spaeth Jr., K.E.,Pierson Jr., F.B., Weltz, M.A., 2003. Rangeland and Pastureland Hydrology and

Erosion. In: Butler, L.D. (Ed.), National Range and Pasture Handbook. UnitedStates Department of Agriculture, Natural Resources Conservation Service, Graz-ing Lands Technology Institute, Fort Worth, Texas.

alder, J.A., Kirkpatrick, J.B., 2008. Climate change and other factors influencing thedecline of the Tasmanian cider gum (Eucalyptus gunnii). Aust. J. Bot. 56, 684–692.

nd Environment 148 (2012) 44– 64

Carnahan, J.A., 1977. Vegetation. In: Jeans, D.N. (Ed.), Australia: A Geography. SydneyUniversity Press, Sydney, pp. 175–195.

Carnicer, J., Coll, M., Ninyerola, M., Pons, X., Sánchez, G., Penuelas, J., 2011.Widespread crown condition decline, food web disruption, and amplified treemortality with increased climate change-type drought. PNAS 108, 1474–1478,doi:10.1073/pnas.1010070108.

Carrera, A., Ares, J., Labraga, J., Thurner, S., Bertiller, M., 2007. Scenarios of futureclimate and land-management effects on carbon stocks in northern Patagonianshrublands. Environ. Manage. 40, 944–957, doi:10.1007/s00267-007-9007-0.

Carter, J.O., Wynn, J.G., Bird, M.I., 2004. Estimating NPP and impacts of climate changeon sandy soils in Australia using a soil carbon model calibrated to measure-ments of soil carbon, nitrogen and carbon isotopes. In: Singh, B. (Ed.), Supersoil2004: Program and abstracts for the 3rd Australian New Zealand Soil Conference.The Regional Institute Ltd., Gosford, New South Wales, University of Sydney,Australia, 5–9 December 2004, pp. 1–8.

Chartres, C.J., Helyar, K.R., Fitzpatrick, R.W., Williams, J., 1992. Land degradation as aresult of European settlement of Australia and its influence on soil properties. In:Gifford, R.M., Barson, M.M. (Eds.), Australia’s Renewable Resources: Sustainabil-ity and Global Change. International Geosphere-Biosphere Programme AustraliaPlanning Workshop. Bureau of Rural Resources Proceedings No. 14. AustralianGovernment Publishing Service, Canberra, 3–4 October 1990, pp. 3–33.

Caseldine, C.J., Hatton, J., 1993. The development of high moorland on Dartmoor:fire and the influence of Mesolithic activity on vegetation change. In: Cham-bers, F. (Ed.), Climatic Change and Human Impact on the Landscape: Studies inPalaeoecology and Environmental Archaeology. Chapman & Hall, London, pp.119–131.

Chave, J., Condit, R., Muller-Landau, H.C., Thomas, S.C., Ashton, P.S., Bunyavejchewin,S., Co, L.L., Dattaraja, H.S., Davies, S.J., Esufali, S., Ewango, C.E.N., Feeley, K.J.,Foster, R.B., Gunatilleke, N., Gunatilleke, S., Hall, P., Hart, T.B., Hernaındez, C.,Hubbell, S.P., Itoh, A., Kiratiprayoon, S., LaFrankie, J.V., de Lao, S.L., Makana,J.-R., Noor, M.N.S., Kassim, A.R., Samper, C., Sukumar, R., Suresh, H.S., Tan, S.,Thompson, J., Tongco, M.D.C., Valencia, R., Vallejo, M., Villa, G., Yamakura, T.,Zimmerman, J.K., Losos, E.C., 2008. Assessing evidence for a pervasive alterationin tropical tree communities. PLoS Biol. 6, doi:10.1371/journal.pbio.0060045.

Christian, C.S., Perry, R.A., 1969. Arid-land studies in Australia. In: McGinnes, W.G.,Goldman, B.J. (Eds.), Arid Lands in Perspective. The University of Arizona Press,Tuscon, Arizona, pp. 207–225.

CIE, 2000. The Rangelands. A Synthesis of Three Reports on SustainableNatural Resource Management. Prepared for Agriculture, Fisheries andForestry—Australia. Centre for International Economics, Canberra andSydney.

Clarke, P.J., Latz, P.K., Albrecht, D.E., 2005. Long-term changes in semi-arid veg-etation: invasion of an exotic perennial grass has larger effects than rainfallvariability. J. Veg. Sci. 16, 237–248.

Cloudsley-Thompson, J.L., 1975. Desert expansion and the adaptive problems of theinhabitants. In: Hadley, N.F. (Ed.), Environmental Physiology of Desert Animals.Dowden, Hutchinson and Ross, Stroudsburg, Pennsylvania, pp. 255–268.

Condon, R.W., 1972. Soil conservation in dryland Australia. In: Hall, N.H., Boden,R.W., Christian, C.S., Condon, R.W., Dale, F.A., Hart, A.J., Leigh, J.H., Marshall, J.K.,McArthur, A.G., Russell, V., Turnbull, J.W. (Eds.), The Use of Trees and Shrubsin the Dry Country of Australia,. Australian Government Publishing Service,Canberra, pp. 138–161.

Cook, G.D., Liedloff, A.C., Eager, R.W., Chen, X., Williams, R.J., O’Grady, A.P.O., Hutley,L.B., 2005. The estimation of carbon budgets of frequently burnt tree stands insavannas of northern Australia, using allometric analysis and isotopic discrimi-nation. Aust. J. Bot. 53, 621–630.

Cook, G.D., Williams, R.J., Stokes, C.J., Hutley, L.B., Ash, A.J., Richards, A.E., 2010.Managing sources and sinks of greenhouse gases in Australia’s rangelands andtropical savannas. Rangeland Ecol. Manage. 63, 137–146, doi:10.2111/08-101.1.

Cowling, S.A., Shin, Y., 2006. Simulated ecosystem threshold responses to co-varyingtemperature, precipitation and atmospheric CO2 within a region of the Amazon.Global Ecol. Biogeogr. 15, 553–566.

CSIRO, 2007. Welcome to Ozclim. Exploring Climate Change Scenarios for Australia.Available online at: http://www.csiro.au/ozclim/advanced.do? [accessed 20August 2009].

Daly, R.L., Hodgkinson, K.C., 1996. Relationships between grass, shrub and tree coveron four landforms of semi-arid eastern Australia, and prospects for change byburning. Rangeland J. 18, 104–117.

D’Antonio, C.M., Vitousek, P.M., 1992. Biological invasions by exotic grasses, thegrass/fire cycle, and global change. Annu. Rev. Ecol. Syst. 23, 63–87.

Davidson, A., Beare, S., Gooday, P., Kokic, P., Lawson, K., Elliston, L., 2006. Nativevegetation. Public conservation on private land. Cost of forgone rangelandsdevelopment in southern and western Queensland. ABARE Research Report06.13. Australian Bureau of Agricultural and Resource Economics, Common-wealth of Australia, Canberra. Available online at http://www.abare.gov.au/publications html/crops/crops 06/native veg.pdf. [accessed 2 April 2011].

Dawson, N.M., Turner, E.J., 1982. Land Degradation in the Western Arid Lands ofQueensland. Information Series Q182010. Queensland Department of PrimaryIndustries, Brisbane.

DCCEE, 2010. Australian National Greenhouse Accounts. National InventoryReport 2008 Volume 2. The Australian Government Submission to the UN

Framework Convention on Climate Change May 2010. Department of ClimateChange and Energy Efficiency, Commonwealth of Australia, Canberra. Availableonline at http://www.climatechange.gov.au/publications/greenhouse-acctg/∼/media/publications/greenhouse-acctg/national-inventory-report-2008-vol2-part-a.ashx [accessed 7 May 2011].

ems a

D

D

D

D

D

D

D

D

D

D

D

D

D

E

E

E

E

E

E

E

E

E

F

F

C. Dean et al. / Agriculture, Ecosyst

CCEE, 2011. Quarterly Update of Australia’s National Greenhouse Gas Inven-tory. June Quarter 2011 Department of Climate Change and EnergyEfficiency, Commonwealth of Australia, Canberra. Available online athttp://www.climatechange.gov.au/∼/media/climate-change/emissions/2011-06/NGGI-June-quarter-2011-PDF.pdf [accessed 19 November 2011].

ean, C., 2005. Monitoring of Pastoral Rangeland Condition in the Southern North-ern Territory (Australia) by Remote Sensing—Status and Prospects. In: Achim,R., Hill, J. (Eds.), Proceedings of the 1st International Conference on RemoteSensing and Geoinformation Processing in the Assessment and Monitoring ofLand Degradation and Desertification. Trier, Germany, 7–9 September 2005, pp.151–158.

ean, C., Roxburgh, S.H., Harper, R.J., Eldridge, D.J., Watson, I.W., Wardell-Johnson,G.W., 2012. Accounting for space and time in timbered rangelands. Ecol. Eng.38, 51–64.

ean, C., Wardell-Johnson, G., 2010. Old-growth forests, carbon and climate change:functions and management for tall open-forests in two hotspots of temperateAustralia. Plant Biosyst. 144, 180–193.

enman, K.L., Brasseur, G., Chidthaisong, A., Ciais, P., Cox, P.M., Dickinson, R.E.,Hauglustaine, D., Heinze, C., Holland, E., Jacob, D., Lohmann, U., Ramachan-dran, S., da Silva Dias, P.L., Wofsy, S.C., Zhang, X., 2007. Couplings BetweenChanges in the Climate System and Biogeochemistry. In: Solomon, S., Qin,D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L.(Eds.), Climate Change 2007: The Physical Science Basis. Contribution of Work-ing Group I to the Fourth Assessment Report of the Intergovernmental Panelon Climate Change. Cambridge University Press, Cambridge and New York,pp. 499–587.

EWR, 2007. Australia’s Native Vegetation. A summary of Australia’s MajorVegetation Groups. Department of the Environment and Water Resources, Com-monwealth Government, Canberra.

ijkstra, F., Cheng, W., 2007. Interactions between soil and tree roots acceleratelong-term soil carbon decomposition. Ecol. Lett. 10, 1046–1053.

iochon, A., Kellman, L., Beltrami, H., 2009. Looking deeper: an investigation ofsoil carbon losses following harvesting from a managed northeastern redspruce (Picea rubens Sarg.) forest chronosequence. For. Ecol. Manage. 257,413–420.

NRW, 2008. Land cover change in Queensland 2005–2006: a Statewide Land-cover and Trees Study (SLATS) Report Feb 2008. Statewide Landcover andTrees Study (SLATS) Report Feb 2008. Department of Natural Resources andWater, Queensland Government, Brisbane, Queensland. Available online athttp://www.nrw.qld.gov.au/slats/pdf/land cover change 0506/land coverchange 0506.pdf [accessed 28 February 2009].

oering, M., Uehlinger, U., Ackermann, T., Woodtli, M., Tockner, K., 2011. Spatiotem-poral heterogeneity of soil and sediment respiration in a river-floodplain mosaic(Tagliamento. NE Italy). Freshwater Biol. 56, 1297–1311, doi:10.1111/j.1365-2427.2011.02569.x.

oherty, R.M., Sitch, S., Smith, B., Lewis, S.L., Thornton, P.K., 2009. Implicationsof future climate and atmospheric CO2 content for regional biogeochemistry,biogeography and ecosystem services across East Africa. Global Change Biol.,doi:10.1111/j. 1365-2486.2009.01997.x.

onohue, R., Hill, M.J., Holloway, J., Houlder, P., Lesslie, R., Smith, J., Thackway, R.,2005. Australia’s Rangelands: An Analysis of Natural Resources, Patterns of Useand Community Assets, Report to the Natural Resource Management BusinessUnit, Department of Agriculture Fisheries and Forestry. Bureau of Rural Sciences,Australian Government, Canberra, ACT, Australia.

regne, H.E., 1995. Erosion and soil productivity in Australia and New Zealand. LandDegrad. Dev. 6, 71–78.

delman, P.D., McFarland, D.C., Mironov, V.A., Matheny, J.G., 2005. In vitro-culturedmeat production. Tissue Eng. 11, 659–662.

dney, P.A., Kershaw, A.P., De Decker, P., 1990. A Late Pleistocene and Holocenevegetation and environmental record from Lake Wangoom, Western Plains ofVictoria, Australia. Palaeogeogr., Palaeoclimatol., Palaeoecol. 80, 325–343.

ldridge, D.J., Greene, R.S.B., 1994. Microbiotic soil crusts – a review of their roles insoil and ecological processes in the rangelands of Australia. Aust. J. Soil Res. 32,389–415.

ldridge, D.J., Wong, V.N.L., 2005. Clumped and isolated trees influence soil nutri-ent levels in an Australian temperate box woodland. Plant Soil 270, 331–342,doi:10.1007/s11104-004-1774-2.

ldridge, D.J., Bowker, M.A., Maestre, F.T., Roger, E., Reynolds, J.F., Whitford, W.G.,2011a. Impacts of shrub encroachment on ecosystem structure and functioning:towards a global synthesis. Ecol. Lett. 2011, 1–14, doi:10.1111/j.1461-0248.2011.01630.x.

ldridge, D.J., Greene, R.S.B., Dean, C., 2011b. Soil Health and Climate Change. In:Singh, B.P., Cowie, A.L., Yin Chan, K. (Eds.), Climate Change Impacts on SoilProcesses in Rangelands. Soil Biology, vol. 29. Springer-Verlag, Heidelberg, pp.237–255.

llis, E.C., 2011. Anthropogenic transformation of the terrestrial biosphere. Philos.Trans. R. Soc. A 369, 1010–1035, doi:10.1098/rsta.2010.0331.

lmegreen Rafelski, L.A., Piper, S.C., Keeling, R.F., 2009. Climate effects on atmo-spheric carbon dioxide over the last century. Tellus 61B, 718–731.

yre, T.J., Wang, J., Venz, M.F., Chilcott, C., Whish, G., 2009. Buffel grass in Queens-land‘s semi-arid woodlands: response to local and landscape scale variables, and

relationship with grass, forb and reptile species. Rangeland J. 31, 293–305.

airfax, R.J., Fensham, R.J., 2000. The effect of exotic pasture development on floristicdiversity in central Queensland, Australia. Biol. Conserv. 94, 11–21.

eller, M.C., 1980. Biomass and nutrient distribution in two eucalypt forest ecosys-tems. Aust. J. Ecol. 5, 309–333.

nd Environment 148 (2012) 44– 64 61

Fensham, R.J., Guymer, G.P., 2009. Carbon accumulation through ecosystem recov-ery. Environ. Sci. Policy 12, 367–372.

Fensham, R.J., Powell, O., Horne, J., 2011. Rail survey plans to remote sensing: veg-etation change in the Mulga Lands of eastern Australia and its implications forland use. Rangeland J. 33, 1–10, doi:10.1071/RJ11007.

Fontaine, S., Barot, S., Barré, P., Bdioui, N., Mary, B., Rumpel, C., 2007. Stability oforganic carbon in deep soil layers controlled by fresh carbon supply. Nature450, 277–280, doi:10.1038/nature06275.

Foran, B.D., 2007. Sifting the future from the past: a personal assess-ment of trends impacting the Australian rangelands. Rangeland J. 29,3–11.

Foxcroft, L.C., Richardson, D.M., Rejmánek, M., Pysek, P., 2010. Alien plant invasionsin tropical and sub-tropical savannas: patterns, processes and prospects. Biol.Invasions 12, 3913–3933, doi:10.1007/s10530-010-9823-7.

Franklin, K.A., Lyons, K., Nagler, P.L., Lampkin, D., Glenn, E.P., Molina-Freaner,F., Markow, T., Huete, A.R., 2006. Buffelgrass (Pennisetum ciliare) land con-version and productivity in the plains of Sonora, Mexico. Biol. Conserv. 127,62–71.

Freifelder, R.R., Vitousek, P.M., D’Antonio, C.M., 1996. Microclimate change and effecton fire following forest-grass conversion in seasonally dry tropical woodland.Biotropica 30, 286–297.

Friedel, M., Marshall, N., Klinken, R.D.V., Grice, T., 2008. Quantifying costsand benefits of buffel grass. A report prepared for the AustralianGovernment, Land and Water Australia. CSIRO, Canberra. Availableonline at: http://lwa.gov.au/files/products/defeating-weed-menace/pn22409/pn22409.pdf [accessed 19 November 2011].

Gallant, A.J.E., Hennessy, K.J., Risbey, J., 2007. Trends in rainfall indices for six Aus-tralian regions: 1910–2005. Aust. Met. Mag. 56, 223–239.

Garnaut, R., 2008. Garnaut Climate Change Review. Cambridge University Press,Melbourne.

George, R.J., 2001. WA soil erosion under investigation. J. Agric. 42, 15–18.Gifford, R.M., McIvor, J., 2009. Rehabilitate overgrazed rangelands, restoring soil

and vegetation carbon-balance. In: Eady, S., Grundy, M., Battaglia, M., Keating,B. (Eds.), An Analysis of Greenhouse Gas Mitigation and Carbon Seques-tration Opportunities from Rural Land Use. CSIRO, St Lucia, QLD, Australia,pp. 60–76.

Gill, R.A., Anderson, L.J., Polley, H.W., Johson, H.B., Jackson, R.B., 2006. Potentialnitrogen constraints on soil carbon sequestration under low and elevated atmo-spheric CO2. Ecology 87, 41–52.

Glenn, E.P., Squires, V.R., Olsen, M., Fyre, R., 1993. Potential for carbon sequestrationin the drylands. J. Water Air Soil Pollut. 70, 341–355.

Gloor, M., Sarmiento, J.L., Gruber, N., 2010. What can be learned about carboncycle climate feedbacks from the CO2 airborne fraction? Atmos. Chem. Phys.10, 7739–7751.

Gonzalez, P., Neilson, R.P., Leniham, J.M., Drapek, R.J., 2010. Global patterns in thevulnerability of ecosystems to vegetation shifts due to climate change. GlobalEcol. Biogeogr. 19, 755–768.

Gough, C.M., Vogel, C.S., Kazanski, C., Nagel, L., Flower, C.E., Curtis, P.S., 2007. Coarsewoody debris and the carbon balance of a north temperate forest. For. Ecol.Manage. 244, 60–70.

Grace, P.R., Post, W.M., Hennessy, K., 2006. The potential impact of climate changeon Australia’s soil organic carbon resources. Carbon Balance Manage. 1, 1–14,doi:10.1186/1750-0680-1-14.

Grandy, A.S., Robertson, G.P., 2007. Land-use intensity effects on soil organic carbonaccumulated rates and mechanisms. Ecosystems 10, 58–73.

Gregorich, E.G., Greer, K.J., Anderson, D.W., Liang, B.C., 1998. Carbon distribution andlosses: erosion and deposition effects. Soil Tillage Res. 47, 291–302.

Grice, T., Martin, T., 2006. The Management of Weeds and Their Impact on Bio-diversity in the Rangelands. Cooperative Research Centre for Australian WeedManagement, Townsville. Available online at http://www.environment.gov.au/land/publications/pubs/weed-management.pdf [accessed 18 August 2011].

Harms, B.P., Dalal, R.C., Cramp, A.P., 2005. Changes in soil carbon and soil nitrogenafter tree-clearing in the semi-arid rangelands of Queensland. Aust. J. Bot. 53,639–650.

Harper, R.J., Beck, A.C., Ritson, P., Hill, M.J., Mitchell, C.D., Barrett, D.J., Smetton, K.R.J.,Mann, S.S., 2007. The potential of greenhouse sinks to underwrite improved landmanagement. Ecol. Eng. 29, 329–341.

Harrington, G.N., Oxley, R.E., Tongway, D.J., 1979. The effects of European settle-ment and domestic livestock on the biological system in poplar box (Eucalyptuspopulnea) lands. Aust. Rangeland J. 1, 271–279.

Heath, L.S., Kauppi, P.E., Burschel, P., Gregor, H.-D., Guderian, R., Kohlmaier, G.H.,Lorenz, S., Overdieck, D., Scholz, F., Thomasius, H., Weber, M., 1993. Contributionof temperate forests to the world‘s carbon budget. Water Air Soil Pollut. 70,55–69.

Heathcote, R.L., 1969. The pastoral ethic. A comparative study of pastoralresource appraisals in Australia and America. In: McGinnes, W.G., Goldman,B.J. (Eds.), Arid Lands in Perspective. University of Arizona Press, Tucson,pp. 311–324.

Henry, B.K., Danaher, T., McKeon, G.M., Burrows, W.H., 2002. A review of thepotential role of greenhouse gas abatment in native vegetation managementin Queensland’s rangelands. Rangeland J. 24, 112–132.

Hibbard, K.A., Schimel, D.S., Archer, S., Ojima, D.S., Parton, W., 2003. Grassland towoodland transitions: integrating changes in landscape structure and biogeo-chemistry. Ecol. Appl. 13, 911–926.

Hill, M.J., Roxburgh, S., McKeon, G.M., Carter, J.O., Barrett, J.D., Davies, I., 2006. Range-ASSESS. Environ. Modell. Software 21, 799–801.

6 ems a

H

H

H

H

H

H

H

H

I

I

J

J

J

J

J

J

K

K

K

K

KK

K

K

K

K

K

K

2 C. Dean et al. / Agriculture, Ecosyst

odgkinson, K.C., Ludlow, M.M., Mott, J.J., Baruch, Z., 1989. Comparative responsesof the Savanna grasses Cenchrus ciliaris and Themeda triandra to defoliation.Ocealogia 79, 45–52, doi:10.1007/BF00378238.

ofmockel, K.S., Zak, D.R., Moran, K.K., Jastrow, J.D., 2011. Changes in forest soilorganic matter pools after a decade of elevated CO2 and O3. Soil Biol. Biochem.43, 1518–1527, doi:10.1016/j.soilbio.2011.03.030.

opkins, P.D., Dacey, A., 2008. Vegetarian meat: could technology save animals andsatisfy meat eaters? J. Agric. Environ. Ethics 21, 579–596.

oughton, J., 1997. Global Warming: The Complete Briefing. Cambridge UniversityPress, Cambridge, England.

oughton, R.A., 2008. Carbon Flux to the Atmosphere From Land-Use Changes:1850–2005. TRENDS: A Compendium of Data on Global Change. Car-bon Dioxide Information Analysis Center, Oak Ridge National Labora-tory, U.S. Department of Energy, Oak Ridge, Tennessee, U.S.A. Avail-able online at http://cdiac.ornl.gov/trends/landuse/houghton/houghton.html[accessed August 14, 2009].

ouse, J.I., Prentice, I.C., Le Quéré, C., 2002. Maximum impacts of futurereforestation or deforestation on atmospheric CO2. Global Change Biol. 8,1047–1052.

owden, S.M., McKeon, G.M., Scanlan, J.C., Carter, J.O., White, D.H.I.E.G., 1991. Man-aging pastures in northern Australia to minimise greenhouse gas emissions:adaptation of an existing simulation model. In: Proceedings of the AustralianSimulation Society, Gold Coast, December 1991, pp. 168–178.

owden, S.M., Moore, J.L., McKeon, G.M., Carter, J.O., 2001. Global change and themulga woodlands of southwest Queensland: greenhouse gas emissions, impactsand adaptation. Environ. Int. 27, 161–166.

barra-Flores, F., Cox, J.R., Martin-Rivera, M., Crowl, T.A., Norton, B.E., Banner, R.E.,Miller, R.W., 1999. Soil physicochemical changes following buffelgrass estab-lishment in Mexico. Arid Soil Res. Rehabil. 13, 39–52.

lic, J., Boland, D., McDonald, M., Downes, G., Blakemore, P., 2000. Woody Den-sity Phase 1—State of Knowledge. Technical Report No. 18. National CarbonAccounting System. Australian Greenhouse Office, Commonwealth of Australia,Canberra. Available online at http://pandora.nla.gov.au/pan/23322/20020220-0000/www.greenhouse.gov.au/ncas/files/pdfs/tr18final.pdf [accessed 4 April2011].

acinthe, P.A., Lal, R., Kimble, J.M., 2001. Assessing water erosion impacts on soilcarbon pools and fluxes. In: Lal, R., Kimble, J.M., Follett, R.F., Stewart, B.A. (Eds.),Assessment Methods For Soil Carbon. Lewis Publishers, Baco Raton, Florida, pp.427–449.

acinthe, P.A., Lal, R., Kimble, J.M., 2002. Carbon dioxide evolution in runoff fromsimulated rainfall on long-term no-till plowed soils in southwestern Ohio. SoilTillage Res. 66, 23–33.

ackson, J., 2004. Impacts and Management of Cenchrus ciliaris (buffel grass) as anInvasive Species in Northern Queensland. Tropical Plant Sciences. James CookUniversity, Cairns, Queensland, p. 237.

ackson, J., 2005. Is there a relationship between herbaceous species richness andbuffel grass (Cenchrus ciliaris)? Aust. Ecol. 50, 505–517.

ackson, J., Ash, A.J., 1998. Tree-grass relationships in open eucalypt woodlands ofnortheastern Australia: influence of trees on pasture productivity, forage qualityand species distribution. Agrofor. Syst. 40, 159–176.

ames, C.D., Landsberg, J., Morton, S.R., 1999. Provision of watering points in theAustralian arid zone: a review of effects on biota. J. Arid Environ. 41, 87–121.

ashian, D.M., Romme, W., Tinker, D.B., Turner, M.G., Ryan, M.G., 2006. Carbon stor-age on landscapes with stand-replacing fires. Bioscience 56, 598–606.

aul, M., Dadhwal, V.K., Mohren, G.M.J., 2009. Land use change and net C flux inIndian forests. For. Ecol. Manage. 258, 100–108.

aur, K., Midmore, D.J., Jalota, R.K., Ashwath, N., 2006. Pasture composition in clearedand uncleared woodlands. Aust. J. Bot. 54, 459–470.

aur, K., Jalota, R.K., Midmore, D.J., 2007. Impact of tree clearing on soil attributesfor a pastoral property in central Queensland, Australia. Soil Sci. 172, 516–533.

ealley, I.G., 1989. Fragrant harvest. Landscope 4, 35–39.ealley, I.G., 1991. The Management of Sandalwood. Wildlife Management Program

No. 8. Department of Conservation and Land Management, Western Australia,Perth, WA.

ing, K., Marsden-Smedley, J.B., Cary, G., Allan, G.E., Bradstock, R., Gill, A.M.,2008. Modelling fire dynamics in the West MacDonnell Range area. Work-ing Paper 20. Desert Knowledge CRC, Alice Springs. Available online at:http://www.desertknowledgecrc.com.au/publications/downloads/DKCRC-Working-Paper-20-Modelling-fire-dynamics-in-the-West-MacDonnell-Ranges-area.pdf [accessed 4 October 2008].

inloch, J.E., Friedel, M.H., 2005. Soil seed reserves in arid grazing lands of cen-tral Australia. Part 1: seed bank and vegetation dynamics. J. Arid Environ. 60,133–161.

irschbaum, M.U.F., 2004. Soil respiration under prolonged soil warming: are ratereductions caused by acclimation or substrate loss? Global Change Biol. 10,1870–1877, doi:10.1111/j. 1365-2486.2004.00852.x.

irschbaum, M.U.F., Harms, B.P., Mathers, N.J., Dalal, R.C., 2008. Soil carbon andnitrogen changes after clearing mulga (Acacia aneura) vegetation in Queensland,Australia: observations, simulations and scenario analysis. Soil Biol. Biochem. 40,392–405.

rawchuk, M.A., Moritz, M.A., Parisien, M.-A., Van Dorn, J., Hayhoe, K., 2009. Global

pyrogeography: the current and future distribution of wildfire. PLoSOne 4,e5102, doi:10.1371/journal.pone.0005102.

urz, W.A., Stinson, G., Rampley, G., 2008. Could increased boreal forest ecosystemproductivity offset carbon losses from increased disturbances? Philos. Trans. R.Soc. B 363, 2259–2268.

nd Environment 148 (2012) 44– 64

Lal, R., 2001. Soil erosion and carbon dynamics on grazing land. In: Follett, R.F.,Kimble, J.M.a.R.L. (Eds.), The Potential of U. S. Grazing Lands to Sequester Carbonand Mitigate the Greenhouse Effect,. Lewis Publishers, Baco Raton, Florida, pp.231–247.

Lal, R., 2004. Carbon sequestration in dryland ecosystems. Environ. Manage. 33,528–544.

Lal, R., Griffin, M., Apt, J., Lave, L., Morgan, G., 2004. Response to comments onmanaging soil carbon. Science 10, 1567, doi:10.1126/science.1101271.

Leys, J., McTainsh, G., Shao, Y., 2001. Wind erosion monitoring and modelling tech-niques in Australia. In: Stott, D.E., Mohtar, R.H., Steinhardt, G.C. (Eds.), Sustainingthe Global Farm. Selected papers from the 10th International Soil ConservationOrganization Meeting. Purdue University and USDA-ARS National Soil ErosionResearch Laboratory, 24–29 May 1999, pp. 940–950.

Li, D., Komiyama, H., Kurihara, K., Sato, Y., 2000. Case studies of the impact oflandscape changes on weather modification in western Australia summer. J.Geophys. Res. 105, 12, 303-312.315.

Liao, J.D., Boutton, T.W., Jastrow, J.D., 2006. Organic matter turnover in soil physicalfractions following woody plant invasion of grassland: evidence from natural 13Cand 15N. Soil Biol. Biochem. 38, 3197–3210, doi:10.1016/j.soilbio.2006.04.004.

Liedloff, A.C., Coughenour, M.B., Ludwig, J.A., Dyer, R., 2001. Modelling the trade-offbetween fire and grazing in a tropical savanna landscape, northern Australia.Environ. Int. 27, 173–180.

Little, R.J., 1964. Official Year Book of Western Australia. No. 4. CommonwealthBureau of Census and Statistics, Perth, Australia.

Liu, Y., Stanturf, J., Goodrick, S., 2010. Trends in global wildfire potential in a changingclimate. For. Ecol. Manage. 259, 685–697.

Lorenz, K., Lal, R., Preston, C.M., Nierop, K.G.J., 2007. Strengthening the soilorganic carbon pool by increasing contributions from recalcitrant aliphaticbio(macro)molecules. Geoderma 142, 1–10.

Loughran, R.J., Elliott, G.L., McFarlane, D.J., Campbell, B.L., 2004. A survey of soilerosion in Australia using Ceasium-137. Aust. Geogr. Stud. 42, 221–233.

Ludwig, J.A., Tongway, D.J., 2002. Clearing savannas for use as rangelands in Queens-land: altered landscapes and water-erosion processes. Rangeland J. 24, 83–95.

Luly, J.G., 1993. Holocene paleoenvironments near Lake Tyrrell, semiarid northwest-ern Victoria, Australia. J. Biogeogr. 20, 587–598.

McAlpine, C.A., Syktus, J., Ryan, J.G., Deo, R.C., McKeon, G.M., McGowan, H.A., Phinn,S.R., 2009. A continent under stress: interactions, feedbacks and risks associatedwith impact of modified land cover on Australia’s climate. Global Change Biol.15, 2206–2223, doi:10.1111/j. 1365-2486.2009.01939.x.

McCarty, G.W., Ritchie, J.C., 2002. Impact of soil movement on carbon sequestrationin agricultural ecosystems. Environ. Pollut. 116, 423–430.

McDonald, C.J., McPherson, G.R., 2011. Fire behavior characteristics of buffelgrass-fueled fires and native plant community composition in invaded patches. J. AridEnviron. 75, 1147–1154, doi:10.1016/j.jaridenv.2011.04.024.

McIntosh, P.D., Laffan, M.D., Hewitt, A.E., 2005. The role of fire and nutrient loss inthe genesis of the forest soils of Tasmania and southern New Zealand. For. Ecol.Manage. 220, 185–215.

McIntyre, R.E.S., Adams, M.A., Ford, D.J., Grierson, P.F., 2009. Rewetting and litteraddition influence mineralisation and microbial communities in soils from asemi-arid intermittent stream. Soil Biol. Biochem. 41, 92–101.

McIvor, J.C., Williams, J., Gardener, C.J., 1995. Pasture management influences runoffand soil movement in the semi-arid tropics. Aust. J. Exp. Agric. 35, 55–65.

McKeon, G.M., Howden, S.M., Stafford Smith, D.M., 1992. The management ofextensive agriculture: greenhouse gas emissions and climate change. Assessingtechnologies and management systems for agriculture and forestry in relationto global climate change. Intergovernmental Panel on Climate Change, responsestrategies working group , Australian Government Publishing Service, Canberra,Australia, 20–23 January 1992, pp. 42–47.

McKeon, G.M., Carter, J.O., Day, K., Hall, W.B., Howden, S.M., 1998. Evaluationof the Impact of Climate Change on Northern Australian Grazing Industries:Final Report for the Rural Industries Research and Development Corporation.DAQ 139A. Department of Natural Resources, Brisbane, Queensland. Availableonline at http://www.regional.org.au/au/asit/compendium/i-03.htm [accessed12 August 2011].

McKeon, G.M., Hall, W., Henry, B., Stone, G., Watson, I.W., 2004. Pasture Degradationand Recovery in Australia’s Rangelands. Learning From History. Depart-ment of Natural Resources, Mines and Energy, Queensland GovernmentBrisbane.

McMahon, T.A., Gan, K.C., Finlayson, B.L., 1992. Anthropogenic changes tothe hydrologic cycle in Australia. In: Gifford, R.M., Barson, M.M. (Eds.),Australia’s Renewable Resources: Sustainability and Global Change. Interna-tional Geosphere-Biosphere Programme Australia Planning Workshop. Bureauof Rural Resources Proceedings No. 14. Australian Government Publishing Ser-vice, Canberra, 3–4 October 1990, pp. 35–66.

Metz, J., 2009. Deforestation. In: Kitchin, R., Thrift, N. (Eds.), International Ency-clopaedia of Human Geography, 1. Elsevier, Oxford, pp. 39–50.

Miles, R., McTainsh, G., 1994. Wind erosion and land management in the mulga landsof Queensland. Aust. J. Soil Water Conserv. 7, 41–45.

Miller, G., Friedel, M., Adam, P., Chewings, V., 2010. Ecological impacts of buffel grass(Cenchrus ciliaris L.) invasion in central Australia – does field evidence supporta fire-invasion feedback? Rangeland J. 32, 353–365.

Mokany, K., Raison, R.J., Prokushkin, A.S., 2006. Critical analysis of root:shoot ratiosin terrestrial biomass. Global Change Biol. 12, 84–96.

Moore, J.L., Howden, S.M., McKeon, G.M., Carter, J.O., Scanlan, J.C., 2001. The dynam-ics of grazed woodlands in southwest Queensland, Australia and their effect ongreenhouse gas emissions. Environ. Int. 27, 147–153.

ems a

M

MM

N

N

O

P

P

P

P

P

P

P

P

P

P

P

P

P

P

P

P

Q

R

R

R

R

R

R

R

R

C. Dean et al. / Agriculture, Ecosyst

ora, J.L., Guerra, J.A., Armas, C.M., Rodrìguez- Rodrìguez, A., Arbelo, C.D., Notario,J.S., 2007. Mineralization rate of eroded organic C in Andosols of the CanaryIslands. Sci. Total Environ. 378, 143–146.

uller-Landau, H.C., 2009. Sink in the African jungle. Nature 457, 969–970.urrell, B., 1984. Hydrological regimes in the Australian arid zone with notes on the

current changes in hydrological regime in the Willouran Ranges, South Australia.In: Cogger, H.G., Cameron, E.E. (Eds.), Arid Australia. Australian Museum, Sydney,pp. 327–334.

eff, J.C., Barger, N.N., Baisden, W.T., Fernandez, D.P., Asner, G.P., 2009. Soil carbonstorage responses to expanding pinyon-juniper populations in southern Utah.Ecol. Appl. 19, 1405–1416.

iklas, K.J., Enquist, B.J., 2002. Canonical rules for plant organ biomass partitioningand annual allocation. Am. J. Bot. 89, 812–819.

lofsson, J., Hickler, T., 2008. Effects of human land-use on the global carbon cycleduring the last 6000 years. Veg. Hist. Archaeobot. 17, 605–615.

aull, C.K., Ussler III, W., Mitts, P.J., Caress, D.W., West, G.J., 2006. Discordant 14C-stratigraphies in upper Monterey Canyon: a signal of anthropogenic disturbance.Marine Geol. 233, 21–36.

eng, C.H., Guiot, J., Van Campo, E., 1998. Past and future carbon balance of Europeanecosystems from pollen data and climatic models simulations. Global Planet.Change 18, 189–200.

erkins, S.E., Pitman, A.J., Holbrook, N.J., McAneney, J., 2007. Evaluation of the AR4 cli-mate models’ simulated daily maximum temperature, minimum temperature,and precipitation over Australia using probability density functions. J. Clim. 20,4356–4375.

erry, R.A., 1977. Rangeland management for livestock production in semi-arid andarid Australia. In: Proceedings of the Second United States/Australian RangelandPanel, Adelaide, 1972, Australian Rangelands Society, Perth, pp. 311–316.

iao, S., Wang, X., Ciais, P., Zhu, B., Wang, T., Liu, J., 2011. Changes in satellite-derivedvegetation growth trend in temperate and boreal Eurasia from 1982 to 2006.Global Change Biol. 17, doi:10.1111/j.1365-2486.2011.02419.x.

ickup, G., 1990. Impact of soil erosion on Australia’s rangelands. In: Proceedings ofthe 6th Australian Rangeland Conference. Australian Rangeland Society, Carnar-von, Western Australia, 3–6 September 1990, pp. 159–169.

ickup, G., 1998. Desertification and climate change—the Australian perspective.Clim. Res. 11, 51–63.

ickup, G., Chewings, V.H., 1986. Mapping and forecasting soil erosion patterns fromLANDSAT on a microcomputer-based image processing facility. Aust. RangelandJ. 8, 57–62.

inter, N., Fiedel, S., Keeley, J.E., 2011. Fire and vegetation shifts in the Americas atthe vanguard of Paleoindian migration. Quaternary Sci. Rev. 30, 269–272.

itman, A.J., Perkins, S.E., 2008. Regional projections of future seasonal and annualchanges in rainfall and temperature over Australia based on skill-selected AR4models. Earth Interactions 12, 1–50, doi:10.1175/2008EI260.1.

itman, A.J., Narisma, G.T., McAneney, J., 2007. The impact of climate change onthe risk of forest and grassland fires in Australia. Clim. Change 84, 383–401,doi:10.1007/s10584-007-9243-6.

ittock, A.B., 2009. Climate Change. The Science, Impacts and Solutions. CSIRO,Collingwood, Victoria, Australia.

olley, H.W., Morgan, J.A., Fay, P.A., 2011. Application of a conceptual framework tointerpret variability in rangeland responses to atmospheric CO2 enrichment. J.Agric. Sci. 149, 1–14, doi:10.1017/S0021859610000717.

olyakov, V.O., Lal, R., 2008. Soil organic and CO2 emission as affected by watererosion on field runoff plots. Geoderma 143, 216–222.

ressland, A.J., 1984. Productivity and management of western Queensland’s range-lands. Aust. Rangeland J. 6, 26–45.

rober, S.M., Standish, R.J., Wiehl, G., 2011. After the fence: vegetation and topsoilcondition in grazed, fenced and benchmark eucalypt woodlands of fragmentedagricultural landscapes. Aust. J. Bot. 59, 338–369, doi:10.1071/BT11026.

ueensland Department of Natural Resources, 2000. Land Cover Change in Queens-land 1997–1999. A Statewide Landcover and Trees Study (SLATS) Report. Bris-bane, Queensland. Available online at www.nrw.qld.gov.au/slats/final9799.html[accessed 28 February 2008].

adford, B.J., Thornton, C.M., Cowie, B.A., Stephens, M.L., 2007. The Brigalow Catch-ment Study: III. Productivity changes on brigalow land cleared for long-termcropping and for grazing. Aust. J. Soil Res. 45, 512–523.

aison, J., Witte, C., Walls, J., Kitchen, J., Carter, J.O., Goulevitch, B., Clark, A.,Guymer, G.P., 2009. Carbon positive management of regrowth vegetation andremnant forest. An Analysis of Greenhouse Gas Mitigation and Carbon Biose-questration Opportunities from Rural Land Use. CSIRO Available online at:http://www.csiro.au/files/files/prdz.pdf [accessed 10 March 2011].

ao, L.E., Allen, E.B., 2010. Combined effects of precipitation and nitrogen depositionon native and invasive winter annual production in California deserts. Ocealogia162, 1035–1046, doi:10.1007/s00442-009-1516-5.

ichards, G.P., Brack, C., 2004. A continental biomass stock and stock change esti-mation approach for Australia. Aust. For. 67, 284–288.

obertson, G.A., 2003. Global influences on rangelands of Australia. Rangeland J. 25,128–139.

olfe, J., 2002. Economics of vegetation clearing in Queensland. Rangeland J. 24,152–169.

oose, E., Barthès, B., 2006. Soil carbon erosion and its selectivity at the plot scale in

tropical and Mediterranean regions. In: Roose, E.J., Lal, R., Feller, C., Barthès, B.,Stewart, B.A. (Eds.), Soil Erosion and Carbon Dynamics, Advances in Soil Science.CRC Press, Boca Raton, Florida, pp. 55–72.

oxburgh, S.H., Mackey, B.G., Dean, C., Randall, L.A., Lee, A., Austin, J., 2006.Organic carbon partitioning in soil and litter in subtropical woodlands and

nd Environment 148 (2012) 44– 64 63

open forests: a case study from the Brigalow Belt, Queensland. Rangeland J. 28,115–125.

Rudiman, W.F., 2003. The anthropogenic greenhouse era began thousands of yearsago. Clim. Change 61, 261–293.

SA-ALNRM, 2009. Pest Management Fact Sheet. Buffel Grass. Available onlineat: http://www.saalnrm.sa.gov.au/Portals/8/pdf/FactSheets/fs buffelgrass.pdf[accessed 19 August 2011].

Salinger, M.J., 2007. Agriculture’s influence on climate change during the Holocene.Agric. For. Meteorol. 142, 96–102.

Sangha, K.K., Jalota, R.K., Midmore, D.J., 2005. Impact of tree clearing on soil pH andnutrient availability in grazing systems of central Queensland, Australia. Aust. J.Soil Res. 43, 51–60.

Saxe, H., Ellsworth, D.S., Heath, J., 1998. Tree and forest functioning in an enrichedCO2 atmosphere. New Phytol. 139, 395–436.

Scanlan, J.C., 2002. Some aspects of tree-grass dynamics in Queensland’s grazinglands. Rangeland J. 24, 56–82.

Scanlan, J.C., Pressland, A.J., Myles, D.J., 1991. Interaction Between Grazing Man-agement and Erosion in the Upper Burdekin River Catchment of Queensland.Final Report to Land and Water Research Development Corporation. QueenslandDepartment of Primary Industries, Brisbane.

Scarth, P., Armston, J., Danaher, T., 2008. On the Relationship between CrownCover, Foliage Cover and Leaf Area Index. In: Proceedings of the 14thAustralasian Remote Sensing and Photogrammetry Conference, Darwin, October2008.

Schimel, D.S., 1995. Terretrial ecosystems and the carbon cycle. Global Change Biol.1, 77–91.

Scholze, M., Knorr, W., Arnell, N.W., Prentice, I.C., 2006. A climate-change risk anal-ysis for world ecosystems. Proc. Natl. Acad. Sci. U.S.A. 103, 13116–13120.

Scott, A., 2001. Water erosion in the Murray-Darling Basin: Learning from the past.CSIRO Land & Water Technical Report No. 43/01. CSIRO Land & Water, Canberra.

Semple, W.S., Koen, T.B., 2003. Effect of pasture type on regeneration of euca-lypts in the woodland zone of south-eastern Australia. Cunninghamia 8,76–84.

Shao, Y., Raupach, M.R., Short, D., 1994. Preliminary assessment of wind erosionpatterns in the Murray-Darling Basin. Aust. J. Soil Water Conserv. 7, 46–51.

Singh, S.K., Singh, A.K., Sharma, B.K., Tarafadar, J.C., 2007. Carbon stock and organiccarbon dynamics in soils of Rajasthan. India. J. Arid Environ. 68, 408–421.

Sitch, S., Brovkin, V., von Bloh, E., van Vuuren, D., Eichkout, B., Ganopolski, A., 2005.Impacts of future land cover changes on atmospheric CO2 and climate. GlobalBiogeochem. Cycles 19, GB2013, doi:10.1029/2004GB002311.

Sivakumar, M.V.K., 2007. Interactions between climate and desertification. Agric.For. Meteorol. 142, 143–155.

Smith, S.V., Sleezer, R.O., Renwick, W.H., Buddemeier, R.W., 2005. Fates oferoded soil organic carbon; Mississippi Basin case study. Ecol. Appl. 15,1929–1940.

Smittenberg, R.H., Eglinton, T.I., Schouten, S., Sinninghe Damsté, J.S., 2006. Ongoingbuildup of refractory organic carbon in boreal soils during the Holocene. Science314, 1283–1286.

Smyth, A., Friedel, M., O‘Malley, C.O., 2009. The influence of buffel grass (Cenchrusciliaris) on biodiversity in an arid Australian landscape. Rangeland J. 31,307–320.

Solomon, A.M., 1986. Transient response of forests to CO2-induced climate change:simulation modeling experiments in eastern North America. Oecologia 68,567–579.

Sparrow, A.D., Friedel, M.H., Stafford Smith, M.D., 1997. A landscape-scale modelof shrub and herbage dynamics in Central Australia, validated by satellite data.Ecol. Model. 97, 197–216.

Sparrow, A.D., Friedel, M.H., Tongway, D.J., 2003. Degradation and recovery pro-cesses in arid grazing lands of central Australia Part 3: implications at landscapescale. J. Arid Environ. 55, 349–360.

Spooner, P., Lunt, I., Robinson, W., 2002. Is fencing enough. The short-term effects ofstock exclusion in remnant grassy woodlands in southern NSW. Ecol. Manage.Restor. 3, 117–126.

Steel, J., Kohout, M., Newell, G., 2008. Climate change and potential distributionof weeds. Wither the weeds under climate change? Department of PrimaryIndustries. Biosciences Research Division, Victorian Government, Melbourne.

Steffen, W.L., 2009. Climate change 2009. Faster Change & More Serious Risks.Department of Climate Change, Commonwealth of Australia, Canberra. Avail-able online at http://www.climatechange.gov.au/science/publications/faster-change-more-risks.html [accessed 24 August, 2009].

Stolbovoi, V., 2002. Carbon in Russian soils. Clim. Change 55, 131–156.Street-Perrott, F.A., Huang, Y., Perrott, R.A., Eglinton, G., Barker, P., Khelifa, L.B., Hark-

ness, D.D., Olago, D.O., 1997. Impact of lower atmospheric carbon dioxide ontropical mountain ecosystems. Science 278, 1422–1426.

Thorne, M.S., Trlica, M.J., Leininger, W.C., Child, R.D., Klein, D.A., 2005. Soil CO2 effluxresponses to soil loss on two rangeland ecosystems. Rangeland Ecol. Manage.58, 27–34.

Thornton, C.M., Cowie, B.A., Freebairn, D.M., Playford, C.L., 2007. The Brigalow Catch-ment Study: II. Clearing brigalow (Acacia harpophylla) for cropping or pastureincreases runoff. Aust. J. Soil Res. 45, 496–511.

Throop, H.L., Archer, S.R., 2008. Shrub (Proposis velutina) encroachment in a

semidesert grassland: spatial-temporal changes in organic carbon and nitrogenpools. Global Change Biol. 14, 1–12.

Tongway, D.J., Sparrow, A.D., Friedel, M.H., 2003. Degradation and recovery pro-cesses in arid grazing lands of central Australia. Part 1: soil and land resources.J. Arid Environ. 55, 301–326.

6 ems a

T

T

v

W

W

W

W

W

W

W

W

W

W

W

W

4 C. Dean et al. / Agriculture, Ecosyst

rumbore, S.E., 1997. Potential responses of soil organic carbon to global environ-mental change. PNAS 94, 8284–8291.

urner, E.J., Beeston, G.R., 1978. Resource Use. Western Arid Region Land Use Study.Land Utility Bulletin No. 23. Queensland Department of Primary Industries, Bris-bane.

an Mantgem, P.J., Stephenson, N.L., Byrne, J.C., Daniels, L.D., Franklin, J.F., Fulé,P.Z., Harmon, M.E., Larson, A.J., Smith, J.M., Taylor, A.H., Veblen, T.T., 2009.Widespread increase of tree mortality rates in the western United States. Science323, 521–524.

alker, B.H., Steffen, W.L., 1993. Rangelands and global change. Rangeland J. 15,95–103.

allace, J.F., Behn, G., Furby, S., 2006. Vegetation condition assessment and moni-toring from sequences of satellite imagery. Ecol. Manage. Restor. 71, S31–S36.

allace, J., Li, M., Traylen, A., 2009. Forest vegetation monitoring and runoff in watersupply catchments affected by drying climate. In: Proceedings of the 2009 IEEEInternational Geoscience and Remote Sensing Symposium, IGARSS 2009, CapeTown, 12–17 July 2009, pp. 939–942.

ard, J.P., Smith, S.E., McClaran, M.P., 2006. Water requirements for emergence ofbuffelgrass (Pennisetum ciliare). Weed Sci. 54, 720–725.

aterworth, R.M., Richards, G.P., 2008. Implementing Australian forest manage-ment practices into a full carbon accounting model. For. Ecol. Manage. 255,2434–2443.

atson, I.W., Thomas, P.W.E., Fletcher, W.J., 2007. The first assessment, using arangeland monitoring system, of change in shrub and tree populations acrossthe arid shrublands of Western Australia. Rangeland J. 29, 25–37.

erner, P.A., 2005. Impact of feral water buffalo and fire on growth and survivalof mature savanna trees: an experimental field study in Kakadu National Park,northern Australia. Aust. Ecol. 30, 625–647.

estoby, M., Walker, B., Noy-Meir, I., 1989. Opportunistic management for range-lands not at equilibrium. J. Range Manage. 42, 266–274.

eston, N.B., Joye, S.B., 2005. Temperature-driven decoupling of key phases oforganic matter degradation in marine sediments. PNAS 102, 17036–17040,doi:10.1073/pnas.0508798102.

heeler, J.L., Hutchinson, K.J., 1973. Production and utilization of food for ruminants.In: Alexander, G., Willimas, O.B. (Eds.), The Pastoral Industries of Australia. Prac-tice and Technology of Sheep and Cattle Production, Sydney University Press,Sydney, pp. 201–232.

hite, J.D., Gutzwiller, K.J., Barrow, W.C., Randall, L.J., Swint, P., 2009. Modeling

mechanisms of vegetation change due to fire in a semi-arid ecosystem. Ecol.Model. 214, 181–200.

illiams, D.G., Baruch, Z., 2000. African grass invasion in the Americas:ecosystem consequences and the role of ecophysiology. Biol. Invasions 2,123–140.

nd Environment 148 (2012) 44– 64

Williams, O.B., Calaby, J.H., 1985. The hot deserts of Australia. In: Noy-Meir, A.M.E.,Goodall, I.D.W. (Eds.), Ecosyst. World, 12A. Hot Deserts and Arid Shrublands.Elsevier, Amsterdam, pp. 269–312.

Williams, R.J., Bradstock, R.A., Cary, G.J., Enright, N.J., Gill, A.M., Liedloff, A.C., Lucas,C., Whelan, R.J., Andersen, A.A., Bowman, D.J.M.S., Clarke, P., Cook, G.J., Hennessy,K., York, A., 2009. The Impact of Climate Change on Fire Regimes and Biodiversityin Australia—A Preliminary Assessment. CSIRO, Canberra, p. 257.

Witt, G.B., Luly, J., Fairfax, R.J., 2006. How the west was once: vegetation change insouth-west Queensland from 1930 to 1995. J. Biogeogr. 33, 1585–1596.

Witt, G.B., Noëla, M.V., Bird, M.I., Beetona, R.J.S., Menzies, N.W., 2011. Carbonsequestration and biodiversity restoration potential of semi-arid mulga landsof Australia interpreted from long-term grazing exclosures. Agric. Ecosyst. Env-iron., doi:10.1016/j.agee.2011.02.020.

Woods, L.E., 1984. Land degradation in Australia. Australian Government PublishingService, Canberra.

Woodwell, G.M., Hobbie, J.E., Houghton, R.A., Melillo, J.M., Moore, B., Peterson, B.J.,Shaver, G.R., 1983. Global deforestation: contribution to atmospheric carbondioxide. Science 222, 1081–1086.

Woomer, P.L., Touré, A., Sall, M., 2004. Carbon stocks in Senegal’s Sahel transitionzone. J. Arid Environ. 59, 499–510, doi:10.1016/j.jaridenv.2004.03.027.

Wynn, J.G., Bird, M.T., Vellen, L., Gramd-Clement, E., Carter, J.O., Berry, S.L., 2006.Continental-scale measurements of the soil organic carbon pool with cli-matic, edaphic, and biotic controls. Global Biogeochem. Cycles 20, GB1007,doi:10.1029/2005GB002576.

Xuyang, L.U., Cheng, G., Xiao, F., Fan, J., 2008. Modeling effects of temperature andprecipitation on carbon characteristics and GHGs emissions in Abies fabric forestof subalpine. J. Environ. Sci. 20, 339–346.

Yamada, K., Kojima, T., Abe, Y., Saito, M., Egashira, Y., Takashi, N., Tahara, K., Law,J., 2003. Restructuring and afforestation of hardpan area to sequester carbon. J.Chem. Eng. Japan 36, 328–332.

Young, R., Wilson, B.R., McLeod, M., Alston, C., 2005. Carbon storage in the soils andvegetation of contrasting land uses in northern New South Wales, Australia.Aust. J. Soil Res. 43, 21–31.

Zeng, N., Yoon, J., 2009. Expansion of the world‘s deserts due to vegetation-albedo feedback under global warming. Geophys. Res. Lett. 36, L17401,doi:10.1029/2009GL039699.

Zhao, M., Running, S.W., 2010. Drought-induced reduction in global terrestrial netprimary production from 2000 through 2009. Science 329, 940–943.

Zuazo, V.H.D., Pleguezuelo, C.R.R., 2008. Soil-erosion and runoff prevention by plantcovers: a review. Agron. Sustain. Dev. 28, 65–86, doi:10.1051/agro:2007062.

Zummo, L.M., Friedland, A.J., 2011. Soil carbon release along a gradient of physicaldisturbance in a harvested northern hardwood forest. For. Ecol. Manage. 261,1016–1026.