1

Agriculture, Ecosystems and Environment 1

Influence of land sharing and land sparing strategies on patterns of vegetation and 2

terrestrial vertebrate richness and occurrence in Australian endangered eucalypt 3

woodlands. 4

5

Damian R. Michael*a, b, Jeff T. Wood a, Thea O’Loughlin a and David B. Lindenmayer a, b 6

7

a Fenner School of Environment and Society, The Australian National University, Canberra, 8

ACT 0200 Australia. bNational Environmental Science Programme, Threatened Species 9

Recovery Hub. 10

*Corresponding author: damian.michael@anu.edu.au 11

12

Word count: (7440 - including References, Tables and Figures) 13

14

15

16

Running Head: Influence of land management on biodiversity. 17

18

19

2

Abstract 20

Native vegetation placed under an agri-environment scheme (AES) is purported to support 21

greater biodiversity than vegetation managed for intensive livestock grazing, and 22

conservation reserves are purported to support greater biodiversity than land sharing under 23

AES. These predictions underpin financial incentive delivery programs that enable 24

landholders to adopt environmentally friendly agricultural practices. To evaluate these 25

predictions, we established a biodiversity monitoring program in endangered temperate 26

eucalypt woodland communities in southern Australia. We compared vegetation variables 27

and vertebrate species richness and abundance among sites under different land management 28

between 2010 and 2014. Our sites included: 1) woodland remnants on private property 29

recently placed under an AES land management agreement (land sharing), 2) woodland 30

remnants in State conservation reserves as reference areas (land sparing), and 3) woodland 31

remnants used for intensive livestock production as controls. We used hierarchical 32

generalized linear models to examine patterns of biodiversity among management classes and 33

over time. We found conservation reserves were structurally more complex and floristically 34

richer compared to production sites, and AES supported greater cover of native perennial 35

grass. Reptile and amphibian species richness and abundance, and total bird species richness 36

did not differ significantly among management classes, although AES and reference sites 37

supported more birds of conservation concern. Arboreal marsupials were significantly more 38

species rich in conservation reserves than AES. Temporal patterns in vertebrate species 39

richness were related to post-drought climatic conditions. Our findings suggest that strategies 40

involving land sharing under AES are as effective as land sparing (e.g. conservation reserves) 41

for bird conservation, but alternative strategies may be required to enhance habitat for less 42

mobile species such as frogs and reptiles, or species dependant on old growth vegetation such 43

as arboreal marsupials. 44

3

Keywords: Agricultural landscape; arboreal marsupials; birds; management intervention; 45

reptiles; temperate woodland. 46

47

1. Introduction 48

With the global population approaching nine billion people, there is mounting pressure to 49

provide food security while at the same time arrest the decline of biodiversity (Brussaard et 50

al., 2010; Godfray et al., 2010; Chappell and LaValle, 2011; Tscharntke et al., 2012). 51

However, attempts to integrate production and conservation present a major conservation 52

challenge (Tilman et al., 2002; Habel et al., 2015), as approximately 40% of the earth’s land 53

is used for agriculture and the estimated rates of biodiversity loss are calculated to be 1,000 – 54

10,000 times the pre-human background rate of extinction (Chappell and LaValle, 2011). In 55

recent years, strategies to minimize human impacts on the land include land sharing and land 56

sparing (Fischer et al., 2008; Chappell and LaValle, 2011; Kleijn et al., 2011). The former 57

strategy involves integrating biodiversity conservation and low-yield food production on the 58

same land (Phalan et al., 2011b). An example of this strategy is the European Union’s agri-59

environment scheme, a policy instrument that involves paying farmers to modify agricultural 60

practices to mitigate the negative effects of agricultural intensification on biodiversity (Kleijn 61

et al., 2011; Concepcion et al., 2012). The latter strategy involves separating land for 62

conservation from high-yielding crop land (Fischer et al., 2008). Protected areas are one 63

example of this strategy. These areas represent clearly defined geographical spaces that are 64

recognised, dedicated and managed, through legislation, to achieve the long term 65

conservation of nature, associated ecosystem services and cultural values (Dudley et al., 66

2010). Understanding which of these two strategies can accommodate greater biodiversity in 67

4

agricultural landscapes around the world remains a key question (Kleijn et al., 2011; Habel et 68

al., 2015). 69

70

In 2008, the Australian Government established the Caring for Our Country initiative 71

(Commonwealth of Australia 2009). This initiative delivered grants to State Government 72

organisations and natural resource management agencies. The North East Catchment 73

Management Authority (NECMA) in Victoria, Australia, was a successful recipient of a grant 74

which funded the project ‘Improving landscape scale conservation of threatened grassy 75

woodland ecosystems in the Greater Murray Goulburn catchment’. The aim of this project 76

was to establish contractual agreements with private landholders to manage approximately 77

600 ha of endangered grassy woodland vegetation for biodiversity outcomes. This significant 78

financial investment by the Australian Government in threatened native vegetation on private 79

property in south-eastern Australia provided the motivation for this study (Commonwealth of 80

Australia 2009).This project, governed by NECMA, is analogous to the European Union’s 81

agri-environment scheme (AES) (Kleijn and Sutherland, 2003; Whittingham, 2007), which 82

involves paying landholders to adopt environmentally friendly farming practices. 83

84

Whilst biodiversity in protected areas is relatively well studied and monitored, many AES 85

have been criticized for their lack of monitoring and evaluation (Kleijn and Sutherland, 2003; 86

Tscharntke et al., 2005; Whittingham, 2007; Kleijn et al., 2011; Concepcion et al., 2012). 87

Furthermore, the majority of studies that have examined the effectiveness of AES involve 88

investigations on invertebrates (Fuentes‐Montemayor et al., 2011; Holland et al., 2012; 89

Delattre et al., 2013; Holland et al., 2014; Wood et al., 2015) or birds (Baker et al., 2012; 90

5

Lindenmayer et al., 2012a; MacDonald et al., 2012; Prince et al., 2012; Hiron et al., 2013; 91

Bright et al., 2015). This is, in part, because of the ease in which these taxa can be studied 92

(Whittingham, 2011). By contrast, studies on mammals (Broughton et al., 2014) and 93

ectothermic vertebrates are scarce and represent a key knowledge gap in the ecological 94

literature on agri-environmental schemes (Lindenmayer et al., 2012b; Michael et al., 2014) 95

and would add valuable information to the land sparing versus land sharing debate (Kleijn et 96

al., 2011; Habel et al., 2015). 97

98

We sought to address the question of whether the strategies of land sharing versus land 99

gazetted for conservation (land sparing) had complimentary vegetation patterns and richness 100

of vertebrate fauna in endangered eucalypt woodland communities. In doing so, we explored 101

two questions that underpin biodiversity conservation in commodity production landscapes: 102

1) Does native vegetation placed under AES support greater vertebrate species richness and 103

abundance than vegetation managed for primary production outcomes? 2) Do conservation 104

reserves support greater vertebrate species richness and abundance than sites under AES? 105

Remnant vegetation placed under a management agreement has the potential to support good 106

quality native vegetation and high levels of biodiversity (Lindenmayer et al., 2012a; Michael 107

et al., 2014). We postulated that management interventions such as reducing livestock 108

grazing pressure, controlling weeds of National significance and restricting the removal of 109

fallen timber are likely to result in improved native vegetation structure and condition, which 110

in turn, may lead to improved habitat values and positive biodiversity outcomes. We also 111

predicted that there would be a significant difference in vegetation structure and measures of 112

vertebrate diversity between sites managed for production outcomes, sites placed under 113

management agreements in 2010 (AES) and conservation reserves, in accordance with 114

6

differences in land use history and past disturbance regimes. With these two above questions 115

in mind, we aimed to evaluate the effectiveness of land sharing and land sparing strategies in 116

conserving and improving woodland biodiversity. 117

118

2. Methods 119

2.1. Study Area 120

Our study was conducted in the North East and Goulburn Broken catchment management 121

areas of Victoria. This region is bordered by the Murray River in the north, the township of 122

Merton in the south (36º 58′ 42′′ S 145º 42′ 33′′ E), Wises Creek Flora Reserve in the east 123

(36º 03′ 17′′ S 147º 12′ 58′′ E) and the township of Locksley in the west (36º 49′ 06′′ S 145º 124

18′ 31′′ E) (Figure 1). The predominant type of native vegetation in our study region is 125

temperate eucalypt woodland, of which, over 85% has been cleared for agriculture and 126

livestock grazing (Hobbs and Yates, 2000). The remaining stands off fragmented vegetation 127

include several endangered grassy woodland communities listed under the Environment 128

Protection Biodiversity Conservation Act 1999. The two endangered woodland communities 129

in our study area include Box Gum Grassy Woodland dominated by white box Eucalyptus 130

albens, yellow box E. melliodora and Blakely’s red gum E. blakelyi; and Buloke Woodland 131

dominated by Buloke Allocasuarina luehmannii and grey box E. microcarpa. The quality and 132

quantity of remnant vegetation in our study area varies according to land use history and past 133

clearing, with most remnant vegetation occurring on hillsides or along drainage lines. Sites 134

with minimal livestock grazing pressure are dominated by native grasses and forbs, whereas 135

intensive production sites with a history of fertilizer use are dominated by exotic annual 136

grasses and broad-leaved weeds. 137

7

2.2. Experimental design 138

Under the Australian Government’s Caring for Our Country initiative, the North East and 139

Goulburn Broken Catchment Management Authority (CMA) received a grant to improve and 140

protect 580 ha of endangered grassy woodland vegetation. Expressions of interest were 141

advertised and eligible landholders received funds and entered into management agreements 142

to undertake conservation actions such weed control and fencing native vegetation to exclude 143

or reduce livestock grazing pressure during the spring and summer months. 144

145

In April 2010, 13 discrete landscape units (paddocks containing threatened native vegetation) 146

that ranged in size from 9 ha - 200 ha (mean = 53 ha) were placed under a management 147

agreement (agri-environment scheme) and selected for biophysical monitoring. New fences 148

were constructed where necessary and new grazing regimes were adopted before biophysical 149

surveys were conducted. These areas were selected based on attaining benchmark vegetation 150

condition criteria, such that structurally diverse and floristically rich sites were targeted over 151

poorer condition sites. A single 200 m long transect (monitoring site) was placed randomly 152

within patches of native vegetation at a minimum distance of 50 m from the edge of the 153

remnant. These sites were paired with native vegetation managed for production purposes 154

(control) on the same property to account for inter-farm differences in management practices. 155

Control sites were selected to approximate the same vegetation type as AES but differed in 156

having below benchmark vegetation condition scores. In addition to these paired sites, we 157

selected the nearest conservation reserve/roadside reserve (protected area) within 15 km of 158

the same vegetation type to serve as reference sites. Conservation reserves in this region are 159

relatively small, have not been grazed for more than a decade and are subject to periodic 160

weed control. Thus, our study design included three different management classes: (1) AES 161

8

sites (N = 13) - remnant vegetation that was in former production areas and placed under a 10 162

year conservation management agreement with the CMA in 2010. These areas were formerly 163

set stocked, were not subject to targeted weed control and had no restrictions on timber 164

removal prior to this study. (2) Production sites (N = 10) - remnant vegetation managed for 165

primary production, with a set-stock grazing regime, and (3) Conservation reserves (N = 11) - 166

remnant vegetation in nature reserves managed by Parks Victoria. Thus, our study 167

encompassed 34 permanent sites on 18 farms (Figure 1). 168

169

2.3. Survey Protocols 170

We surveyed vegetation, amphibians, reptiles, arboreal marsupials and birds on four 171

occasions between May 2010 and December 2014. No data were collected in 2013 due to the 172

project transitioning to bi-annual surveys and a long-term monitoring phase.We collected 173

percentage cover abundance estimates for vegetation variables using the point intercept 174

method (Gibbons et al., 2008), whereby the presence/absence of vegetation attributes were 175

recorded ten times at one metre intervals. We applied this method twice along our 200 m 176

monitoring transects (between 0 - 50 m and 150 - 200 m) to calculate a percent score for 177

each attribute. The key vegetation variables we selected were those predicted to fluctuate 178

with season and management; including native grass, exotic grass, broad leaved weeds, bare 179

ground and leaf litter. We measured seedling recruitment of Eucalypt species and native 180

shrubs (e.g. Acacia sp.), and measured the length of logs (greater than 1 m) within two 50 x 181

20 m plots located at opposite ends of the transect. We also measured native plant species 182

richness from a 20 x 20 m plot located mid-way along the transect. 183

184

9

We surveyed amphibians and reptiles at each site once a year during August using time- and 185

area- constrained (30 min x 1 ha) active searches of natural habitat (200 m x 50 m search 186

area) and inspections of artificial refuges (AR) arrays (Michael et al., 2012). Each AR array 187

consisted of four railway sleepers (1.2 m in length), four roof tiles and one double stack of 1 188

m² corrugated steel sheet. AR arrays were established within the 1 ha search area and placed 189

100 m apart. Using a combination of active searches and AR survey techniques increases the 190

probability of detecting cryptic herpetofauna (Michael et al., 2012). Surveys were conducted 191

on clear days between 0900 - 1400 hours. 192

193

We conducted spotlighting surveys of arboreal marsupials once each year during November 194

using a time- and area- constrained (20 min x 1 ha) search protocol. We commenced 195

spotlighting surveys one hour after dusk and terminated five hours later to reduce observer 196

fatigue and potential bias in detectability. We used a hand held (9V) battery powered 197

spotlight to detect animal movement and/or eye shine. Only individuals within 50 m of the 198

transect were included in the analysis to avoid recording species in adjacent fields and 199

potentially under different land use. 200

201

Our bird counting protocols entailed conducting five minute point interval counts (Pyke and 202

Recher, 1983) at the 0 m, 100 m and 200 m points along a permanent transect. All species 203

seen or heard within 50 m of the transect were recorded and used in the analysis. To account 204

for observer bias and weather conditions, each site was surveyed by two different observers 205

on sequential days during June (Austral winter) and October (Austral spring) (Lindenmayer 206

et al., 2009). We completed counts between 0530 - 0930 hours and did not undertake surveys 207

10

during inclement weather. We classified birds of conservation concern based on Reid (1999) 208

or if they were listed under the NSW Threatened Species ACT 1995. 209

210

2.4. Data Analysis 211

To allow for between- and within-site variation, we fitted hierarchical generalized linear 212

models (Lee et al., 2006) to species richness and abundance. We used vegetation variables, 213

amphibian, reptile, bird and arboreal marsupial species richness and abundance as response 214

variables, and management class (treatment effect) and year (temporal effect) as predictors in 215

the analysis. For responses which were percentages, we used a quasi-binomial distribution 216

with a logit link function, and we used a beta distribution with a logit link function for the 217

random effects of site and plots within site where that was appropriate for birds and 218

vegetation variables. For counts, we assumed a quasi-Poisson distribution with a log link, and 219

a gamma distribution with a log link for the random effects. For fallen timber, we assumed 220

normal distributions for both the response and the random effects. The effective degrees of 221

freedom for the estimation of variance components, 25 for site and for plot within site and 222

more for other components, were large enough to confidently estimate variances and standard 223

errors with sufficient accuracy, overcoming the limitations of small sample size. We used 224

Wald tests for assessing the significance of effects. We restricted our analysis of arboreal 225

marsupials and birds to those detected within 50 m of each transect and excluded water birds 226

or species flying overhead unless they were obviously using the site. We combined data from 227

the two endangered woodland vegetation communities and performed all analyses using 228

GenStat v17. 229

230

3. Results 231

3.1. Vegetation variables 232

11

We found significant differences in the mean values of several vegetation variables among 233

management classes. The number of shrub seedlings ( 22χ = 12.27, P = 0.002; Figure 2b), the 234

percent cover of leaf litter ( 22χ = 15.63, P < 0.001; Figure 2g) and native plant species 235

richness ( 22χ = 9.7, P < 0.05) was highest in conservation reserves; the percent cover of exotic 236

broad leaved weeds ( 22χ = 9.98, P = 0.007; Figure 2c) was highest on production sites, and 237

the percent cover of native grass ( 22χ = 6.65, P = 0.036; Figure 2e) was highest on AES. We 238

found significant temporal differences in cover abundance estimates for all variables 239

measured. Overall, the percent cover of broad leaved weeds ( 23χ = 42.86, P < 0.001; Figure 240

2c) and bare ground ( 23χ = 79.19, P < 0.001; Figure 2h) declined between 2010 and 2014, and 241

the amount of fallen timber increased between 2010 and 2014 ( 23χ = 56.82, P < 0.001; Figure 242

2d). The percent cover of native grass ( 23χ = 117.79, P < 0.001; Figure 2e) peaked in 2011, 243

whereas the percent cover of exotic grass ( 23χ = 130.71, P < 0.001; Figure 2f) was lowest in 244

2011. We also found significant positive interactions between management and survey year 245

for several variables. Overstorey eucalypt recruitment was significantly high in conservation 246

reserves in 2012 ( 26χ = 34.34, P < 0.001; Figure 2a), the cover of exotic broad leaved weeds 247

was significantly low in production sites in 2014 ( 26χ = 26.22, P < 0.001; Figure 2c) and 248

amounts of fallen timber was significantly high in conservation reserves in 2014 ( 26χ = 26.83, 249

P < 0.001; Figure 2d). 250

251

3.2. Amphibian species richness and abundance 252

We obtained 186 observations of eight frog species from three families (see supplementary 253

data Table S1). Approximately 75% of all observations were of two species; Eastern Banjo 254

Frog Limnodynastes dumerilii and Spotted Marsh Frog L. tasmaniensis. We found no 255

12

significant difference in amphibian species richness ( 22χ = 1.20, P = 0.548; Figure 3a) or 256

abundance ( 22χ = 3.99, P = 0.136; Figure 3b) among management classes. We identified 257

highly significant differences in species richness ( 23χ = 12.09, P < 0.01; Figure 3a) and 258

abundance ( 23χ = 40.43, P < 0.001; Figure 3b) between years. Amphibian species richness 259

and abundance peaked in 2011 and declined thereafter. We found no significant interaction 260

between management and year for amphibian species richness ( 26χ = 2.59, P = 0.857; Figure 261

3a) or abundance ( 26χ = 5.09, P = 0.532; Figure 3b). 262

263

3.3. Reptile species richness and abundance 264

We obtained 584 observations of 18 reptile species from five families (see supplementary 265

data Table S1). Approximately 70% of all observations were of four species; Boulenger’s 266

Skink Morethia boulengeri (40%), Eastern Striped Skink Ctenotus spaldingi (16%), Three-267

toed Earless Skink Hemiergis talbingoensis (14%) and the Ragged Snake-eyed Skink 268

Cryptoblepharus pannosus (7%). We found reptile species richness increased along a 269

gradient from production sites, AES sites to conservation reserves. However, differences 270

among management types were not statistically significant (species richness; 22χ = 3.91, P = 271

0.142; Figure 3c: species abundance; 22χ = 405, P = 0.132, Figure 3d). By contrast, we 272

identified highly significant differences in reptile species richness ( 23χ = 27.36, P < 0.001; 273

Figure 3c) and abundance ( 23χ = 23.33, P < 0.001; Figure 3d) between years. Reptile species 274

richness and abundance peaked during the 2011 and 2012 surveys. We found no significant 275

interaction between management and year for reptile species richness ( 26χ = 4.03, P = 0.672; 276

Figure 3c) or abundance ( 26χ = 4.56, P = 0.601; Figure 3d). 277

13

3.4. Bird species richness 278

We recorded 4517 observations of 121 bird species from 37 families (see supplementary data 279

Table S1). We found no significant difference in total bird species richness ( 22χ = 2.36, P = 280

0.307; Figure 3e) among management classes. We identified highly significant differences in 281

bird species richness among years ( 23χ = 84.10, P < 0.001; Figure 3e) and seasons ( 2

1χ = 282

12.30, P < 0.001). More bird species were detected in 2011 and during the spring surveys. 283

We found no significant interaction between management and survey year ( 26χ = 8.57, P = 284

0.192; Figure 3e) or between management and season ( 22χ = 1.81, P = 0.418). However, we 285

identified a significant interaction between year and season ( 23χ = 23.46, P < 0.001), where 286

more bird species were detected in spring 2011. 287

288

When we examined 16 bird species of conservation concern, we identified a significant 289

difference in species richness among management classes ( 22χ = 6.63, P = 0.036; Figure 3f) 290

and survey year ( 23χ = 15.375, P = 0.002; Figure 3f), but not between seasons ( 2

1χ = 0.96, P = 291

0.333). Furthermore, we found no significant interaction between management and survey 292

year ( 26χ = 8.09, P = 0.233; Figure 3f), management and season ( 2

2χ = 0.16, P = 0.927), or 293

survey year and season ( 23χ = 2.35, P = 0.501). Overall, significantly more bird species of 294

conservation concern were recorded on AES and conservation reserves than production sites. 295

296

3.5. Arboreal marsupial species richness and abundance 297

14

We recorded 256 observations of six species of arboreal marsupial from four families (see 298

supplementary data Table S1). More than 80% of all observations were of the Common 299

Brushtail Possum Trichosurus vulpechula and the Common Ring-tailed Possum 300

Pseudocheirus peregrinus. We identified a significant difference in arboreal marsupial 301

species richness among management classes ( 22χ = 9.45, P = 0.009; Figure 3g) but not 302

between years ( 23χ = 6.64, P = 0.084) (Figure 6). On average, significantly more species were 303

detected in AES compared to production sites and significantly more species were detected in 304

conservation reserves than AES. We found no significant difference in arboreal marsupial 305

abundance among management classes ( 22χ = 5.49, P = 0.064; Figure 3h), although we 306

identified a significant difference in arboreal marsupial abundance between years ( 23χ = 307

14.24, P < 0.01; Figure 3h). The numbers of individuals recorded were highest in 2010 and 308

2012. We found no significant interaction between management and survey year for arboreal 309

marsupial species richness ( 26χ = 4.04, P = 0.671; Figure 3g). However, we identified a 310

significant interaction between management and survey year for arboreal marsupial 311

abundance ( 26χ = 15.34, P = 0.018; Figure 3h), where significantly fewer individuals were 312

recorded on production sites in 2012. 313

314

4. Discussion 315

We sought to evaluate the influence of different land management uses on vegetation patterns 316

and vertebrate species richness and occurrence in endangered temperate eucalypt woodland 317

vegetation communities. Under both land sparing (conservation reserves) and land sharing 318

(agri-environment scheme) strategies, comparatively high numbers of birds of conservation 319

concern were supported than compared to agricultural production sites. Likewise, arboreal 320

15

marsupial species richness and abundance was highest under a land sparing strategy, but AES 321

still supported significantly more arboreal marsupial species than agricultural production 322

sites. The relative benefits of AES or conservation reserves for supporting high herpetofaunal 323

diversity was not apparent in this study. 324

325

Agri-environment schemes are increasingly being adopted around the world to maintain 326

agricultural productivity while at the same time integrating biodiversity conservation – a key 327

premise of land sharing strategies (Fischer et al., 2008; Phalan et al., 2011b). However, it is 328

recognized that land sharing schemes may not guarantee benefits to all biodiversity in all 329

agricultural landscapes (Kleijn et al., 2001; Kleijn et al., 2006; Phalan et al., 2011a; Michael 330

et al., 2014). Furthermore, conservation reserves may not comprehensively protect 331

biodiversity dependent on some endangered ecological communities (Scott et al., 2001; 332

Rodrigues et al., 2004). We highlight the contrasting responses of different taxa to these two 333

strategies. In the remainder of this paper we discuss the influence of these land use strategies 334

on patterns of vegetation and vertebrate diversity and suggest ways to improve future AES to 335

maximize the protection of biodiversity in agricultural landscapes. 336

337

4.1. Effect of land management on vegetation patterns 338

Protected areas such as conservation reserves and roadsides are predicted to have greater 339

vegetation structural complexity and floristic diversity than remnant vegetation in agricultural 340

landscapes due to the exclusion of livestock grazing, logging and soil disturbances (Bennett, 341

1991). Our results indicate that conservation reserves in our study area are floristically rich, 342

support high percent cover of leaf litter and more recruitment of eucalypt saplings and shrub 343

16

seedlings than AES or production sites (Figure 2a, b). This pattern is consistent with the 344

response of native vegetation to reduced livestock grazing pressure (Dorrough & Moxham, 345

2005). By contrast, we found AES supported greater cover of native grass than conservation 346

reserves or production sites (Figure 2e), a result we expected considering areas of high 347

quality native vegetation were targeted for investment. The temporal changes in vegetation 348

patterns we observed in this study reflect seasonal rainfall patterns; noticeably with a peak 349

and trough in native and exotic grass cover in 2011, a year of above average high rainfall. 350

Our findings also highlight the complex interaction between climate and management. For 351

example, eucalypt recruitment and fallen timber were significantly higher in conservation 352

reserves during 2011 and 2014 respectively (Figure 2a, d). This suggests that some attributes 353

of native vegetation (e.g. eucalypt recruitment) respond markedly better under different 354

management scenarios during climatically favorable years. 355

356

4.2. Effect of land management on patterns of vertebrate diversity 357

At the outset of our study, we predicted that AES would support greater vertebrate richness 358

than production sites. A second prediction was that conservation reserves would support 359

greater vertebrate richness than AES sites. Our data provide inconclusive evidence to 360

determine which strategy is best for amphibians or reptiles. On average, reptile species 361

richness and abundance was highest on AES compared to controls, and conservation reserves 362

supported greater reptile diversity than AES sites (Figure 3c, d). However, the differences 363

among these management classes were not statistically significant. This result is congruent 364

with other studies on ectotherms that report low site-level species richness, an artifact of past 365

habitat loss and widespread declines in herpetofauna (Brown et al., 2008; Mac Nally et al., 366

2014). 367

17

Despite low site-level species richness, our data do not concur with other researchers that 368

small conservation reserves (including small linear road reserves) play a disproportionately 369

more important role in protecting reptiles in fragmented agricultural landscapes than other 370

land tenures (Driscoll, 2004; Brown et al., 2008; Mendenhall et al., 2014). Our findings 371

suggest, by lack of statistical significance among treatments, that reptiles are capable of 372

persisting in some production landscapes (albeit at low densities), particularly if habitat 373

requirements are met (Fischer et al., 2005; Michael et al., 2015) and grazing pressure remains 374

low (Howland et al., 2014). In a similar study, Michael et al. (2014) found that reptile species 375

richness and composition were similar between production areas and sites under an agri-376

environment scheme in southern New South Wales. Considering the lack of difference in 377

reptile diversity between management classes, conservation efforts to improve reptiles will 378

need to extend beyond conservation reserves or patches of remnant vegetation and consider 379

approaches to improve habitat in the interconnecting farming matrix. This will include 380

improving the carrying capacity of crops and grazing land for reptile biota by restoring 381

critical ground cover habitat such as fallen timber and bush rock, two critical resources upon 382

which many reptiles depend (Michael et al. 2015). 383

384

With respect to birds, we found no statistically significant difference in mean species richness 385

among management classes (Figure 3e). However, when we restricted the analysis to birds of 386

conservation concern, we found that AES and conservation reserves supported, on average, 387

three times more species than production sites (Figure 3f). This finding highlights the value 388

of AES and protected areas in protecting a suite of birds that have declined since European 389

settlement (Ried, 1999; Montague-Drake et al., 2009). This finding also emulates patterns 390

reported in another study from southern NSW which found the same suite of birds benefitted 391

18

from AES (Lindenmayer et al., 2012a). In the previous study, modified grazing regimes led 392

to an increase in shrub recruitment and improved structural complexity, key predictors of the 393

presence of declining woodland birds (Montague-Drake et al., 2009; Lindenmayer et al., 394

2012a). However, short-term land management agreements (10 years in most cases) may not 395

provide native vegetation with enough time to recover from intensive grazing pressure. We 396

argue, that inperpetuity conservation covenants with private landholders are required to 397

provide a long-term solution to native vegetation regeneration and habitat rehabilitation on 398

farms. 399

400

Arboreal marsupials were another group to differ statistically in measures of diversity among 401

management classes. In this group, we found a strong gradient in species richness and 402

abundance from production sites to conservation reserves, the difference between 403

management classes being almost twofold (Figure 3g, h). With the exception of the Koala 404

Phascolarctos cinereus, all species of arboreal marsupial detected in this study are hollow-405

dependant (Gibbons and Lindenmayer, 2002). The relationship between arboreal marsupial 406

density, patch size and the number of den sites (tree hollows) is well documented (van der 407

Ree et al., 2001; Harper et al., 2008; Crane et al., 2010) as is the importance of maintaining 408

roadside vegetation and conservation reserves (Crane et al., 2014). However, a key challenge 409

in conserving arboreal marsupials in agricultural landscapes is to improve habitat 410

connectivity (Cunningham et al., 2007). Traditionally, this has proved to be a challenge for 411

land managers as it often requires adopting management practices that facilitate eucalypt 412

regeneration such as modifying grazing regimes at a farm- and landscape-scale (Ikin et al., 413

2015), or undertaking extensive tree plantings. Instead, alternative strategies to increase 414

arboreal marsupial habitat may include increasing the width of roadside vegetation (van der 415

19

Ree et al., 2001; Crane et al., 2014) and expanding vegetation around existing conservation 416

reserves. 417

418

4.3. Trends over time 419

Climate changes associated with El Nino Southern Oscillation can have predictable effects on 420

wildlife populations in landscapes where productivity is strongly limited by precipitation 421

(Holmgren et al., 2006). For most vegetation variables and three of four vertebrates groups, 422

we found highly significant differences in species richness among years (Figures 2 and 3). 423

These patterns are consistent with increased population numbers (and detectability) following 424

the Millennium drought which ended in 2010, and subsequent above average rainfall across 425

the study area in the years 2010 to 2012 (the mean average rainfall for Benalla, a town 426

situated in the centre of our study area declined significantly since 2012, Bureau of 427

Meteorology, 2015). Furthermore, temporal variation in measures of diversity between 428

management types highlights the importance of considering management responses in 429

different years when designing future AES. For example, we found the abundance of arboreal 430

marsupials declined markedly on production sites in 2012 (Figure 3h). This pattern may have 431

been driven by differences in predator-prey relationships between sites where feral predatory 432

animals were controlled (e.g. AES) and sites where predators were not controlled (e.g. 433

production sites). 434

435

The relationship between rainfall parameters and population-level increases is well 436

documented for amphibians (Wassens et al., 2013; Mac Nally et al., 2014). For example, 437

periods of above average rainfall can prolong breeding events and breeding success, which 438

20

can lead to an increase in adult frog densities and detection rates (Mac Nally et al., 2014). 439

Our amphibian data were partly driven by the abundant spotted marsh frog Limnodynastes 440

tasmaniensis, a habitat generalist that has good dispersal ability and opportunistically breeds 441

in rain-fed ponds (Wassens et al., 2013). In addition, the relationship between rainfall 442

patterns and reptile diversity is more complex and less well understood (see Read et al., 2012; 443

Pastro et al., 2013). In one study, Dickman et al. (1999) found juvenile abundance of two 444

agamid species was related to rainfall in the preceding summer and autumn, consistent with 445

enhanced survival, growth and clutch size. However, within-year rainfall also drove changes 446

in vegetation cover, which led to a shift in the relative abundance of the two species 447

(Dickman et al., 1999). The peak and subsequent decline in the detection rate of herpetofauna 448

in this study is likely to be associated with post-drought rainfall patterns, suggesting it may 449

take many years before the causal effects of management can be separated from the 450

background effects of climate. 451

452

4.4. Conclusion 453

Our findings from an agricultural landscape in south-eastern Australia suggest that land 454

sharing and land sparing have contrasting benefits for farm biodiversity. Land sharing 455

strategies, under an agri-environment scheme, and conservation reserves can support 456

relatively species rich assemblages of birds, including many species of conservation concern. 457

However, AES may have only limited benefit for protecting populations of arboreal 458

marsupials due to the lack of hollow-bearing trees in agricultural landscapes. Furthermore, 459

the comparative benefit of land sharing and land sparing for maintaining herpetofaunal 460

diversity remains inconclusive, and future incentive schemes may need to focus on improving 461

the carrying capacity and dispersal potential for reptiles in the broader agricultural matrix. 462

21

Another important consideration for future AES is to take into account interactions between 463

year (and associated climatic conditions) and management because differences in measures of 464

diversity may be masked in poor (drought years) compared to above average rainfall years. 465

466

Acknowledgments 467

This study was funded by the North East and Goulburn Broken Catchment Management 468

Authorities, the Australian Research Council and the Australian Government’s Caring for our 469

Country initiative. We thank Greta Quinlivan, Mary Munroe, Jenny Wilson and Steve Wilson 470

for supporting this work. Jacqui Slingo, Lachlan McBurney, David Blair, Mason Crane, 471

Sachiko Okada, Christopher MacGregor, Daniel Florance, Scott Lucas and Malcom Miles 472

assisted with field work. Surveys were conducted in accordance with The Australian National 473

University Animal Care and Ethics protocols (F.ES.01.10) under a scientific research permit 474

issued by the Department of Sustainability and Environment (10005355). 475

476

References 477

Baker, D.J., Freeman, S.N., Grice, P.V.,Siriwardena, G.M. 2012. Landscape‐scale responses 478

of birds to agri‐environment management: a test of the English Environmental 479

Stewardship scheme. J. Appl. Ecol. 49, 871-882.Bennett, A.F. 1991. Roads, 480

roadsides, and wildlife conservation: a review. In Nature conservation 2: the role of 481

corridors (Eds D.A.Saunders R.J. Hobbs), pp 99 -118, Surrey Beatty and Sons, 482

Chipping Norton, New South Wales, Australia. 483

22

Bright, J.A., Morris, A.J., Field, R.H., Cooke, A.I., Grice, P.V., Walker, L.K., Fern, J.,Peach, 484

W.J. 2015. Higher-tier agri-environment scheme enhances breeding densities of some 485

priority farmland birds in England. Agric. Ecosyst. Environ. 203, 69-79. 486

Broughton, R.K., Shore, R.F., Heard, M.S., Amy, S.R., Meek, W.R., Redhead, J.W., Turk, 487

A., Pywell, R.F. 2014. Agri-environment scheme enhances small mammal diversity 488

and abundance at the farm-scale. Agric. Ecosyst. Environ. 192, 122-129. 489

Brown, G.W., Bennett, A.F., Potts, J.M. 2008. Regional faunal decline–reptile occurrence in 490

fragmented rural landscapes of south-eastern Australia. Wildl. Res. 35, 8-18. 491

Brussaard, L., Caron, P., Campbell, B., Lipper, L., Mainka, S., Rabbinge, R., Babin, D., 492

Pulleman, M. 2010. Reconciling biodiversity conservation and food security: 493

scientific challenges for a new agriculture. Curr. Opin. Environ. Sustain. 2, 34-42. 494

Chappell, M.J., LaValle, L.A. 2011. Food security and biodiversity: can we have both? An 495

agroecological analysis. Agric. Human Values, 28, 3-26. 496

Commonwealth of Australia (2009) Environmental Stewardship Strategic Framework. 497

Canberra: Commonwealth of Australia. 498

Concepcion, E.D., Diaz, M., Kleijn, D., Baldi, A., Batary, P., Clough, Y., Gabriel, D., 499

Herzog, F., Holzschuh, A., Knop, E. 2012. Interactive effects of landscape context 500

constrain the effectiveness of local agri‐environmental management. J. Appl. Ecol., 501

49, 695-705. 502

Crane, M.J., Lindenmayer, D.B., Cunningham, R.B. 2010. The use of den trees by the 503

squirrel glider (Petaurus norfolcensis) in temperate Australian woodlands. Aust. J. 504

Zool. 58, 39-49. 505

Crane, M.J., Lindenmayer, D.B., Cunningham, R.B. 2014. The value of countryside elements 506

in the conservation of a threatened arboreal marsupial Petaurus norfolcensis in 507

23

agricultural landscapes of south-eastern Australia - The disproportional value of 508

scattered trees. PloS One 9, e107178. 509

Cunningham, R.B., Lindenmayer, D.B., Crane, M., Michael, D.,MacGregor, C. 2007. Reptile 510

and arboreal marsupial response to replanted vegetation in agricultural landscapes. 511

Ecol. Appl. 17, 609-619. 512

Delattre, T., Vernon, P., Burel, F. 2013. An agri-environmental scheme enhances butterfly 513

dispersal in European agricultural landscapes. Agric. Ecosyst. Environ. 166, 102-109. 514

Dickman, C.R., Letnic, M., Mahon, P.S. 1999. Population dynamics of two species of dragon 515

lizards in arid Australia: the effects of rainfall. Oecologia 119, 357-366. 516

Dorrough, J., Moxham, C. 2005. Eucalypt establishment in agricultural landscapes and 517

implications for landscape-scale restoration. Biol. Conserv. 123, 55-66. 518

Driscoll, D.A. 2004. Extinction and outbreaks accompany fragmentation of a reptile 519

community. Ecol. Appl. 14, 220-240.Dudley, N., Parrish, J.D., Redford, K., Stolton S. 520

2010. The revised IUCN protected area management categories: the debate and ways 521

forward. Oryx 44, 485-490. 522

Fischer, J., Brosi, B., Daily, G.C., Ehrlich, P.R., Goldman, R., Goldstein, J., Lindenmayer, 523

D.B., Manning, A.D., Mooney, H.A., Pejchar, L., Ranganathan, J., Tallis, H. 2008. 524

Should agricultural policies encourage land sparing or wildlife-friendly farming? 525

Front. Ecol. Environ. 6, 380-385. 526

Fischer, J., Fazey, I., Briese, R., Lindenmayer, D.B. 2005. Making the matrix matter: 527

challenges in Australian grazing landscapes. Biodivers. Conserv. 14, 561-578. 528

Fuentes‐Montemayor, E., Goulson, D., Park, K.J. 2011. The effectiveness of 529

agri‐environment schemes for the conservation of farmland moths: assessing the 530

importance of a landscape‐scale management approach. J. Appl. Ecol. 48, 532-542. 531

24

Gibbons, P., Lindenmayer, D. 2002. Tree hollows and wildlife conservation in Australia. 532

CSIRO publishing. Collingwood, Melbourne. 533

Gibbons, P., Briggs, S.V., Ayers, D.A., Doyle, S., Seddon, J., McElhinny, C., Jones, N., 534

Sims, R. and Doody, J.S. 2008. Rapidly quantifying reference conditions in modified 535

landscapes. Biol. Conserv. 141, 2483-2493. 536

Godfray, H.C.J., Beddington, J.R., Crute, I.R., Haddad, L., Lawrence, D., Muir, J.F., Pretty, 537

J., Robinson, S., Thomas, S.M., Toulmin, C. 2010. Food security: the challenge of 538

feeding 9 billion people. Science 327, 812-818. 539

Habel, J., Teucher, M., Hornetz, B., Jaetzold, R., Kimatu, J., Kasili, S., Mairura, Z., Mulwa, 540

R., Eggermont, H., Weisser, W., Lens, L. 2015. Real-world complexity of food 541

security and biodiversity conservation. Biodivers. Conserv. DOI 10.1007/s10531-015-542

0866-z. 543

Harper, M.J., McCarthy, M.A., van der Ree, R. 2008. Resources at the landscape scale 544

influence possum abundance. Austral Ecol. 33, 243-252. 545

Hiron, M., Berg, A., Eggers, S., Josefsson, J., Part, T. 2013. Bird diversity relates to agri-546

environment schemes at local and landscape level in intensive farmland. Agric. 547

Ecosyst. Environ. 176, 9-16. 548

Hobbs, R.J., Yates, C.J. 2000. Temperate eucalypt woodlands in Australia: biology, 549

conservation, management and restoration. Surrey Beatty and Sons, Chipping Norton, 550

Sydney. 551

Holland, J., Oaten, H., Moreby, S., Birkett, T., Simper, J., Southway, S., Smith, B. 2012. 552

Agri-environment scheme enhancing ecosystem services: A demonstration of 553

improved biological control in cereal crops. Agric. Ecosyst. Environ. 155, 147-152. 554

25

Holland, J.M., Storkey, J., Lutman, P.J.W., Birkett, T.C., Simper, J., Aebischer, N.J. 2014. 555

Utilisation of agri-environment scheme habitats to enhance invertebrate ecosystem 556

service providers. Agric. Ecosyst. Environ. 183, 103-109. 557

Holmgren, M., Stapp, P., Dickman, C.R., Gracia, C., Graham, S., Gutierrez, J.R., Hice, C., 558

Jaksic, F., Kelt, D.A., Letnic, M. 2006. Extreme climatic events shape arid and 559

semiarid ecosystems. Front. Ecol. Environ. 4, 87-95. 560

Howland, B., Stojanovic, D., Gordon, I.J., Manning, A.D., Fletcher, D., Lindenmayer, D.B. 561

2014. Eaten out of house and home: Impacts of grazing on ground-dwelling reptiles in 562

Australian grasslands and grassy woodlands. PloS One 9, e105966. 563

Ikin, K., Mortelliti, A., Stein, J., Michael, D., Crane, M., Okada, S., Wood, J., Lindenmayer, 564

D. 2015. Woodland habitat structures are affected by both agricultural land 565

management and abiotic conditions. Landscape Ecol. 30, 1387-1403. 566

Kleijn, D., Baquero, R., Clough, Y., Diaz, M., Esteban, J.D., Fernandez, F., Gabriel, D., 567

Herzog, F., Holzschuh, A., Johl, R. 2006. Mixed biodiversity benefits of 568

agri‐environment schemes in five European countries. Ecol. Letters 9, 243-254. 569

Kleijn, D., Berendse, F., Smit, R., Gilissen, N. 2001. Agri-environment schemes do not 570

effectively protect biodiversity in Dutch agricultural landscapes. Nature 413, 723-725. 571

Kleijn, D., Sutherland, W.J. 2003. How effective are European agri‐environment schemes in 572

conserving and promoting biodiversity? J. Appl. Ecol. 40, 947-969. 573

Kleijn, D., Rundlof, M., Scheper, J., Smith, H.G., Tscharntke, T. 2011. Does conservation on 574

farmland contribute to halting the biodiversity decline? Trends Ecol. Evol. 26, 474-575

481. 576

Lee, Y., Nelder, J.A., Pawitha, Y. 2006. Generalized linear models with random effects : 577

unified analyses via h-likelihood. Chapman & Hall/CDC, New York. 578

26

Lindenmayer, D.B., Wood, J., MacGregor, C. 2009. Do observer differences in bird 579

detection significantly influence inferences about environmental impacts? Emu 109, 580

100-106. 581

Lindenmayer, D., Wood, J., Montague-Drake, R., Michael, D., Crane, M., Okada, S., 582

MacGregor, C., Gibbons, P. 2012a. Is biodiversity management effective? Cross-583

sectional relationships between management, bird response and vegetation attributes 584

in an Australian agri-environment scheme. Biol. Conserv. 152, 62-73. 585

Lindenmayer, D.B., Zammit, C., Attwood, S.J., Burns, E., Shepherd, C.L., Kay, G., Wood, J. 586

2012b. A novel and cost-effective monitoring approach for outcomes in an Australian 587

biodiversity conservation incentive program. PloS One 7, e50872. 588

Mac Nally, R., Nerenberg, S., Thomson, J.R., Lada, H., Clarke, R.H. 2014. Do frogs bounce, 589

and if so, by how much? Responses to the ‘Big Wet’ following the ‘Big Dry’ in 590

south‐eastern Australia. Global Ecol. Biogeogr. 23, 223-234. 591

MacDonald, M.A., Maniakowski, M., Cobbold, G., Grice, P.V., Anderson, G.Q.A. 2012. 592

Effects of agri-environment management for stone curlews on other biodiversity. 593

Biol. Conserv. 148, 134-145. 594

Mendenhall, C.D., Frishkoff, L.O., Santos-Barrera, G., Pacheco, J., Mesfun, E., Quijano, 595

F.M., Erhlich, P.R., Ceballos, G., Daily, G.C., Pringle, R.M. 2014. Countryside 596

biogeography of Neotropical reptiles and amphibians. Ecology 95, 856-870. 597

Michael, D.R., Cunningham, R.B., Donnelly, C.F., Lindenmayer, D.B. 2012 Comparative use 598

of active searches and artificial refuges to survey reptiles in temperate eucalypt 599

woodlands. Wildl. Res. 39, 149-162. 600

Michael, D.R., Kay, G.M., Crane, M., Florance, D., MacGregor, C., Okada, S., McBurney, 601

L., Blair, D., Lindenmayer, D.B. 2015 Ecological niche breadth and microhabitat 602

27

guild structure in temperate Australian reptiles: Implications for natural resource 603

management in endangered grassy woodland ecosystems. Austral Ecol. 51, 494-504. 604

Michael, D.R., Wood, J.T., Crane, M., Montague-Drake, R., Lindenmayer, D.B. 2014. How 605

effective are agri-environment schemes for protecting and improving herpetofaunal 606

diversity in Australian endangered woodland ecosystems? J. Appl. Ecol. 51, 494-504. 607

Montague-Drake, R., Lindenmayer, D., Cunningham, R. 2009 Factors affecting site 608

occupancy by woodland bird species of conservation concern. Biol. Conserv. 142, 609

2896-2903. 610

Pastro, L. A., Dickman, C. R., Letnic, M. 2013. Effects of wildfire, rainfall and region on 611

desert lizard assemblages: the importance of multi-scale processes. Oecologia 173, 612

603-614. 613

Phalan, B., Balmford, A., Green, R.E, Scharlemann, J.P. 2011a. Minimising the harm to 614

biodiversity of producing more food globally. Food Policy 36, S62-S71. 615

Phalan, B., Onial, M., Balmford, A., Green, R.E. 2011b. Reconciling food production and 616

biodiversity conservation: Land sharing and land sparing compared. Science 333, 617

1289-1291. 618

Prince, K., Moussus, J.P., Jiguet, F. 2012 Mixed effectiveness of French agri-environment 619

schemes for nationwide farmland bird conservation. Agric. Ecosyst. Environ. 149, 74-620

79. 621

Pyke, G.H., Recher, H.F. 1983. Censusing Australian birds: a summary of procedures and a 622

scheme for standardisation of data presentation and storage, in: Davies, S.J. (Ed.), 623

Methods of censusing birds in Australia. Proceedings of a symposium organised by 624

the Zoology section of the ANZAAS and the Western Australian Group of the Royal 625

Australasian Ornithologists Union. Department of Conservation and Environment, 626

Perth, pp. 55-63. 627

28

Read, J.L., Kovac, K.J., Brook, B.W., Fordham, D.A. 2012. Booming during a bust: 628

asynchronous population responses of arid zone lizards to climatic variables. Acta 629

Oecologica 40, 51-61. 630

Reid, J.R.W. 1999. Threatened and declining birds in the New South Wales sheep-wheat belt: 631

I. diagnosis, characteristics and management. Consultancy report to NSW National 632

Parks and Wildlife Service. CSIRO Wildlife and Ecology, Canberra. 633

Rodrigues, A.S., Andelman, S.J., Bakarr, M.I., Boitani, L., Brooks, T.M., Cowling, R.M., 634

Fishpool, L.D., da Fonseca, G.A., Gaston, K.J., Hoffmann, M. 2004. Effectiveness of 635

the global protected area network in representing species diversity. Nature 428, 640-636

643. 637

Scott, J.M., Davis, F.W., McGhie, R.G., Wright, R.G., Groves, C., Estes, J. 2001. Nature 638

reserves: Do they capture the full range of America's biological diversity? Ecol. Appl. 639

11, 999-1007. 640

Tilman, D., Cassman, K.G., Matson, P.A., Naylor, R., Polasky, S. 2002. Agricultural 641

sustainability and intensive production practices. Nature 418, 671-677. 642

Tscharntke, T., Clough, Y., Wanger, T.C., Jackson, L., Motzke, I., Perfecto, I., Vandermeer, 643

J., Whitbread, A. 2012. Global food security, biodiversity conservation and the future 644

of agricultural intensification. Biol. Conserv. 151, 53-59. 645

Tscharntke, T., Klein, A.M., Kruess, A., Steffan-Dewenter, I., Thies, C. 2005. Landscape 646

perspectives on agricultural intensification and biodiversity - Ecosystem service 647

management. Ecol. Letters 8, 857-874. 648

van der Ree, R., Soderquist, T.R.,Bennett, A.F. 2001. Home-range use by the brush-tailed 649

phascogale (Phascogale tapoatafa) (Marsupialia) in high-quality, spatially limited 650

habitat. Wildl. Res. 28, 517-525. 651

29

Wassens, S., Walcott, A., Wilson, A., Freire, R. 2013. Frog breeding in rain-fed wetlands 652

after a period of severe drought: implications for predicting the impacts of climate 653

change. Hydrobiologia 708, 69-80. 654

Whittingham, M.J. 2007. Will agri‐environment schemes deliver substantial biodiversity 655

gain, and if not why not? J. Appl. Ecol. 44, 1-5. 656

Whittingham, M.J. 2011. The future of agri‐environment schemes: biodiversity gains and 657

ecosystem service delivery? J. Appl. Ecol. 48, 509-513. 658

Wood, T.J., Holland, J.M., Hughes, W.O., Goulson, D. 2015. Targeted agri‐environment 659

schemes significantly improve the population size of common farmland bumblebee 660

species. Mol. Ecol. 24, 1668-1680. 661

662

663

664

665

666

667

668

669

670

671

672

30

Figure 1. Location of long-term biodiversity monitoring sites in North-eastern Victoria, 673

south-eastern Australia. Reference sites represent conservation reserves, controls represent 674

intensively grazed production sites, and CMA represent sites funded under an agri-675

environment scheme. 676

Figure 2. Predicted mean values (95% CI) for eight vegetation variables and the interaction 677

between management class and survey year in North East Victoria, south-eastern Australia. 678

Reference sites represent conservation reserves, controls represent intensively grazed 679

production sites, and AES represent sites funded under an agri-environment scheme. 680

Figure 3. Predicted mean species richness and abundance (95% CI) of all vertebrate groups 681

and the interaction between management class and survey year in North East Victoria, south-682

eastern Australia. Reference sites represent conservation reserves, controls represent 683

intensively grazed production sites, and AES represent sites funded under an agri-684

environment scheme. 685

686

687

31

688

32

689

690

33

691

692