SEARCH

SEARCH BY CITATION

Keywords:

  • agricultural practices;
  • community composition;
  • management intervention;
  • reptile conservation;
  • species richness;
  • temperate woodland;
  • vegetation management

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  1. Agricultural intensification is a major cause of reptile and amphibian decline world-wide, prompting concern on how to best protect biodiversity in commodity production landscapes and meet global food demands. Agri-environment schemes (AES) attempt to integrate biodiversity conservation in agricultural landscapes but are often compromised by a lack of baseline data, monitoring and evaluation. Few studies have examined the benefits of AES in protecting biodiversity relative to the wider farming landscape, and no studies have explicitly quantified the effectiveness of AES to increase herpetofaunal diversity.
  2. To examine whether AES protect and increase herpetofauna, we established a landscape-wide biodiversity monitoring programme in threatened semi-arid and temperate woodland communities in south-eastern Australia.
  3. With 31 species recorded, regional herpetofaunal diversity was relatively high, whereas local diversity was low. Herpetofaunal richness and reptile assemblage structure did not differ significantly between sites under AES and sites managed for livestock production. A gradient in species richness as a function of time-since-management intervention was not evident, although the abundance of one lizard species increased under vegetation management. Reptile richness and frog abundance differed significantly among vegetation types.
  4. Herpetofaunal richness was positively related to native plant richness and bare ground cover, whereas Boulenger's skink Morethia boulengeri was negatively affected by bare ground cover. The ragged snake-eyed skink Cryptoblepharus pannosus was positively related to the amount of woody debris.
  5. Synthesis and applications. In this system, strong habitat specificity implies local-scale management interventions under agri-environment schemes (AES) may not significantly increase herpetofaunal diversity in the short term. Vegetation management is likely to increase the abundance of common lizard species rather than increase local species richness due to barrier effects. Future incentive schemes should focus on improving habitat connectivity, enhancing pasture condition and increasing woody debris in the agricultural matrix to dissolve dispersal barriers and mitigate the legacy of historical land-use practices. We propose that AES, which manage mosaics of intergrading vegetation types at multiple spatial scales, will protect maximum herpetofaunal diversity. These recommendations have implications for AES world-wide, many of which currently do not adequately address the habitat requirements of herpetofauna.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Reptile and amphibian populations are experiencing unprecedented declines world-wide (Gibbons et al. 2000; Houlahan et al. 2000; Araujo, Thuiller & Pearson 2006; Sohdi & Erhlich 2010). This has raised concern among conservation biologists regarding our ability to maintain and improve biodiversity in commodity production landscapes (Whitfield et al. 2007; Wilson et al. 2010), an issue which is particularly poignant as increasing food demands are being met by intensifying agricultural practices (Fischer et al. 2008; Phalan et al. 2011; Tscharntke et al. 2012). Habitat fragmentation and agricultural intensification have been linked to herpetofaunal declines in Australia (Brown, Bennett & Potts 2008), Africa (Fabricius, Burger & Hockey 2003), South America (Whitfield et al. 2007) and Europe (Araujo, Thuiller & Pearson 2006; Ribeiro et al. 2009). Agricultural practices can lead to changes in plant species composition (Dorrough & Scroggie 2008), loss of vegetation structure (McIntyre & Tongway 2005) and altered soil chemistry (Benton, Vickery & Wilson 2003). Reptiles and amphibians are particularly sensitive to ground cover modification (Fischer, Lindenmayer & Cowling 2004; Jellinek, Driscoll & Kirkpatrick 2004; Brown, Dorrough & Ramsay 2011), primarily because many species are dependent on terrestrial retreat sites (Valentine, Roberts & Schwarzkof 2007), environments that are easily destroyed by agricultural practices (Cosentino, Schooley & Phillips 2011; Dorrough et al. 2012). Habitat specificity and poor dispersal ability in many species (Schutz & Driscoll 2008) also mean that reptiles and amphibians are prone to local extinction (Moore et al. 2008). Thus, the response of herpetofauna to native vegetation management in agricultural landscapes should be of importance to land managers but requires immediate investigation.

We sought to address this ecological knowledge gap using a landscape-scale study of agri-environment schemes (AES) in threatened semi-arid and temperate woodlands of south-eastern Australia – an area that has been subject to significant changes in vegetation cover over the past 150 years (Lindenmayer, Bennett & Hobbs 2010). The AES, initiated by the European Union's Common Agricultural Policy, were developed as a policy instrument to mitigate negative effects of agricultural intensification on biodiversity (Kleijn & Sutherland 2003). AES involve paying farmers to modify farming practices with the goal of providing environmental benefits such as increased biodiversity (Kleijn et al. 2006; Concepcion et al. 2012). However, many AES have been criticized for their lack of rigorous assessment, monitoring and evaluation (Zammit, Attwood & Burns 2010; Concepcion et al. 2012). Most studies that have evaluated the effectiveness of AES have focused on vegetation, invertebrate functional groups or particular avifaunal guilds (Whittingham 2007; Lindenmayer et al. 2012). Furthermore, few studies have evaluated the benefits of AES in protecting biodiversity relative to the broader farming landscape (Merckx et al. 2009), and no studies have explicitly evaluated the effectiveness of AES in protecting or increasing herpetofaunal diversity (Kleijn & Sutherland 2003; Kampmann et al. 2012). Thus, the value of AES in providing environmental benefits remains one of the most important policy-relevant ecological questions in recent times (Sutherland et al. 2006; Mauchline et al. 2012).

In south-eastern Australia, changes in the extent and quality of native vegetation occurred rapidly following European settlement (Yates & Hobbs 1997), with catastrophic effects on biodiversity (Lindenmayer, Bennett & Hobbs 2010). One agency responsible for managing natural resources in New South Wales is the Murray Catchment Management Authority (MCMA). The primary role of the MCMA is to prepare catchment action plans (CAP), manage financial incentive delivery programmes and allocate funds to develop property vegetation plans (PVP) to achieve natural resource management targets. Under the PVP, funds allocated by the Australian Government are awarded to landholders via a competitive grants process with the aim of, for example, improving vegetation condition and biodiversity outcomes (Murray Catchment Management Authority 2012).

We sought to determine whether the AES in the Murray catchment protect greater herpetofaunal diversity compared to the broader farming landscape, and whether levels of herpetofaunal diversity increase as a function of time-since-management intervention. We selected herpetofauna for detailed study because the group is relatively species rich within the Murray catchment (Michael & Lindenmayer 2010), includes threatened species of conservation concern, and a paucity of empirical data exists to evaluate responses to AES. Previous work in the woodlands of south-eastern Australia has shown that reptiles in particular are associated with local-scale attributes such as woody debris and native grass cover (Michael, Cunningham & Lindenmayer 2008; Brown, Dorrough & Ramsay 2011). These variables can be altered by agricultural practices and are expected to improve under AES. To evaluate the effectiveness of AES, we posed three main questions:

  1. Does herpetofaunal diversity differ among management regimes and broad vegetation communities?
  2. Are there particular environmental variables that relate to herpetofaunal occurrence patterns?
  3. To what extent can AES meet the objective of reversing herpetofaunal declines in rural landscapes?

Our questions were motivated by the knowledge that the MCMA invest in high-quality remnant vegetation, and these sites have the potential to support high levels of biodiversity (Lindenmayer et al. 2012). Management interventions such as reducing livestock grazing pressure and controlling invasive exotic plants are predicted to further improve the structure and condition of native vegetation condition, which in turn, may have positive benefits for herpetofauna. Based on habitat complexity theory (sensu MacArthur & MacArthur 1961), we expected to find a significant difference in species diversity between structurally complex vegetation under AES compared to structurally simple vegetation in the farming landscape, as well as gradients in species diversity in relation to time-since-management intervention. However, it is unclear what influence historical land-use practices and the prior filtering of species pools had on shaping patterns of herpetofauna in the region. Hence, the capacity for herpetofauna to respond to management remains unknown and is based on the assumption that travelling stock reserves (TSR) adequately reflect the biogeographic history of the region (Lindenmayer et al. 2012).

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Study Area

We conducted our study in threatened semi-arid and temperate woodlands within the Murray catchment of New South Wales, an area defined by the Murray River in the south, Billabong Creek water shed in the north, and the towns of Urana in the east (35°44′10′′S 146°02′58′′E) and Moulamein (35°01′59′′S 143°43′42′′E) in the west (Fig. 1). The annual average rainfall in the region is 320 mm and is uniformly distributed throughout the year. The average minimum and maximum summer temperature ranges from 17 to 33 °C, and the average minimum and maximum winter temperature ranges from 0 to 14 °C. The catchment has a diverse agricultural sector with grazing, cropping and horticulture being the main industries (Murray Catchment Management Authority 2012). Over 100 species of herpetofauna have been recorded in the Murray catchment (Michael & Lindenmayer 2010), including the nationally endangered pink-tailed worm lizard Aprasia parapulchella and the vulnerable Sloane's froglet Crinia sloanei. Historical land-use practices undoubtedly played a crucial role in shaping the distribution and abundance of the region's herpetofauna. However, quantitative baseline data on herpetofauna are lacking in the region, including an understanding of source–sink population dynamics and dispersal abilities, information which is important for understanding the capacity for biological communities to contribute species to restored areas.

image

Figure 1. Location of survey sites in the Murray catchment of New South Wales showing the spatial relationship of woodland communities (Keith Class).

Download figure to PowerPoint

Experimental Design

Between December 2007 and February 2008, we visited 53 farms across the Murray catchment and inspected remnant native vegetation placed under a MCMA management agreement in 2007. We selected sites that were >2 ha and were representative of a threatened vegetation community. We stratified sites by Keith's Class vegetation community (sensu Keith 2004) and management regime (Fig. 2) and ensured independence by selecting sites >500 m apart. We interviewed landholders to verify control sites were representative of historical land-use practices (i.e. intensively grazed by livestock with no strategic rest periods). Our management classes are as follows.

image

Figure 2. Images of Riverine sandhill woodland (RSW) in the Murray catchment showing key structural differences in native vegetation cover between (a) travelling stock reserves (TSR), (b) long-term conversion site, (c) short-term conversion site and (d) livestock production site.

Download figure to PowerPoint

  1. Production sites (PRD) included woodland used for livestock production outcomes. These areas have historically experienced set stocking rates, occasional fertilizer application, limited or no pest plant or animal control and serve as controls.
  2. Short-term conversion sites (STC) included woodland placed under a MCMA management agreement in 2007. Management interventions include removing or reducing livestock grazing pressure (e.g. pulse grazing during winter), planting understorey species and controlling invasive exotic plants and feral animals.
  3. Long-term conversion sites (LTC) included woodland managed for biodiversity conservation outcomes prior to 2003. In these cases, funding to exclude or reduce livestock grazing pressure and restore shrub diversity was provided by other agencies such as Greening Australia.
  4. TSR included woodland managed by the Livestock Health Protection Authority. TSR form part of a network established more than 150 years ago to facilitate the movement of domestic livestock between properties and markets, and because they were periodically grazed and spared from vegetation clearing (Lentini et al. 2011), they best represent woodland condition states prior to the arrival of European settlers. Nature reserves were not available; therefore, we selected TSR as reference sites.

Our woodland communities (Keith Class) included as follows: (i) floodplain transition woodland (FTW) dominated by grey box E. microcarpa; (ii) inland floodplain woodland (IFW) dominated by black box E. largiflorens, (iii) Riverine plain woodland (RPW) dominated by boree Acacia pendula and (iv) Riverine sandhill woodland (RSW) dominated by white cypress pine Callitris glaucophylla and yellow box E. melliodora. Our final design encompassed 105 sites on 31 farms (Table 1).

Table 1. Stratification of reptile monitoring sites in the Murray catchment of New South Wales
Vegetation type (Keith class) / management typeProduction site (control)Short-term conversion site (treatment)Long-term conversion site (treatment)Travelling stock reserve (reference)Total
  1. FTW, floodplain transitional woodland; IFW, inland floodplain woodland; RPW, Riverine plains woodland; RSW, Riverine sandhill woodland.

Grey box (FTW)949426
Black box (IFW)958426
Boree (RPW)1226424
Sandhill (RSW)1059529
Total 40163217105

Survey Protocol

During October 2008, August 2009 and August 2010, we recorded frogs and reptiles along a 200 × 50 m transect using a time- and area-constrained (30 min × 1 ha) survey protocol involving active searches of natural habitat and inspections of artificial refuge (AR) arrays. Each AR array consisted of four railway sleepers (1·2 m in length), four roof tiles and one double stack of 1-m² corrugated steel sheet. AR arrays were established within the 1-ha search area and placed 100-m apart along the 200-m transect (at the 0-m and 100-m points). Using a combination of survey techniques increases the probability of detecting cryptic species (Michael et al. 2012). Surveys were conducted once per year over a one-week period on clear days between 0900 and 1400 hours by the same group of ecologists.

Environmental and Vegetation Measurements

We surveyed vegetation during November 2008 and 2010. We collected percentage cover abundance measures for native and exotic grass, native and exotic shrubs (including subshrubs), bare ground, soil crust, leaf litter, overstorey and mid-storey using the line-intercept method, whereby the presence/absence of biotic and abiotic attributes (see Appendix S1, Supporting information) was recorded every metre along a 50-m transect. At 1-m intervals, a decision was made as to whether the ‘vegetation’ attribute intersected with a point corresponding to the metre interval on the tape. More than one attribute can intersect a single point (e.g. organic litter and grass can intersect the same point). Percentage cover for an attribute was calculated by adding all presences for an attribute, dividing this total by the total number of points measured and multiplying by 100 to convert to%. We applied the step-point count method between 0–50 m and 150–200 m. We measured native plant species richness in a 20 × 20 m plot located midway along the transect and recorded overstorey and mid-storey recruitment, length of logs and number of hollow bearing trees within two 50 × 20 m plots located at opposite ends of the transect.

Data Analysis

We used generalized linear models (GLMs) (McCullagh & Nelder 1989) and hierarchical generalized linear models (HGLMs) (Lee, Nelder & Pawitan 2006) to examine the relationship between species diversity and site-level environmental variables using a quasi-Poisson distribution and log link function. Candidate vegetation variables were selected from a larger set of environmental attributes collected at each site (see Appendix S1, Supporting information) based on the literature and expert knowledge. We tested for collinearity and found correlations to be weak (Pearson correlation coefficient <0·53). Thus, the following design variables (vegetation type and management class) and candidate variables were included in regression models: native grass cover, native shrub cover, exotic plant cover, bare ground, organic litter, cryptogamic crust, native overstorey cover, native mid-storey cover, native plant species richness, total length of logs and number of trees with hollows. We linked the first vegetation survey to the first two fauna surveys, and we linked the last fauna survey to the second vegetation survey. We restricted analysis of amphibians to data collected in 2010 due to the paucity of records in preceding years.

We fitted GLMs to the average number of individuals per year at each site (= 105 sites), and we fitted HGLMs with site as a random term to the numbers at each site × year combination (= 291 as some sites were not surveyed in 2010 due to flooding). We fitted models for total herpetofauna, total reptile species, total frog species and two common lizard species. After developing a model based on vegetation variables, we checked to see whether any further improvement could be made by adding vegetation type and treatment. For site × year data, we also included year as a candidate predictor. The most parsimonious model was selected using all possible subsets of the candidate predictors using Schwarz information criteria (δSIC < 2). For models with random effects, we fitted GLMs to choose appropriate terms for the model and then fitted an HGLM. To examine reptile assemblage structure at the site level, we performed correspondence analysis using a chi-squared similarity metric (Greenacre 2007). The effect of the design variables and candidate variables on the site scores from the correspondence analysis was investigated using linear regression analysis. The conclusions were verified using anosim (Clarke1993) and permanova (Anderson 2001). We examined distance decay of similarity by performing Mantel tests of the relationship between Bray–Curtis distance and geographical distance among sites (Soininen, McDonald & Hillebrand 2007). However, the association was non-significant. Statistical analyses were performed using genstat release 15.1 (VSN International 2012) and R (R Core Development Team 2012).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Summary Statistics

We recorded 31 species of herpetofauna from nine families (Appendix S2, Supporting information). No threatened species were recorded. Lizards from the family Scincidae accounted for 84% of records and were dominated by Morethia boulengeri (58·7% of observations) and Cryptoblepharus pannosus (16·7% of observations). We recorded 58% of all reptile species on 10 or less occasions. Frog detections were low and dominated by two species (Appendix S2, Supporting information).

Effect of Management and Vegetation Type

Our analysis revealed few management effects, several vegetation type effects and few interaction effects. Herpetofauna richness (= 0·25), reptile richness (= 0·22) and frog richness (= 0·37) did not differ significantly among management classes (Fig. 3). Herpetofauna abundance (= 0·02) differed among management classes due to the high abundance of M. boulengeri (= 0·04) and C. pannosus (= 0·01) in grey box (FTW) and black box (IFW) TSR (Fig. 3).

image

Figure 3. Estimated mean species richness and abundance (±1 SE) for herpetofauna, reptiles, frogs, Morethia boulengeri and Cryptoblepharus pannosus by vegetation community (FTW, floodplain transitional woodland; IFW, inland floodplain woodland; RPW, Riverine plains woodland; RSW, Riverine sandhill woodland) and management regime; PRD, production site; STC, short-term conversion site; LTC, long-term conversion site; TSR, travelling stock reserve in the Murray catchment.

Download figure to PowerPoint

We found herpetofaunal richness (= 0·002), total reptile richness (= 0·003) and frog abundance (< 0·001) differed significantly among vegetation types (Fig. 3). Boree woodland (RPW) supported twice as many herpetofauna (4·01 species cf. 2·29 species) and reptile species (3·75 species cf. 2·17 species) than grey box (FTW). Frogs were more abundant in boree woodland (< 0·001), particularly in LTC and TSR treatments (= 0·012).

Predictors of Species Richness and Abundance

The most parsimonious models for herpetofaunal richness, herpetofaunal abundance and reptile richness included native plant species richness and bare ground as explanatory variables (Table 2). Native plant species richness was also a significant predictor in lower-ranked models (Appendix S3, Supporting information). Cryptoblepharus pannosus abundance was positively related to native plant species richness and total length of fallen logs, and M. boulengeri abundance was positively related to TSR and negatively with bare ground (Table 2). When we included year as a fixed effect, herpetofaunal richness, reptile richness and abundance were negatively related to exotic grass (Table 3). Native plant richness, length of logs and TSR were also significant predictors in lower-ranked models (Appendix S4, Supporting information). Native mid-storey cover and TSR were positive predictors of herpetofauna and M. boulengeri abundance, whereas C. pannosus was positively related to year and length of logs, and negatively with organic litter. We graphically present the mean cover abundance values for two key explanatory variables, bare ground and organic leaf litter, in Fig. 4.

Table 2. Site-based models for herpetofauna richness and abundance, reptile richness and abundance, and two common lizard species in the Murray catchment (Based on Schwarz information criteria, δSIC < 2)
ResponsePredictorP-valueCoefficientδSIC
  1. TSR, travelling stock reserves.

Total herpetofauna richnessNative plant richness<0·0010·040 ± 0·01100·00
Bare ground<0·0010·0097 ± 0·00221
Total herpetofauna abundanceNative plant richness0·0300·035 ± 0·01590·00
TSR0·0190·34 ± 0·144
Total reptile richnessNative plant richness<0·0010·040 ± 0·01090·00
Bare ground<0·0010·0090 ± 0·00220
M. boulengeri abundanceBare ground0·006−0·012 ± 0·00420·00
TSR0·0080·42 ± 0·156
C. pannosus abundanceNative plant richness0·0130·076 ± 0·03040·00
Total length of fallen logs<0·0010·018 ± 0·0048
Table 3. Site-by-year models for herpetofauna richness and abundance, reptile richness and abundance, and two common lizard species in the Murray catchment (based on Schwarz information criteria, δSIC < 2)
ResponsePredictorP-valueCoefficientδSIC
  1. TSR, travelling stock reserves.

Total herpetofauna richnessExotic plant cover0·005−0·005 ± 0·00200·00
Total herpetofauna abundanceNative mid-storey cover0·0260·038 ± 0·01690·00
TSR0·0350·33 ± 0·155
Total reptile richnessExotic plant cover<0·001−0·0067 ± 0·00190·00
Total reptile abundanceTotal length of fallen logs<0·0010·0083 ± 0·002320·00
Exotic plant cover0·009−0·0069 ± 0·0026
TSR0·0090·387 ± 0·147
M. boulengeri abundanceTSR0·0090·43 ± 0·1660·00
C. pannosus abundanceTSR0·0900·499 ± 0·2940·00
Total length of fallen logs<0·0010·0165 ± 0·0048
Organic litter0·008−0·0171 ± 0·0065
2009 vs. 2008<0·001−0·762 ± 0·154
image

Figure 4. Percentage cover abundance estimates (±95 CI) for bare ground and organic litter among four management classes in four vegetation communities in the Murray catchment (Data represent the average of two vegetation surveys).

Download figure to PowerPoint

Assemblage Structure

We found no significant difference in reptile assemblage structure between management classes (permanova,= 0·13 and anosim,= 0·74). Similar results were obtained if we included vegetation type as a stratum when testing for management effects (permanova,= 0·11 and anosim,= 0·72). permanova revealed significant differences in reptile assemblage structure between vegetation types (= 0·007), as did anosim (= 0·012). A correspondence analysis of species and sites (shown as treatment means) in ordination space illustrate these findings (Fig. 5a, b). We found the first component from the correspondence analysis of species accounted for 18·1% and was significantly related to vegetation type (F3,100 = 9·7, < 0·001) and percentage of bare ground (F1,102 = 52·2, < 0·001). After adjusting for bare ground, the vegetation type effect was reduced (F3,99 = 3·6, = 0·016). This is illustrated in Fig. 6, which shows the first component scores for individual reptile species as well as the average scores for the four vegetation types. The sites were divided into three approximately equally spaced groups according to the percentage of bare ground: low (0–27%), medium (27–54%) and large (54–81%). The second component accounted for 13·8% and was significantly related to vegetation type (F3,100 = 5·7, = 0·001) and percentage organic litter (F1,102 = 4·8, = 0·03). After adjusting for organic litter cover, the vegetation effect was much greater (F3,99 = 4·9, = 0·003).

image

Figure 5. Correspondence analysis showing ordination of: (a) the reptile assemblage and (b) survey sites (open circles) showing treatment means. The first component of the species ordination related to Keith Class and bare ground and explained 18% of the variation. The second component was related to Keith Class and organic litter and explained 13% of the variation. (See Appendix S2, Supporting information for species abbreviations – first letter of the genus and first three letters of species. Treatment abbreviations are explained in the text).

Download figure to PowerPoint

image

Figure 6. Strip plot for the first component of the correspondence analysis of reptile species showing the relationship with bare ground and Keith Class (divisions for bare ground: low = 0–27%, medium = 27–54%, high = 54–81%).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

This study evaluated the effectiveness of an agri-environment scheme for protecting and improving herpetofaunal diversity. Several key findings emerged from this work, including: (i) relatively high regional reptile diversity, but low site-level diversity, (ii) no evidence to indicate that AES protect significantly more herpetofauna compared to the broader farming landscape, (iii) vegetation management increases the abundance of common species, (iv) vegetation type rather than management regime influences reptile assemblage structure, and (v) native plant diversity and woody debris are significant predictors of reptile occurrence. We further discuss the implications of our findings below in the context of improving AES.

Effect of Management on Species Diversity

At the onset of our study, we predicted that sites under AES would support higher species richness and that species richness would increase as a function of time-since-management intervention, with production areas and STC sites having lower species richness than LTC and TSR, a pattern reported for the region's avifauna (Lindenmayer et al. 2012). However, we found no significant differences in mean species richness among management classes for any response variable (Fig. 3). Only M. boulengeri showed a gradient in abundance, being recorded more frequently in TSR (Fig. 3). This implies that herpetofauna (reptiles in particular) may not be responsive to AES in the short term (<10 years). Benefits to reptiles from vegetation management appear to operate at the population level by increasing the abundance of common species and not at the community level by increasing local-scale species richness. For example, gradients in reptile abundance from PRD to TSR were evident across several vegetation types (e.g. FTW, IFW and RPW Fig. 3). This pattern is common among AES, where improvements in vegetation under AES benefit widespread and common species (Kleijn et al. 2006; Fuentes-Montemayor, Goulson & Park 2011; although see Merckx et al. 2010; Perkins et al. 2011). Two reasons why AES may fail to increase herpetofaunal diversity in this system could be due to the legacy of historical land-use practices (Harding et al. 1998) and the presence of dispersal barriers (Cosentino, Schooley & Phillips 2011). Many species of herpetofauna in temperate Australia have limited dispersal ability in modified landscapes (Schutz & Driscoll 2008; Williams, Driscoll & Bull 2012). Thus, the effectiveness of AES to improve herpetofaunal diversity is likely to depend on the distance to regional species pools, interpatch suitability and resource availability in the agricultural matrix (Donald & Evans 2006). Thus, a key management recommendation to improve AES is to restore habitat connectivity in the broader agricultural landscape by linking remnant vegetation.

Predictors of Species Richness and Abundance

We found several vegetation attributes to be positive predictors of herpetofaunal diversity (Tables 2 and 3), particularly native plant richness and bare ground cover (Fig. 4). Positive relationships between native plant richness and reptile diversity are well-established (Michael, Cunningham & Lindenmayer 2008; Schutz & Driscoll 2008; Brown, Dorrough & Ramsay 2011), as are negative relationships with exotic plant cover (Martin & Murray 2011). Several mechanisms may explain the avoidance of weed-infested areas by reptiles in particular, including unsuitable thermal conditions and low prey availability (Valentine, Roberts & Schwarzkof 2007). In grassy woodland ecosystems, native plant diversity and sward structure can be degraded by grazing pressure, fertilizer use and soil disturbance (McIntyre & Tongway 2005). Furthermore, it may take several decades before soil nutrients return to levels that can support native grasses and forbs (Dorrough & Scroggie 2008). Adopting a farm-scale approach to sustainable grazing may improve grassland condition and potentially benefit herpetofauna in the long term. Developing incentives around holistic rotational grazing systems should therefore be a high priority in future AES.

Assemblage Structure

We found grazed woodland remnants (PRD) supported a similar number of reptile species compared to sites under AES (e.g. 16 species from 40 PRD cf. 19 species from 32 LTC sites). In grey box (FTW), species richness was highest in PRD (Fig. 3). Furthermore, the abundance of D. tessellatus, S. intermedius, C. pannosus, M. adelaidensis and Suta suta was also highest in PRD (Appendix S2, Supporting information), although low detections prevented formal analysis. When we examined reptile assemblages in more detail, we found no evidence to suggest that management influenced community composition (Fig. 5a, b). These findings are congruent with other studies which illustrate the value of remnant vegetation in production areas for maintaining reptile diversity (Fischer et al. 2005), but also highlights a limitation of AES, whereby parts of the landscape potentially rich in reptiles are not protected because they fail to meet investment criteria (e.g. minimum patch size and vegetation condition benchmarks). An alternative approach to conserving herpetofauna under AES may be to protect vegetation communities in various condition states, and at multiple spatial scales (Nicholson et al. 2006). This approach is likely to protect greater herpetofaunal diversity per unit area than investing in structurally homogeneous, high-quality vegetation.

In contrast to the lack of management effects, we found reptile assemblage structure varied significantly among vegetation types (Fig. 5a, b), whereby M. adelaidensis, Tiliqua rugosa and D. tessellatus were associated with boree woodland (RPW). Correspondence analysis revealed vegetation type, bare ground and organic litter to be key discriminating variables influencing assemblage structure. This relationship is underpinned by broad zoogeographical distribution patterns and life-form strategies. For example, the fossorial Lerista timida was associated with sandhill woodland (Fig. 6), a vegetation community found on sandy soils (Keith 2004). The arboreal C. pannosus was associated with black box (IFW) and grey box (FTW), eucalypt-dominated communities which produce large amounts of woody debris, and the arboreal S. intermedius (Fig. 6) shelters behind the bark of Callitris trees, a species associated with sandhill woodland (Keith 2004). Hence, strong habitat affiliations and vegetation heterogeneity may create natural dispersal barriers and limit the number of species potentially able to respond to AES in botanically diverse landscapes.

Conclusion

This study represents the first empirical investigation on the effectiveness of AES in protecting and improving herpetofaunal diversity. However, this study shares several limitations common to reptile studies in particular, notably, low site-level species richness. Some researchers have interpreted this pattern as reflecting an impoverished fauna (Brown, Bennett & Potts 2008; Brown, Dorrough & Ramsay 2011) resulting from prior filtering of regional species pools (Dorrough et al. 2012). However, because baseline data are lacking for herpetofauna in temperate woodlands, inferring pre-European assemblage structure is difficult. Nevertheless, our findings suggest that AES are unlikely to improve herpetofaunal diversity in the short term, emphasizing the need to develop new approaches to conserve low-vagility organisms such as frogs and reptiles in agricultural landscapes.

Our key recommendations for improving AES are to restore habitat connectivity to enhance local-scale diversity and protect mosaics of different vegetation types at multiple spatial scales (i.e. farm level to landscape level) to preserve regional diversity. These approaches are more likely to protect greater herpetofaunal diversity (Nicholson et al. 2006) than focusing solely on patch-scale management interventions. Future incentive schemes also should focus on improving sward structure, floristic diversity and increasing woody debris in the broader farming landscape to dissolve dispersal barriers created by past and present agricultural practices.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

This study was funded by the Australian Government's Caring for our Country scheme, the Murray Catchment Management Authority and the Australian Research Council. We thank David Leslie, Jack Chubb, Helen Wilson, Emmo Willinck and David Costello for supporting this work. Christopher MacGregor, Sachiko Okada, Lachlan McBurney, David Blair and Geoffrey Kay assisted with data collection. Rowan Winsemius produced the location map. Professor Ross Cunningham provided guidance in the initial stages of the experimental design of this study. This study was approved by the Australian National University Animal Care and Ethics Committee under the scientific licence (S12604) from the New South Wales National Parks and Wildlife Service.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
jpe12215-sup-0001-AppendixS1-S4.docWord document273K

Appendix S1. Vegetation attributes recorded at each long-term monitoring site.

Appendix S2. Abundance of herpetofauna by management class and vegetation type (sensu Keith 2004). Bold values represent high standardized values.

Appendix S3. Lower-ranked site-based models for total herpetofauna richness and abundance, total reptile richness and abundance, and two common lizard species in the NSW Murray catchment (model selection based on Schwarz information criteria, δSIC < 2).

Appendix S4. Lower-ranked site-by-year based models for total herpetofauna richness and abundance, total reptile richness and abundance, and two common lizard species in the NSW Murray catchment (model selection based on Schwarz information criteria, δSIC < 2).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.